Use vAny type to get rid of Neon intrinsics that differed only in whether
the overloaded vector types allowed floating-point or integer vector elements.
Most of these operations actually depend on the element type, so bitcasting
was not an option.
If you include the vpadd intrinsics that I updated earlier, this gets rid
of 20 intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@78646 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/ARM/vld4.ll b/test/CodeGen/ARM/vld4.ll
index a0f41cf..48125be 100644
--- a/test/CodeGen/ARM/vld4.ll
+++ b/test/CodeGen/ARM/vld4.ll
@@ -8,7 +8,7 @@
define <8 x i8> @vld4i8(i8* %A) nounwind {
;CHECK: vld4i8:
;CHECK: vld4.8
- %tmp1 = call %struct.__builtin_neon_v8qi4 @llvm.arm.neon.vld4i.v8i8(i8* %A)
+ %tmp1 = call %struct.__builtin_neon_v8qi4 @llvm.arm.neon.vld4.v8i8(i8* %A)
%tmp2 = extractvalue %struct.__builtin_neon_v8qi4 %tmp1, 0
%tmp3 = extractvalue %struct.__builtin_neon_v8qi4 %tmp1, 2
%tmp4 = add <8 x i8> %tmp2, %tmp3
@@ -18,7 +18,7 @@
define <4 x i16> @vld4i16(i16* %A) nounwind {
;CHECK: vld4i16:
;CHECK: vld4.16
- %tmp1 = call %struct.__builtin_neon_v4hi4 @llvm.arm.neon.vld4i.v4i16(i16* %A)
+ %tmp1 = call %struct.__builtin_neon_v4hi4 @llvm.arm.neon.vld4.v4i16(i16* %A)
%tmp2 = extractvalue %struct.__builtin_neon_v4hi4 %tmp1, 0
%tmp3 = extractvalue %struct.__builtin_neon_v4hi4 %tmp1, 2
%tmp4 = add <4 x i16> %tmp2, %tmp3
@@ -28,7 +28,7 @@
define <2 x i32> @vld4i32(i32* %A) nounwind {
;CHECK: vld4i32:
;CHECK: vld4.32
- %tmp1 = call %struct.__builtin_neon_v2si4 @llvm.arm.neon.vld4i.v2i32(i32* %A)
+ %tmp1 = call %struct.__builtin_neon_v2si4 @llvm.arm.neon.vld4.v2i32(i32* %A)
%tmp2 = extractvalue %struct.__builtin_neon_v2si4 %tmp1, 0
%tmp3 = extractvalue %struct.__builtin_neon_v2si4 %tmp1, 2
%tmp4 = add <2 x i32> %tmp2, %tmp3
@@ -38,14 +38,14 @@
define <2 x float> @vld4f(float* %A) nounwind {
;CHECK: vld4f:
;CHECK: vld4.32
- %tmp1 = call %struct.__builtin_neon_v2sf4 @llvm.arm.neon.vld4f.v2f32(float* %A)
+ %tmp1 = call %struct.__builtin_neon_v2sf4 @llvm.arm.neon.vld4.v2f32(float* %A)
%tmp2 = extractvalue %struct.__builtin_neon_v2sf4 %tmp1, 0
%tmp3 = extractvalue %struct.__builtin_neon_v2sf4 %tmp1, 2
%tmp4 = add <2 x float> %tmp2, %tmp3
ret <2 x float> %tmp4
}
-declare %struct.__builtin_neon_v8qi4 @llvm.arm.neon.vld4i.v8i8(i8*) nounwind readonly
-declare %struct.__builtin_neon_v4hi4 @llvm.arm.neon.vld4i.v4i16(i8*) nounwind readonly
-declare %struct.__builtin_neon_v2si4 @llvm.arm.neon.vld4i.v2i32(i8*) nounwind readonly
-declare %struct.__builtin_neon_v2sf4 @llvm.arm.neon.vld4f.v2f32(i8*) nounwind readonly
+declare %struct.__builtin_neon_v8qi4 @llvm.arm.neon.vld4.v8i8(i8*) nounwind readonly
+declare %struct.__builtin_neon_v4hi4 @llvm.arm.neon.vld4.v4i16(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2si4 @llvm.arm.neon.vld4.v2i32(i8*) nounwind readonly
+declare %struct.__builtin_neon_v2sf4 @llvm.arm.neon.vld4.v2f32(i8*) nounwind readonly