Revert "[TTI] Reduction costs only need to include a single extract element cost" This reverts commit r346970. It was causing PR39774, a crash in slp-vectorizer on a rather simple loop with just a bunch of 'and's in the body. llvm-svn: 347541

commit: 8cd9d1b5cebe8a694089ac983038966a2fe6a516 [log] [tgz]
author: Fedor Sergeev <fedor.sergeev@azul.com> Mon Nov 26 10:17:27 2018 +0000
committer: Fedor Sergeev <fedor.sergeev@azul.com> Mon Nov 26 10:17:27 2018 +0000
tree: 0627872d1fcb453f0793656b68509333d84a24ac
parent: 0e0cd5be403a83de607f88e469a4de06051ac39c [diff]
diff --git a/llvm/test/Analysis/CostModel/AArch64/vector-reduce.ll b/llvm/test/Analysis/CostModel/AArch64/vector-reduce.ll
index 3945dcf..c268a18e 100644
--- a/llvm/test/Analysis/CostModel/AArch64/vector-reduce.ll
+++ b/llvm/test/Analysis/CostModel/AArch64/vector-reduce.ll

@@ -47,7 +47,7 @@
 }
 
 ; COST-LABEL: umin.i8.v8i8
-; COST:       Found an estimated cost of 217 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> %v)
 ; CODE-LABEL: umin.i8.v8i8
 ; CODE:       uminv b0, v0.8b
 define i8 @umin.i8.v8i8(<8 x i8> %v) {
@@ -56,7 +56,7 @@
 }
 
 ; COST-LABEL: umin.i8.v16i8
-; COST:       Found an estimated cost of 609 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> %v)
+; COST:       Found an estimated cost of 744 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> %v)
 ; CODE-LABEL: umin.i8.v16i8
 ; CODE:       uminv b0, v0.16b
 define i8 @umin.i8.v16i8(<16 x i8> %v) {
@@ -65,7 +65,7 @@
 }
 
 ; COST-LABEL: umin.i16.v4i16
-; COST:       Found an estimated cost of 65 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> %v)
+; COST:       Found an estimated cost of 92 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> %v)
 ; CODE-LABEL: umin.i16.v4i16
 ; CODE:       uminv h0, v0.4h
 define i16 @umin.i16.v4i16(<4 x i16> %v) {
@@ -74,7 +74,7 @@
 }
 
 ; COST-LABEL: umin.i16.v8i16
-; COST:       Found an estimated cost of 217 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> %v)
 ; CODE-LABEL: umin.i16.v8i16
 ; CODE:       uminv h0, v0.8h
 define i16 @umin.i16.v8i16(<8 x i16> %v) {
@@ -83,7 +83,7 @@
 }
 
 ; COST-LABEL: umin.i32.v4i32
-; COST:       Found an estimated cost of 35 for instruction: %r = call i32 @llvm.experimental.vector.reduce.umin.i32.v4i32(<4 x i32> %v)
+; COST:       Found an estimated cost of 62 for instruction: %r = call i32 @llvm.experimental.vector.reduce.umin.i32.v4i32(<4 x i32> %v)
 ; CODE-LABEL: umin.i32.v4i32
 ; CODE:       uminv s0, v0.4s
 define i32 @umin.i32.v4i32(<4 x i32> %v) {
@@ -92,7 +92,7 @@
 }
 
 ; COST-LABEL: umax.i8.v8i8
-; COST:       Found an estimated cost of 217 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> %v)
 ; CODE-LABEL: umax.i8.v8i8
 ; CODE:       umaxv b0, v0.8b
 define i8 @umax.i8.v8i8(<8 x i8> %v) {
@@ -101,7 +101,7 @@
 }
 
 ; COST-LABEL: umax.i8.v16i8
-; COST:       Found an estimated cost of 609 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> %v)
+; COST:       Found an estimated cost of 744 for instruction: %r = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> %v)
 ; CODE-LABEL: umax.i8.v16i8
 ; CODE:       umaxv b0, v0.16b
 define i8 @umax.i8.v16i8(<16 x i8> %v) {
@@ -110,7 +110,7 @@
 }
 
 ; COST-LABEL: umax.i16.v4i16
-; COST:       Found an estimated cost of 65 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> %v)
+; COST:       Found an estimated cost of 92 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> %v)
 ; CODE-LABEL: umax.i16.v4i16
 ; CODE:       umaxv h0, v0.4h
 define i16 @umax.i16.v4i16(<4 x i16> %v) {
@@ -119,7 +119,7 @@
 }
 
 ; COST-LABEL: umax.i16.v8i16
-; COST:       Found an estimated cost of 217 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> %v)
 ; CODE-LABEL: umax.i16.v8i16
 ; CODE:       umaxv h0, v0.8h
 define i16 @umax.i16.v8i16(<8 x i16> %v) {
@@ -128,7 +128,7 @@
 }
 
 ; COST-LABEL: umax.i32.v4i32
-; COST:       Found an estimated cost of 35 for instruction: %r = call i32 @llvm.experimental.vector.reduce.umax.i32.v4i32(<4 x i32> %v)
+; COST:       Found an estimated cost of 62 for instruction: %r = call i32 @llvm.experimental.vector.reduce.umax.i32.v4i32(<4 x i32> %v)
 ; CODE-LABEL: umax.i32.v4i32
 ; CODE:       umaxv s0, v0.4s
 define i32 @umax.i32.v4i32(<4 x i32> %v) {
@@ -137,7 +137,7 @@
 }
 
 ; COST-LABEL: smin.i8.v8i8
-; COST:       Found an estimated cost of 217 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> %v)
 ; CODE-LABEL: smin.i8.v8i8
 ; CODE:       sminv b0, v0.8b
 define i8 @smin.i8.v8i8(<8 x i8> %v) {
@@ -146,7 +146,7 @@
 }
 
 ; COST-LABEL: smin.i8.v16i8
-; COST:       Found an estimated cost of 609 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> %v)
+; COST:       Found an estimated cost of 744 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> %v)
 ; CODE-LABEL: smin.i8.v16i8
 ; CODE:       sminv b0, v0.16b
 define i8 @smin.i8.v16i8(<16 x i8> %v) {
@@ -155,7 +155,7 @@
 }
 
 ; COST-LABEL: smin.i16.v4i16
-; COST:       Found an estimated cost of 65 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> %v)
+; COST:       Found an estimated cost of 92 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> %v)
 ; CODE-LABEL: smin.i16.v4i16
 ; CODE:       sminv h0, v0.4h
 define i16 @smin.i16.v4i16(<4 x i16> %v) {
@@ -164,7 +164,7 @@
 }
 
 ; COST-LABEL: smin.i16.v8i16
-; COST:       Found an estimated cost of 217 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> %v)
 ; CODE-LABEL: smin.i16.v8i16
 ; CODE:       sminv h0, v0.8h
 define i16 @smin.i16.v8i16(<8 x i16> %v) {
@@ -173,7 +173,7 @@
 }
 
 ; COST-LABEL: smin.i32.v4i32
-; COST:       Found an estimated cost of 35 for instruction: %r = call i32 @llvm.experimental.vector.reduce.smin.i32.v4i32(<4 x i32> %v)
+; COST:       Found an estimated cost of 62 for instruction: %r = call i32 @llvm.experimental.vector.reduce.smin.i32.v4i32(<4 x i32> %v)
 ; CODE-LABEL: smin.i32.v4i32
 ; CODE:       sminv s0, v0.4s
 define i32 @smin.i32.v4i32(<4 x i32> %v) {
@@ -182,7 +182,7 @@
 }
 
 ; COST-LABEL: smax.i8.v8i8
-; COST:       Found an estimated cost of 217 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> %v)
 ; CODE-LABEL: smax.i8.v8i8
 ; CODE:       smaxv b0, v0.8b
 define i8 @smax.i8.v8i8(<8 x i8> %v) {
@@ -191,7 +191,7 @@
 }
 
 ; COST-LABEL: smax.i8.v16i8
-; COST:       Found an estimated cost of 609 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> %v)
+; COST:       Found an estimated cost of 744 for instruction: %r = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> %v)
 ; CODE-LABEL: smax.i8.v16i8
 ; CODE:       smaxv b0, v0.16b
 define i8 @smax.i8.v16i8(<16 x i8> %v) {
@@ -200,7 +200,7 @@
 }
 
 ; COST-LABEL: smax.i16.v4i16
-; COST:       Found an estimated cost of 65 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> %v)
+; COST:       Found an estimated cost of 92 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> %v)
 ; CODE-LABEL: smax.i16.v4i16
 ; CODE:       smaxv h0, v0.4h
 define i16 @smax.i16.v4i16(<4 x i16> %v) {
@@ -209,7 +209,7 @@
 }
 
 ; COST-LABEL: smax.i16.v8i16
-; COST:       Found an estimated cost of 217 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> %v)
+; COST:       Found an estimated cost of 280 for instruction: %r = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> %v)
 ; CODE-LABEL: smax.i16.v8i16
 ; CODE:       smaxv h0, v0.8h
 define i16 @smax.i16.v8i16(<8 x i16> %v) {
@@ -218,7 +218,7 @@
 }
 
 ; COST-LABEL: smax.i32.v4i32
-; COST:       Found an estimated cost of 35 for instruction: %r = call i32 @llvm.experimental.vector.reduce.smax.i32.v4i32(<4 x i32> %v)
+; COST:       Found an estimated cost of 62 for instruction: %r = call i32 @llvm.experimental.vector.reduce.smax.i32.v4i32(<4 x i32> %v)
 ; CODE-LABEL: smax.i32.v4i32
 ; CODE:       smaxv s0, v0.4s
 define i32 @smax.i32.v4i32(<4 x i32> %v) {
@@ -227,7 +227,7 @@
 }
 
 ; COST-LABEL: fmin.f32.v4f32
-; COST:       Found an estimated cost of 35 for instruction: %r = call nnan float @llvm.experimental.vector.reduce.fmin.f32.v4f32(<4 x float> %v)
+; COST:       Found an estimated cost of 62 for instruction: %r = call nnan float @llvm.experimental.vector.reduce.fmin.f32.v4f32(<4 x float> %v)
 ; CODE-LABEL: fmin.f32.v4f32
 ; CODE:       fminnmv s0, v0.4s
 define float @fmin.f32.v4f32(<4 x float> %v) {
@@ -236,7 +236,7 @@
 }
 
 ; COST-LABEL: fmax.f32.v4f32
-; COST:       Found an estimated cost of 35 for instruction: %r = call nnan float @llvm.experimental.vector.reduce.fmax.f32.v4f32(<4 x float> %v)
+; COST:       Found an estimated cost of 62 for instruction: %r = call nnan float @llvm.experimental.vector.reduce.fmax.f32.v4f32(<4 x float> %v)
 ; CODE-LABEL: fmax.f32.v4f32
 ; CODE:       fmaxnmv s0, v0.4s
 define float @fmax.f32.v4f32(<4 x float> %v) {

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-add.ll b/llvm/test/Analysis/CostModel/X86/reduce-add.ll
index 39272e5..f926062 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-add.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-add.ll

@@ -11,18 +11,18 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.add.i64.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.add.i64.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.add.i64.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.add.i64.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.add.i64.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.add.i64.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.add.i64.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.add.i64.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.add.i64.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.add.i64.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -45,8 +45,8 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.add.i64.v1i64(<1 x i64> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.add.i64.v2i64(<2 x i64> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.add.i64.v4i64(<4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.add.i64.v8i64(<8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.add.i64.v16i64(<16 x i64> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.add.i64.v1i64(<1 x i64> undef)
@@ -59,19 +59,19 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.add.i32.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.add.i32.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.add.i32.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.add.i32.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.add.i32.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.add.i32.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.add.i32.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.add.i32.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i32'
@@ -94,8 +94,8 @@
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.add.i32.v2i32(<2 x i32> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.add.i32.v4i32(<4 x i32> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.add.i32.v8i32(<8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.add.i32.v16i32(<16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.add.i32.v32i32(<32 x i32> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i32 @llvm.experimental.vector.reduce.add.i32.v2i32(<2 x i32> undef)
@@ -108,19 +108,19 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
@@ -134,41 +134,41 @@
 ; AVX1-LABEL: 'reduce_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.add.i16.v8i16(<8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.add.i16.v16i16(<16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.add.i16.v32i16(<32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.add.i16.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V4  = call i16 @llvm.experimental.vector.reduce.add.i16.v4i16(<4 x i16> undef)
@@ -181,67 +181,67 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 85 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.add.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.add.i8.v32i8(<32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.add.i8.v64i8(<64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.add.i8.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V8   = call i8 @llvm.experimental.vector.reduce.add.i8.v8i8(<8 x i8> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-and.ll b/llvm/test/Analysis/CostModel/X86/reduce-and.ll
index 3079b54..052b2ad 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-and.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-and.ll

@@ -11,34 +11,34 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE-LABEL: 'reduce_i64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.and.i64.v1i64(<1 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.and.i64.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.and.i64.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.and.i64.v1i64(<1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.and.i64.v2i64(<2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.and.i64.v4i64(<4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.and.i64.v8i64(<8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.and.i64.v16i64(<16 x i64> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.and.i64.v1i64(<1 x i64> undef)
@@ -51,35 +51,35 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE-LABEL: 'reduce_i32'
-; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i32'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i32'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.and.i32.v4i32(<4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.and.i32.v8i32(<8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.and.i32.v16i32(<16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.and.i32.v32i32(<32 x i32> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i32 @llvm.experimental.vector.reduce.and.i32.v2i32(<2 x i32> undef)
@@ -92,67 +92,67 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i16'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i16'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.and.i16.v8i16(<8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.and.i16.v16i16(<16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.and.i16.v32i16(<32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.and.i16.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V4  = call i16 @llvm.experimental.vector.reduce.and.i16.v4i16(<4 x i16> undef)
@@ -165,67 +165,67 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i8'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.and.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.and.i8.v32i8(<32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.and.i8.v64i8(<64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.and.i8.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V8   = call i8 @llvm.experimental.vector.reduce.and.i8.v8i8(<8 x i8> undef)
@@ -239,90 +239,90 @@
 define i32 @reduce_i1(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i1'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i1'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i1'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i1'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i1'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i1'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 135 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 154 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i1'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 775 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 777 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 357 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 838 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 840 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i1'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 135 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.and.i1.v2i1(<2 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.and.i1.v4i1(<4 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.and.i1.v8i1(<8 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.and.i1.v16i1(<16 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.and.i1.v32i1(<32 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 154 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.and.i1.v64i1(<64 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.and.i1.v128i1(<128 x i1> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1   = call i1 @llvm.experimental.vector.reduce.and.i1.v1i1(<1 x i1> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-mul.ll b/llvm/test/Analysis/CostModel/X86/reduce-mul.ll
index fe65501..0aee277 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-mul.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-mul.ll

@@ -11,50 +11,50 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE-LABEL: 'reduce_i64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 122 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 123 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 149 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i64'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i64'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i64'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.mul.i64.v2i64(<2 x i64> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.mul.i64.v4i64(<4 x i64> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.mul.i64.v8i64(<8 x i64> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.mul.i64.v16i64(<16 x i64> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.mul.i64.v1i64(<1 x i64> undef)
@@ -67,67 +67,67 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 102 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 102 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i32'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i32'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i32'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i32'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i32'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.mul.i32.v4i32(<4 x i32> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.mul.i32.v8i32(<8 x i32> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.mul.i32.v16i32(<16 x i32> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.mul.i32.v32i32(<32 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i32 @llvm.experimental.vector.reduce.mul.i32.v2i32(<2 x i32> undef)
@@ -140,67 +140,67 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i16'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i16'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.mul.i16.v8i16(<8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.mul.i16.v16i16(<16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.mul.i16.v32i16(<32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.mul.i16.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V4  = call i16 @llvm.experimental.vector.reduce.mul.i16.v4i16(<4 x i16> undef)
@@ -213,67 +213,67 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 113 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 161 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 257 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 272 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 221 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 236 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 221 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 236 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 171 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 223 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 327 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 202 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 254 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 358 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 208 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 137 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 171 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 239 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i8'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 117 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 115 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 137 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 178 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 200 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 164 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.mul.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 117 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.mul.i8.v32i8(<32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.mul.i8.v64i8(<64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 195 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.mul.i8.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V8   = call i8 @llvm.experimental.vector.reduce.mul.i8.v8i8(<8 x i8> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-or.ll b/llvm/test/Analysis/CostModel/X86/reduce-or.ll
index d742a28..0360850 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-or.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-or.ll

@@ -11,34 +11,34 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE-LABEL: 'reduce_i64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.or.i64.v1i64(<1 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.or.i64.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.or.i64.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.or.i64.v1i64(<1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.or.i64.v2i64(<2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.or.i64.v4i64(<4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.or.i64.v8i64(<8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.or.i64.v16i64(<16 x i64> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.or.i64.v1i64(<1 x i64> undef)
@@ -51,35 +51,35 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE-LABEL: 'reduce_i32'
-; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i32'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i32'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.or.i32.v4i32(<4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.or.i32.v8i32(<8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.or.i32.v16i32(<16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.or.i32.v32i32(<32 x i32> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i32 @llvm.experimental.vector.reduce.or.i32.v2i32(<2 x i32> undef)
@@ -92,67 +92,67 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i16'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i16'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.or.i16.v8i16(<8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.or.i16.v16i16(<16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.or.i16.v32i16(<32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.or.i16.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V4  = call i16 @llvm.experimental.vector.reduce.or.i16.v4i16(<4 x i16> undef)
@@ -165,67 +165,67 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i8'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.or.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.or.i8.v32i8(<32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.or.i8.v64i8(<64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.or.i8.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V8   = call i8 @llvm.experimental.vector.reduce.or.i8.v8i8(<8 x i8> undef)
@@ -239,90 +239,90 @@
 define i32 @reduce_i1(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i1'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i1'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i1'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i1'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i1'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i1'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 135 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 154 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i1'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 775 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 777 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 357 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 838 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 840 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i1'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 135 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.or.i1.v2i1(<2 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.or.i1.v4i1(<4 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.or.i1.v8i1(<8 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.or.i1.v16i1(<16 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.or.i1.v32i1(<32 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 154 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.or.i1.v64i1(<64 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.or.i1.v128i1(<128 x i1> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1   = call i1 @llvm.experimental.vector.reduce.or.i1.v1i1(<1 x i1> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-smax.ll b/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
index c516fc5..3b1a09e 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-smax.ll

@@ -11,18 +11,18 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smax.i64.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.i64.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.i64.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.i64.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.i64.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.i64.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.i64.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.i64.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.i64.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smax.i64.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.i64.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.i64.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.i64.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.i64.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.i64.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.i64.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.i64.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.i64.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -36,9 +36,9 @@
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smax.i64.v1i64(<1 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
@@ -67,19 +67,19 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.i32.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.i32.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.i32.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.i32.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smax.i32.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.i32.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.i32.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.i32.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.i32.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smax.i32.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.i32.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.i32.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.i32.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.i32.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smax.i32.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.i32.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.i32.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.i32.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.i32.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smax.i32.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i32'
@@ -124,19 +124,19 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
@@ -150,9 +150,9 @@
 ; AVX1-LABEL: 'reduce_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
@@ -175,8 +175,8 @@
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.i16.v4i16(<4 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.i16.v8i16(<8 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
@@ -197,40 +197,40 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
@@ -238,7 +238,7 @@
 ;
 ; AVX512F-LABEL: 'reduce_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
@@ -246,15 +246,15 @@
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 253 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 257 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.i8.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.i8.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.i8.v64i8(<64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.i8.v128i8(<128 x i8> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-smin.ll b/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
index 69476c4..077f0c5 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-smin.ll

@@ -11,18 +11,18 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smin.i64.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.i64.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.i64.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.i64.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.i64.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.i64.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.i64.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.i64.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.i64.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smin.i64.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.i64.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.i64.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.i64.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.i64.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.i64.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.i64.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.i64.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.i64.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -36,9 +36,9 @@
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smin.i64.v1i64(<1 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
@@ -67,19 +67,19 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.i32.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.i32.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.i32.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.i32.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smin.i32.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.i32.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.i32.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.i32.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.i32.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smin.i32.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.i32.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.i32.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.i32.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.i32.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smin.i32.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.i32.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.i32.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.i32.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.i32.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smin.i32.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i32'
@@ -124,19 +124,19 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
@@ -150,9 +150,9 @@
 ; AVX1-LABEL: 'reduce_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
@@ -175,8 +175,8 @@
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.i16.v4i16(<4 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.i16.v8i16(<8 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
@@ -197,40 +197,40 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
@@ -238,7 +238,7 @@
 ;
 ; AVX512F-LABEL: 'reduce_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
@@ -246,15 +246,15 @@
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 253 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 257 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.i8.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.i8.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.i8.v64i8(<64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.i8.v128i8(<128 x i8> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-umax.ll b/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
index 629d716..f8eb9b6 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-umax.ll

@@ -11,18 +11,18 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umax.i64.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.i64.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.i64.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.i64.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.i64.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.i64.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.i64.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.i64.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.i64.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umax.i64.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.i64.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.i64.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.i64.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.i64.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.i64.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.i64.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.i64.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.i64.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -36,9 +36,9 @@
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umax.i64.v1i64(<1 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
@@ -67,19 +67,19 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.i32.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.i32.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.i32.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.i32.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umax.i32.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.i32.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.i32.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.i32.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.i32.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umax.i32.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.i32.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.i32.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.i32.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.i32.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umax.i32.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.i32.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.i32.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.i32.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.i32.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umax.i32.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i32'
@@ -124,19 +124,19 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
@@ -150,9 +150,9 @@
 ; AVX1-LABEL: 'reduce_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
@@ -175,8 +175,8 @@
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.i16.v4i16(<4 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.i16.v8i16(<8 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
@@ -197,40 +197,40 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
@@ -238,7 +238,7 @@
 ;
 ; AVX512F-LABEL: 'reduce_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
@@ -246,15 +246,15 @@
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 253 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 257 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.i8.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.i8.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.i8.v64i8(<64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.i8.v128i8(<128 x i8> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-umin.ll b/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
index 08732a1..ce2cbe5 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-umin.ll

@@ -11,18 +11,18 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umin.i64.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.i64.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.i64.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.i64.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.i64.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.i64.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.i64.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.i64.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.i64.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umin.i64.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.i64.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.i64.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.i64.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.i64.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.i64.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.i64.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 71 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.i64.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 143 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.i64.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -36,9 +36,9 @@
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umin.i64.v1i64(<1 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
@@ -67,19 +67,19 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.i32.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.i32.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.i32.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.i32.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umin.i32.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.i32.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.i32.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.i32.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.i32.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umin.i32.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.i32.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.i32.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.i32.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.i32.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umin.i32.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.i32.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.i32.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.i32.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.i32.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umin.i32.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i32'
@@ -124,19 +124,19 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
@@ -150,9 +150,9 @@
 ; AVX1-LABEL: 'reduce_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 111 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 131 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
@@ -175,8 +175,8 @@
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.i16.v4i16(<4 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.i16.v8i16(<8 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
@@ -197,40 +197,40 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 97 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 101 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 109 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 99 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 172 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
@@ -238,7 +238,7 @@
 ;
 ; AVX512F-LABEL: 'reduce_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
@@ -246,15 +246,15 @@
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 253 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 257 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.i8.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.i8.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.i8.v64i8(<64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.i8.v128i8(<128 x i8> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduce-xor.ll b/llvm/test/Analysis/CostModel/X86/reduce-xor.ll
index 784b602..35c3e3e 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-xor.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-xor.ll

@@ -11,34 +11,34 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE-LABEL: 'reduce_i64'
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v1i64(<1 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.xor.i64.v1i64(<1 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.xor.i64.v2i64(<2 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.xor.i64.v4i64(<4 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.xor.i64.v8i64(<8 x i64> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.xor.i64.v16i64(<16 x i64> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.xor.i64.v1i64(<1 x i64> undef)
@@ -51,35 +51,35 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE-LABEL: 'reduce_i32'
-; SSE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
-; SSE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
 ; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i32'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i32'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i32'
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.xor.i32.v4i32(<4 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.xor.i32.v8i32(<8 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.xor.i32.v16i32(<16 x i32> undef)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.xor.i32.v32i32(<32 x i32> undef)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i32 @llvm.experimental.vector.reduce.xor.i32.v2i32(<2 x i32> undef)
@@ -92,67 +92,67 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i16'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i16'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i16'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i16'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.xor.i16.v8i16(<8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.xor.i16.v16i16(<16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.xor.i16.v32i16(<32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.xor.i16.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V4  = call i16 @llvm.experimental.vector.reduce.xor.i16.v4i16(<4 x i16> undef)
@@ -165,67 +165,67 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i8'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i8'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i8'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i8'
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i8'
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.xor.i8.v16i8(<16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.xor.i8.v32i8(<32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.xor.i8.v64i8(<64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.xor.i8.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V8   = call i8 @llvm.experimental.vector.reduce.xor.i8.v8i8(<8 x i8> undef)
@@ -239,90 +239,90 @@
 define i32 @reduce_i1(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i1'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i1'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i1'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i1'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i1'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'reduce_i1'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 135 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 154 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i1'
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 326 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 775 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 777 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 357 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 838 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 840 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; AVX512BW-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512DQ-LABEL: 'reduce_i1'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 133 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 135 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 139 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i1 @llvm.experimental.vector.reduce.xor.i1.v2i1(<2 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i1 @llvm.experimental.vector.reduce.xor.i1.v4i1(<4 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8 = call i1 @llvm.experimental.vector.reduce.xor.i1.v8i1(<8 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V16 = call i1 @llvm.experimental.vector.reduce.xor.i1.v16i1(<16 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %V32 = call i1 @llvm.experimental.vector.reduce.xor.i1.v32i1(<32 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 154 for instruction: %V64 = call i1 @llvm.experimental.vector.reduce.xor.i1.v64i1(<64 x i1> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 162 for instruction: %V128 = call i1 @llvm.experimental.vector.reduce.xor.i1.v128i1(<128 x i1> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V1   = call i1 @llvm.experimental.vector.reduce.xor.i1.v1i1(<1 x i1> undef)

diff --git a/llvm/test/Analysis/CostModel/X86/reduction.ll b/llvm/test/Analysis/CostModel/X86/reduction.ll
index f379714..6025dd94 100644
--- a/llvm/test/Analysis/CostModel/X86/reduction.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduction.ll

@@ -15,7 +15,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSSE3-LABEL: 'reduction_cost_float'
@@ -23,7 +23,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSE42-LABEL: 'reduction_cost_float'
@@ -59,7 +59,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.2 = add <8 x i32> %bin.rdx, %rdx.shuf.2
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf.3 = shufflevector <8 x i32> %bin.rdx.2, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.3 = add <8 x i32> %bin.rdx.2, %rdx.shuf.3
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <8 x i32> %bin.rdx.3, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x i32> %bin.rdx.3, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSSE3-LABEL: 'reduction_cost_int'
@@ -69,7 +69,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.2 = add <8 x i32> %bin.rdx, %rdx.shuf.2
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf.3 = shufflevector <8 x i32> %bin.rdx.2, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.3 = add <8 x i32> %bin.rdx.2, %rdx.shuf.3
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <8 x i32> %bin.rdx.3, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x i32> %bin.rdx.3, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSE42-LABEL: 'reduction_cost_int'
@@ -127,7 +127,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r2 = fadd float %r, %f1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r2
 ;
@@ -138,7 +138,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r2 = fadd float %r, %f1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r2
 ;
@@ -188,7 +188,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r2 = fadd float %r, %f1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r2
 ;
@@ -199,7 +199,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r2 = fadd float %r, %f1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r2
 ;
@@ -248,7 +248,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.1 = fadd <4 x float> %bin.rdx.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r2 = fadd float %r, %f1
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r2
 ;
@@ -258,7 +258,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx.1 = fadd <4 x float> %bin.rdx.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx.1, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %r2 = fadd float %r, %f1
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r2
 ;
@@ -300,13 +300,13 @@
 ; SSE2-LABEL: 'no_pairwise_reduction2double'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 1, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <2 x double> %rdx, %rdx.shuf
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %r = extractelement <2 x double> %bin.rdx, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x double> %bin.rdx, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction2double'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 1, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <2 x double> %rdx, %rdx.shuf
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %r = extractelement <2 x double> %bin.rdx, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x double> %bin.rdx, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction2double'
@@ -334,7 +334,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction4float'
@@ -342,7 +342,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction4float'
@@ -376,7 +376,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx = fadd <4 x double> %rdx, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf7 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx8 = fadd <4 x double> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction4double'
@@ -384,7 +384,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx = fadd <4 x double> %rdx, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf7 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx8 = fadd <4 x double> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction4double'
@@ -428,7 +428,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx = fadd <8 x float> %bin.rdx4, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf7 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx8 = fadd <8 x float> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x float> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %r = extractelement <8 x float> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction8float'
@@ -438,7 +438,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx = fadd <8 x float> %bin.rdx4, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf7 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx8 = fadd <8 x float> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x float> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %r = extractelement <8 x float> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction8float'
@@ -486,13 +486,13 @@
 ; SSE2-LABEL: 'no_pairwise_reduction2i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 1, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx = add <2 x i64> %rdx, %rdx.shuf
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %r = extractelement <2 x i64> %bin.rdx, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x i64> %bin.rdx, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction2i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 1, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx = add <2 x i64> %rdx, %rdx.shuf
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %r = extractelement <2 x i64> %bin.rdx, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x i64> %bin.rdx, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction2i64'
@@ -520,7 +520,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx = add <4 x i32> %rdx, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <4 x i32> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction4i32'
@@ -528,7 +528,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx = add <4 x i32> %rdx, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <4 x i32> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction4i32'
@@ -562,7 +562,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = add <4 x i64> %rdx, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf7 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = add <4 x i64> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction4i64'
@@ -570,7 +570,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = add <4 x i64> %rdx, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf7 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = add <4 x i64> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction4i64'
@@ -614,7 +614,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx = add <8 x i16> %bin.rdx4, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %rdx.shuf7 = shufflevector <8 x i16> %bin.rdx, <8 x i16> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <8 x i16> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %r = extractelement <8 x i16> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %r = extractelement <8 x i16> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i16 %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction8i16'
@@ -624,7 +624,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx = add <8 x i16> %bin.rdx4, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf7 = shufflevector <8 x i16> %bin.rdx, <8 x i16> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <8 x i16> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <8 x i16> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %r = extractelement <8 x i16> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i16 %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction8i16'
@@ -666,7 +666,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = add <8 x i32> %bin.rdx4, %rdx.shuf
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf7 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = add <8 x i32> %bin.rdx, %rdx.shuf7
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <8 x i32> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x i32> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSSE3-LABEL: 'no_pairwise_reduction8i32'
@@ -676,7 +676,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = add <8 x i32> %bin.rdx4, %rdx.shuf
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf7 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = add <8 x i32> %bin.rdx, %rdx.shuf7
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <8 x i32> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x i32> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSE42-LABEL: 'no_pairwise_reduction8i32'
@@ -725,14 +725,14 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 0, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 1, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <2 x double> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x double> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <2 x double> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction2double'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 0, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 1, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <2 x double> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x double> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <2 x double> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction2double'
@@ -765,7 +765,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction4float'
@@ -775,7 +775,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %r = extractelement <4 x float> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction4float'
@@ -817,7 +817,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.1.1 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx8 = fadd <4 x double> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction4double'
@@ -827,7 +827,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.1.1 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx8 = fadd <4 x double> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <4 x double> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret double %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction4double'
@@ -882,7 +882,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.2.0 = shufflevector <8 x float> %bin.rdx8, <8 x float> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf.2.1 = shufflevector <8 x float> %bin.rdx8, <8 x float> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx9 = fadd <8 x float> %rdx.shuf.2.0, %rdx.shuf.2.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %r = extractelement <8 x float> %bin.rdx9, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %r = extractelement <8 x float> %bin.rdx9, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction8float'
@@ -895,7 +895,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.2.0 = shufflevector <8 x float> %bin.rdx8, <8 x float> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf.2.1 = shufflevector <8 x float> %bin.rdx8, <8 x float> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx9 = fadd <8 x float> %rdx.shuf.2.0, %rdx.shuf.2.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %r = extractelement <8 x float> %bin.rdx9, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %r = extractelement <8 x float> %bin.rdx9, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret float %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction8float'
@@ -956,14 +956,14 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 0, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 1, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <2 x i64> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x i64> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <2 x i64> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction2i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 0, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 1, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <2 x i64> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %r = extractelement <2 x i64> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %r = extractelement <2 x i64> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction2i64'
@@ -996,7 +996,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <4 x i32> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction4i32'
@@ -1006,7 +1006,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx8 = add <4 x i32> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <4 x i32> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction4i32'
@@ -1048,7 +1048,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.1.1 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = add <4 x i64> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction4i64'
@@ -1058,7 +1058,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.1.0 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 0, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.1.1 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx8 = add <4 x i64> %rdx.shuf.1.0, %rdx.shuf.1.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %r = extractelement <4 x i64> %bin.rdx8, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i64 %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction4i64'
@@ -1113,7 +1113,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.2.0 = shufflevector <8 x i16> %bin.rdx8, <8 x i16> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %rdx.shuf.2.1 = shufflevector <8 x i16> %bin.rdx8, <8 x i16> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx9 = add <8 x i16> %rdx.shuf.2.0, %rdx.shuf.2.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %r = extractelement <8 x i16> %bin.rdx9, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %r = extractelement <8 x i16> %bin.rdx9, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i16 %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction8i16'
@@ -1126,7 +1126,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.2.0 = shufflevector <8 x i16> %bin.rdx8, <8 x i16> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.2.1 = shufflevector <8 x i16> %bin.rdx8, <8 x i16> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %bin.rdx9 = add <8 x i16> %rdx.shuf.2.0, %rdx.shuf.2.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %r = extractelement <8 x i16> %bin.rdx9, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %r = extractelement <8 x i16> %bin.rdx9, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i16 %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction8i16'
@@ -1180,7 +1180,7 @@
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.2.0 = shufflevector <8 x i32> %bin.rdx8, <8 x i32> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf.2.1 = shufflevector <8 x i32> %bin.rdx8, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx9 = add <8 x i32> %rdx.shuf.2.0, %rdx.shuf.2.1
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <8 x i32> %bin.rdx9, i32 0
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %r = extractelement <8 x i32> %bin.rdx9, i32 0
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSSE3-LABEL: 'pairwise_reduction8i32'
@@ -1193,7 +1193,7 @@
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %rdx.shuf.2.0 = shufflevector <8 x i32> %bin.rdx8, <8 x i32> undef, <8 x i32> <i32 0, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %rdx.shuf.2.1 = shufflevector <8 x i32> %bin.rdx8, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx9 = add <8 x i32> %rdx.shuf.2.0, %rdx.shuf.2.1
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %r = extractelement <8 x i32> %bin.rdx9, i32 0
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %r = extractelement <8 x i32> %bin.rdx9, i32 0
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 %r
 ;
 ; SSE42-LABEL: 'pairwise_reduction8i32'
commit	8cd9d1b5cebe8a694089ac983038966a2fe6a516	[log] [tgz]
author	Fedor Sergeev <fedor.sergeev@azul.com>	Mon Nov 26 10:17:27 2018 +0000
committer	Fedor Sergeev <fedor.sergeev@azul.com>	Mon Nov 26 10:17:27 2018 +0000
tree	0627872d1fcb453f0793656b68509333d84a24ac
parent	0e0cd5be403a83de607f88e469a4de06051ac39c [diff]