Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / ba883e980a9c0ce0edfddb0737b2a30a1dec0ef7 / . / llvm / test / Analysis / CostModel / X86 / arith-fp.ll
blob: 1e95e9a5116d27a1fad11b883a4274e85b529b90 [file] [log] [blame]
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]2; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=-sse2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE1
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]3; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE2
4; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+sse4.2 | FileCheck %s --check-prefixes=CHECK,SSE,SSE42
Simon Pilgrim3ede11b2018-04-13 19:12:32 +0000[diff] [blame]5; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx | FileCheck %s --check-prefixes=CHECK,AVX,AVX1
6; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx2 | FileCheck %s --check-prefixes=CHECK,AVX,AVX2
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]7; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512F
8; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=CHECK,AVX512,AVX512BW
Simon Pilgrim34b397a2018-04-13 19:30:15 +0000[diff] [blame]9;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]10; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=slm | FileCheck %s --check-prefixes=SLM
11; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=goldmont | FileCheck %s --check-prefixes=GLM
12; RUN: opt < %s -enable-no-nans-fp-math -cost-model -analyze -mtriple=x86_64-apple-macosx10.8.0 -mcpu=btver2 | FileCheck %s --check-prefixes=BTVER2
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]13
14target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
15target triple = "x86_64-apple-macosx10.8.0"
16
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]17define i32 @fadd(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]18; SSE1-LABEL: 'fadd'
Simon Pilgrimc2054142019-01-04 16:55:57 +0000[diff] [blame]19; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]20; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fadd <4 x float> undef, undef
21; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fadd <8 x float> undef, undef
22; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fadd <16 x float> undef, undef
23; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fadd double undef, undef
24; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fadd <2 x double> undef, undef
25; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fadd <4 x double> undef, undef
26; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fadd <8 x double> undef, undef
27; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
28;
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]29; SSE2-LABEL: 'fadd'
30; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fadd float undef, undef
31; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fadd <4 x float> undef, undef
32; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fadd <8 x float> undef, undef
33; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fadd <16 x float> undef, undef
34; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fadd double undef, undef
35; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fadd <2 x double> undef, undef
36; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fadd <4 x double> undef, undef
37; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fadd <8 x double> undef, undef
38; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
39;
40; SSE42-LABEL: 'fadd'
41; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
42; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
43; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fadd <8 x float> undef, undef
44; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fadd <16 x float> undef, undef
45; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
46; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
47; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fadd <4 x double> undef, undef
48; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fadd <8 x double> undef, undef
49; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
50;
51; AVX1-LABEL: 'fadd'
52; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
53; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
54; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fadd <8 x float> undef, undef
55; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fadd <16 x float> undef, undef
56; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
57; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
58; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fadd <4 x double> undef, undef
59; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fadd <8 x double> undef, undef
60; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
61;
62; AVX2-LABEL: 'fadd'
63; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
64; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
65; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fadd <8 x float> undef, undef
66; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fadd <16 x float> undef, undef
67; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
68; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
69; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fadd <4 x double> undef, undef
70; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fadd <8 x double> undef, undef
71; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
72;
73; AVX512-LABEL: 'fadd'
74; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
75; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
76; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fadd <8 x float> undef, undef
77; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fadd <16 x float> undef, undef
78; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
79; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
80; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fadd <4 x double> undef, undef
81; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fadd <8 x double> undef, undef
82; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
83;
84; SLM-LABEL: 'fadd'
85; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
86; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
87; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fadd <8 x float> undef, undef
88; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fadd <16 x float> undef, undef
89; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
90; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fadd <2 x double> undef, undef
91; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fadd <4 x double> undef, undef
92; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fadd <8 x double> undef, undef
93; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
94;
95; GLM-LABEL: 'fadd'
96; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
97; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
98; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fadd <8 x float> undef, undef
99; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fadd <16 x float> undef, undef
100; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
101; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
102; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fadd <4 x double> undef, undef
103; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fadd <8 x double> undef, undef
104; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
105;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]106; BTVER2-LABEL: 'fadd'
107; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
108; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fadd <4 x float> undef, undef
109; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fadd <8 x float> undef, undef
110; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fadd <16 x float> undef, undef
111; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fadd double undef, undef
112; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fadd <2 x double> undef, undef
113; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fadd <4 x double> undef, undef
114; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fadd <8 x double> undef, undef
115; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
116;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]117 %F32 = fadd float undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]118 %V4F32 = fadd <4 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]119 %V8F32 = fadd <8 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]120 %V16F32 = fadd <16 x float> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]121
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]122 %F64 = fadd double undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]123 %V2F64 = fadd <2 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]124 %V4F64 = fadd <4 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]125 %V8F64 = fadd <8 x double> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]126
127 ret i32 undef
128}
129
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]130define i32 @fsub(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]131; SSE1-LABEL: 'fsub'
Simon Pilgrimc2054142019-01-04 16:55:57 +0000[diff] [blame]132; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]133; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fsub <4 x float> undef, undef
134; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fsub <8 x float> undef, undef
135; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fsub <16 x float> undef, undef
136; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fsub double undef, undef
137; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fsub <2 x double> undef, undef
138; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fsub <4 x double> undef, undef
139; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fsub <8 x double> undef, undef
140; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
141;
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]142; SSE2-LABEL: 'fsub'
143; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fsub float undef, undef
144; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fsub <4 x float> undef, undef
145; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fsub <8 x float> undef, undef
146; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fsub <16 x float> undef, undef
147; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fsub double undef, undef
148; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fsub <2 x double> undef, undef
149; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fsub <4 x double> undef, undef
150; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fsub <8 x double> undef, undef
151; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
152;
153; SSE42-LABEL: 'fsub'
154; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
155; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
156; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> undef, undef
157; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> undef, undef
158; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
159; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
160; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> undef, undef
161; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> undef, undef
162; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
163;
164; AVX1-LABEL: 'fsub'
165; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
166; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
167; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> undef, undef
168; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> undef, undef
169; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
170; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
171; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> undef, undef
172; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> undef, undef
173; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
174;
175; AVX2-LABEL: 'fsub'
176; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
177; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
178; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> undef, undef
179; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fsub <16 x float> undef, undef
180; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
181; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
182; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> undef, undef
183; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fsub <8 x double> undef, undef
184; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
185;
186; AVX512-LABEL: 'fsub'
187; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
188; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
189; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> undef, undef
190; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> undef, undef
191; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
192; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
193; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> undef, undef
194; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> undef, undef
195; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
196;
197; SLM-LABEL: 'fsub'
198; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
199; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
200; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> undef, undef
201; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> undef, undef
202; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
203; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fsub <2 x double> undef, undef
204; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fsub <4 x double> undef, undef
205; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fsub <8 x double> undef, undef
206; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
207;
208; GLM-LABEL: 'fsub'
209; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
210; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
211; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> undef, undef
212; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> undef, undef
213; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
214; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
215; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> undef, undef
216; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> undef, undef
217; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
218;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]219; BTVER2-LABEL: 'fsub'
220; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
221; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> undef, undef
222; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> undef, undef
223; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> undef, undef
224; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double undef, undef
225; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> undef, undef
226; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> undef, undef
227; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> undef, undef
228; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
229;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]230 %F32 = fsub float undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]231 %V4F32 = fsub <4 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]232 %V8F32 = fsub <8 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]233 %V16F32 = fsub <16 x float> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]234
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]235 %F64 = fsub double undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]236 %V2F64 = fsub <2 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]237 %V4F64 = fsub <4 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]238 %V8F64 = fsub <8 x double> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]239
240 ret i32 undef
241}
242
Craig Topperba883e92019-05-27 05:27:57 +0000[diff] [blame^]243define i32 @fneg_idiom(i32 %arg) {
244; SSE1-LABEL: 'fneg_idiom'
245; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
246; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
247; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
248; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
249; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fsub double -0.000000e+00, undef
250; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
251; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
252; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
253; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
254;
255; SSE2-LABEL: 'fneg_idiom'
256; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fsub float -0.000000e+00, undef
257; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
258; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
259; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
260; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fsub double -0.000000e+00, undef
261; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
262; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
263; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
264; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
265;
266; SSE42-LABEL: 'fneg_idiom'
267; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
268; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
269; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
270; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
271; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
272; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
273; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
274; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
275; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
276;
277; AVX1-LABEL: 'fneg_idiom'
278; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
279; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
280; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
281; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
282; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
283; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
284; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
285; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
286; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
287;
288; AVX2-LABEL: 'fneg_idiom'
289; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
290; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
291; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
292; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
293; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
294; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
295; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
296; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
297; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
298;
299; AVX512-LABEL: 'fneg_idiom'
300; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
301; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
302; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
303; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
304; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
305; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
306; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
307; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
308; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
309;
310; SLM-LABEL: 'fneg_idiom'
311; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
312; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
313; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
314; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
315; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
316; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
317; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
318; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
319; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
320;
321; GLM-LABEL: 'fneg_idiom'
322; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
323; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
324; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
325; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
326; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
327; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
328; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
329; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
330; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
331;
332; BTVER2-LABEL: 'fneg_idiom'
333; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float -0.000000e+00, undef
334; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
335; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
336; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, undef
337; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fsub double -0.000000e+00, undef
338; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, undef
339; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
340; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fsub <8 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, undef
341; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
342;
343 %F32 = fsub float -0.0, undef
344 %V4F32 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, undef
345 %V8F32 = fsub <8 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
346 %V16F32 = fsub <16 x float> <float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0, float -0.0>, undef
347
348 %F64 = fsub double -0.0, undef
349 %V2F64 = fsub <2 x double> <double -0.0, double -0.0>, undef
350 %V4F64 = fsub <4 x double> <double -0.0, double -0.0, double -0.0, double -0.0>, undef
351 %V8F64 = fsub <8 x double> <double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0>, undef
352
353 ret i32 undef
354}
355
356define i32 @fneg(i32 %arg) {
357; CHECK-LABEL: 'fneg'
358; CHECK-NEXT: Cost Model: Unknown cost for instruction: %F32 = fneg float undef
359; CHECK-NEXT: Cost Model: Unknown cost for instruction: %V4F32 = fneg <4 x float> undef
360; CHECK-NEXT: Cost Model: Unknown cost for instruction: %V8F32 = fneg <8 x float> undef
361; CHECK-NEXT: Cost Model: Unknown cost for instruction: %V16F32 = fneg <16 x float> undef
362; CHECK-NEXT: Cost Model: Unknown cost for instruction: %F64 = fneg double undef
363; CHECK-NEXT: Cost Model: Unknown cost for instruction: %V2F64 = fneg <2 x double> undef
364; CHECK-NEXT: Cost Model: Unknown cost for instruction: %V4F64 = fneg <4 x double> undef
365; CHECK-NEXT: Cost Model: Unknown cost for instruction: %V8F64 = fneg <8 x double> undef
366; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
367;
368; SLM-LABEL: 'fneg'
369; SLM-NEXT: Cost Model: Unknown cost for instruction: %F32 = fneg float undef
370; SLM-NEXT: Cost Model: Unknown cost for instruction: %V4F32 = fneg <4 x float> undef
371; SLM-NEXT: Cost Model: Unknown cost for instruction: %V8F32 = fneg <8 x float> undef
372; SLM-NEXT: Cost Model: Unknown cost for instruction: %V16F32 = fneg <16 x float> undef
373; SLM-NEXT: Cost Model: Unknown cost for instruction: %F64 = fneg double undef
374; SLM-NEXT: Cost Model: Unknown cost for instruction: %V2F64 = fneg <2 x double> undef
375; SLM-NEXT: Cost Model: Unknown cost for instruction: %V4F64 = fneg <4 x double> undef
376; SLM-NEXT: Cost Model: Unknown cost for instruction: %V8F64 = fneg <8 x double> undef
377; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
378;
379; GLM-LABEL: 'fneg'
380; GLM-NEXT: Cost Model: Unknown cost for instruction: %F32 = fneg float undef
381; GLM-NEXT: Cost Model: Unknown cost for instruction: %V4F32 = fneg <4 x float> undef
382; GLM-NEXT: Cost Model: Unknown cost for instruction: %V8F32 = fneg <8 x float> undef
383; GLM-NEXT: Cost Model: Unknown cost for instruction: %V16F32 = fneg <16 x float> undef
384; GLM-NEXT: Cost Model: Unknown cost for instruction: %F64 = fneg double undef
385; GLM-NEXT: Cost Model: Unknown cost for instruction: %V2F64 = fneg <2 x double> undef
386; GLM-NEXT: Cost Model: Unknown cost for instruction: %V4F64 = fneg <4 x double> undef
387; GLM-NEXT: Cost Model: Unknown cost for instruction: %V8F64 = fneg <8 x double> undef
388; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
389;
390; BTVER2-LABEL: 'fneg'
391; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %F32 = fneg float undef
392; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %V4F32 = fneg <4 x float> undef
393; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %V8F32 = fneg <8 x float> undef
394; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %V16F32 = fneg <16 x float> undef
395; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %F64 = fneg double undef
396; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %V2F64 = fneg <2 x double> undef
397; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %V4F64 = fneg <4 x double> undef
398; BTVER2-NEXT: Cost Model: Unknown cost for instruction: %V8F64 = fneg <8 x double> undef
399; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
400;
401 %F32 = fneg float undef
402 %V4F32 = fneg <4 x float> undef
403 %V8F32 = fneg <8 x float> undef
404 %V16F32 = fneg <16 x float> undef
405
406 %F64 = fneg double undef
407 %V2F64 = fneg <2 x double> undef
408 %V4F64 = fneg <4 x double> undef
409 %V8F64 = fneg <8 x double> undef
410
411 ret i32 undef
412}
413
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]414define i32 @fmul(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]415; SSE1-LABEL: 'fmul'
416; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fmul float undef, undef
417; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fmul <4 x float> undef, undef
418; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fmul <8 x float> undef, undef
419; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fmul <16 x float> undef, undef
420; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fmul double undef, undef
421; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fmul <2 x double> undef, undef
422; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fmul <4 x double> undef, undef
423; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fmul <8 x double> undef, undef
424; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
425;
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]426; SSE2-LABEL: 'fmul'
427; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fmul float undef, undef
428; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fmul <4 x float> undef, undef
429; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fmul <8 x float> undef, undef
430; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fmul <16 x float> undef, undef
431; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fmul double undef, undef
432; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fmul <2 x double> undef, undef
433; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fmul <4 x double> undef, undef
434; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fmul <8 x double> undef, undef
435; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
436;
437; SSE42-LABEL: 'fmul'
438; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
439; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
440; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
441; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
442; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
443; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
444; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
445; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
446; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
447;
448; AVX1-LABEL: 'fmul'
449; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
450; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
451; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
452; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
453; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
454; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
455; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
456; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
457; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
458;
459; AVX2-LABEL: 'fmul'
460; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
461; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
462; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
463; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fmul <16 x float> undef, undef
464; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
465; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
466; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
467; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fmul <8 x double> undef, undef
468; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
469;
470; AVX512-LABEL: 'fmul'
471; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
472; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
473; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
474; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fmul <16 x float> undef, undef
475; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
476; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
477; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
478; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fmul <8 x double> undef, undef
479; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
480;
481; SLM-LABEL: 'fmul'
482; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
483; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fmul <4 x float> undef, undef
484; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fmul <8 x float> undef, undef
485; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fmul <16 x float> undef, undef
486; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fmul double undef, undef
487; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fmul <2 x double> undef, undef
488; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fmul <4 x double> undef, undef
489; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fmul <8 x double> undef, undef
490; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
491;
492; GLM-LABEL: 'fmul'
493; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
494; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
495; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
496; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
497; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
498; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
499; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
500; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
501; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
502;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]503; BTVER2-LABEL: 'fmul'
504; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
505; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
506; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
507; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
508; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
509; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
510; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
511; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
512; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
513;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]514 %F32 = fmul float undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]515 %V4F32 = fmul <4 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]516 %V8F32 = fmul <8 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]517 %V16F32 = fmul <16 x float> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]518
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]519 %F64 = fmul double undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]520 %V2F64 = fmul <2 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]521 %V4F64 = fmul <4 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]522 %V8F64 = fmul <8 x double> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]523
524 ret i32 undef
525}
526
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]527define i32 @fdiv(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]528; SSE1-LABEL: 'fdiv'
529; SSE1-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %F32 = fdiv float undef, undef
530; SSE1-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V4F32 = fdiv <4 x float> undef, undef
531; SSE1-NEXT: Cost Model: Found an estimated cost of 68 for instruction: %V8F32 = fdiv <8 x float> undef, undef
532; SSE1-NEXT: Cost Model: Found an estimated cost of 136 for instruction: %V16F32 = fdiv <16 x float> undef, undef
533; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fdiv double undef, undef
534; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
535; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fdiv <4 x double> undef, undef
536; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fdiv <8 x double> undef, undef
537; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
538;
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]539; SSE2-LABEL: 'fdiv'
540; SSE2-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %F32 = fdiv float undef, undef
541; SSE2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4F32 = fdiv <4 x float> undef, undef
542; SSE2-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8F32 = fdiv <8 x float> undef, undef
543; SSE2-NEXT: Cost Model: Found an estimated cost of 156 for instruction: %V16F32 = fdiv <16 x float> undef, undef
544; SSE2-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %F64 = fdiv double undef, undef
545; SSE2-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %V2F64 = fdiv <2 x double> undef, undef
546; SSE2-NEXT: Cost Model: Found an estimated cost of 138 for instruction: %V4F64 = fdiv <4 x double> undef, undef
547; SSE2-NEXT: Cost Model: Found an estimated cost of 276 for instruction: %V8F64 = fdiv <8 x double> undef, undef
548; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
549;
550; SSE42-LABEL: 'fdiv'
551; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
552; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
553; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
554; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
555; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
556; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
557; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
558; SSE42-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
559; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
560;
561; AVX1-LABEL: 'fdiv'
562; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
563; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
564; AVX1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
565; AVX1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
566; AVX1-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
567; AVX1-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
568; AVX1-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
569; AVX1-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
570; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
571;
572; AVX2-LABEL: 'fdiv'
573; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = fdiv float undef, undef
574; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = fdiv <4 x float> undef, undef
575; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = fdiv <8 x float> undef, undef
576; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = fdiv <16 x float> undef, undef
577; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = fdiv double undef, undef
578; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = fdiv <2 x double> undef, undef
579; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = fdiv <4 x double> undef, undef
580; AVX2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = fdiv <8 x double> undef, undef
581; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
582;
583; AVX512-LABEL: 'fdiv'
584; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = fdiv float undef, undef
585; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = fdiv <4 x float> undef, undef
586; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = fdiv <8 x float> undef, undef
587; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fdiv <16 x float> undef, undef
588; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = fdiv double undef, undef
589; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = fdiv <2 x double> undef, undef
590; AVX512-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = fdiv <4 x double> undef, undef
591; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fdiv <8 x double> undef, undef
592; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
593;
594; SLM-LABEL: 'fdiv'
595; SLM-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %F32 = fdiv float undef, undef
596; SLM-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4F32 = fdiv <4 x float> undef, undef
597; SLM-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8F32 = fdiv <8 x float> undef, undef
598; SLM-NEXT: Cost Model: Found an estimated cost of 156 for instruction: %V16F32 = fdiv <16 x float> undef, undef
599; SLM-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %F64 = fdiv double undef, undef
600; SLM-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %V2F64 = fdiv <2 x double> undef, undef
601; SLM-NEXT: Cost Model: Found an estimated cost of 138 for instruction: %V4F64 = fdiv <4 x double> undef, undef
602; SLM-NEXT: Cost Model: Found an estimated cost of 276 for instruction: %V8F64 = fdiv <8 x double> undef, undef
603; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
604;
605; GLM-LABEL: 'fdiv'
606; GLM-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %F32 = fdiv float undef, undef
607; GLM-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %V4F32 = fdiv <4 x float> undef, undef
608; GLM-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %V8F32 = fdiv <8 x float> undef, undef
609; GLM-NEXT: Cost Model: Found an estimated cost of 140 for instruction: %V16F32 = fdiv <16 x float> undef, undef
610; GLM-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %F64 = fdiv double undef, undef
611; GLM-NEXT: Cost Model: Found an estimated cost of 65 for instruction: %V2F64 = fdiv <2 x double> undef, undef
612; GLM-NEXT: Cost Model: Found an estimated cost of 130 for instruction: %V4F64 = fdiv <4 x double> undef, undef
613; GLM-NEXT: Cost Model: Found an estimated cost of 260 for instruction: %V8F64 = fdiv <8 x double> undef, undef
614; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
615;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]616; BTVER2-LABEL: 'fdiv'
617; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
618; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
619; BTVER2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
620; BTVER2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
621; BTVER2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
622; BTVER2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
623; BTVER2-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
624; BTVER2-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
625; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
626;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]627 %F32 = fdiv float undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]628 %V4F32 = fdiv <4 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]629 %V8F32 = fdiv <8 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]630 %V16F32 = fdiv <16 x float> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]631
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]632 %F64 = fdiv double undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]633 %V2F64 = fdiv <2 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]634 %V4F64 = fdiv <4 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]635 %V8F64 = fdiv <8 x double> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]636
637 ret i32 undef
638}
639
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]640define i32 @frem(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]641; SSE1-LABEL: 'frem'
642; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
643; SSE1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
644; SSE1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
645; SSE1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
646; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
647; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
648; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = frem <4 x double> undef, undef
649; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = frem <8 x double> undef, undef
650; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
651;
652; SSE2-LABEL: 'frem'
653; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
654; SSE2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
655; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
656; SSE2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
657; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
658; SSE2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
659; SSE2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
660; SSE2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
661; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
662;
663; SSE42-LABEL: 'frem'
664; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
665; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
666; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
667; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
668; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
669; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
670; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
671; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
672; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]673;
674; AVX-LABEL: 'frem'
675; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
676; AVX-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
677; AVX-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
678; AVX-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = frem <16 x float> undef, undef
679; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
680; AVX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
681; AVX-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
682; AVX-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = frem <8 x double> undef, undef
683; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
684;
685; AVX512-LABEL: 'frem'
686; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
687; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
688; AVX512-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
689; AVX512-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16F32 = frem <16 x float> undef, undef
690; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
691; AVX512-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
692; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
693; AVX512-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F64 = frem <8 x double> undef, undef
694; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
695;
696; SLM-LABEL: 'frem'
697; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
698; SLM-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
699; SLM-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
700; SLM-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
701; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
702; SLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
703; SLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
704; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
705; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
706;
707; GLM-LABEL: 'frem'
708; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
709; GLM-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
710; GLM-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
711; GLM-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
712; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
713; GLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
714; GLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
715; GLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
716; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
717;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]718; BTVER2-LABEL: 'frem'
719; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
720; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
721; BTVER2-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
722; BTVER2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = frem <16 x float> undef, undef
723; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
724; BTVER2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
725; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
726; BTVER2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = frem <8 x double> undef, undef
727; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
728;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]729 %F32 = frem float undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]730 %V4F32 = frem <4 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]731 %V8F32 = frem <8 x float> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]732 %V16F32 = frem <16 x float> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]733
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]734 %F64 = frem double undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]735 %V2F64 = frem <2 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]736 %V4F64 = frem <4 x double> undef, undef
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]737 %V8F64 = frem <8 x double> undef, undef
Simon Pilgrim3cd61a02016-08-21 18:34:47 +0000[diff] [blame]738
739 ret i32 undef
740}
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]741
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]742define i32 @fsqrt(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]743; SSE1-LABEL: 'fsqrt'
744; SSE1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
745; SSE1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
746; SSE1-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
747; SSE1-NEXT: Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
748; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
749; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
750; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
751; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
752; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
753;
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]754; SSE2-LABEL: 'fsqrt'
755; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
756; SSE2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
757; SSE2-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
758; SSE2-NEXT: Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
759; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
760; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
761; SSE2-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
762; SSE2-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
763; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
764;
765; SSE42-LABEL: 'fsqrt'
766; SSE42-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
767; SSE42-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
768; SSE42-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
769; SSE42-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
770; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
771; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
772; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
773; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
774; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
775;
776; AVX1-LABEL: 'fsqrt'
777; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
778; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
779; AVX1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
780; AVX1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
781; AVX1-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
782; AVX1-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
783; AVX1-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
784; AVX1-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
785; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
786;
787; AVX2-LABEL: 'fsqrt'
788; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
789; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
790; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
791; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
792; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
793; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
794; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
795; AVX2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
796; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
797;
798; AVX512-LABEL: 'fsqrt'
799; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
800; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
801; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
802; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
803; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
804; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
805; AVX512-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
806; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
807; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
808;
809; SLM-LABEL: 'fsqrt'
810; SLM-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
811; SLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
812; SLM-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
813; SLM-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
814; SLM-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
815; SLM-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
816; SLM-NEXT: Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
817; SLM-NEXT: Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
818; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
819;
820; GLM-LABEL: 'fsqrt'
821; GLM-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
822; GLM-NEXT: Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
823; GLM-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
824; GLM-NEXT: Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
825; GLM-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
826; GLM-NEXT: Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
827; GLM-NEXT: Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
828; GLM-NEXT: Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
829; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
830;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]831; BTVER2-LABEL: 'fsqrt'
832; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
833; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
834; BTVER2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
835; BTVER2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
836; BTVER2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
837; BTVER2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
838; BTVER2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
839; BTVER2-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
840; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
841;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]842 %F32 = call float @llvm.sqrt.f32(float undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]843 %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]844 %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]845 %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]846
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]847 %F64 = call double @llvm.sqrt.f64(double undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]848 %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]849 %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]850 %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]851
852 ret i32 undef
853}
854
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]855define i32 @fabs(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]856; SSE1-LABEL: 'fabs'
857; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
858; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
859; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
860; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
861; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
862; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
863; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
864; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
865; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
866;
867; SSE2-LABEL: 'fabs'
868; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
869; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
870; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
871; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
872; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
873; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
874; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
875; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
876; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
877;
878; SSE42-LABEL: 'fabs'
879; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
880; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
881; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
882; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
883; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
884; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
885; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
886; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
887; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]888;
889; AVX-LABEL: 'fabs'
890; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
891; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
892; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
893; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
894; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
895; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
896; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
897; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
898; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
899;
900; AVX512-LABEL: 'fabs'
901; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
902; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
903; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
904; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
905; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
906; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
907; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
908; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
909; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
910;
911; SLM-LABEL: 'fabs'
912; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
913; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
914; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
915; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
916; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
917; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
918; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
919; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
920; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
921;
922; GLM-LABEL: 'fabs'
923; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
924; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
925; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
926; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
927; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
928; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
929; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
930; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
931; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
932;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]933; BTVER2-LABEL: 'fabs'
934; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
935; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
936; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
937; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
938; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
939; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
940; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
941; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
942; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
943;
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]944 %F32 = call float @llvm.fabs.f32(float undef)
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]945 %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]946 %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]947 %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
948
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]949 %F64 = call double @llvm.fabs.f64(double undef)
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]950 %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]951 %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]952 %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
953
954 ret i32 undef
955}
956
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]957define i32 @fcopysign(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]958; SSE1-LABEL: 'fcopysign'
959; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
960; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
961; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
962; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
963; SSE1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
964; SSE1-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
965; SSE1-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
966; SSE1-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
967; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
968;
969; SSE2-LABEL: 'fcopysign'
970; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
971; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
972; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
973; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
974; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
975; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
976; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
977; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
978; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
979;
980; SSE42-LABEL: 'fcopysign'
981; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
982; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
983; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
984; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
985; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
986; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
987; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
988; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
989; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]990;
991; AVX-LABEL: 'fcopysign'
992; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
993; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
994; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
995; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
996; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
997; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
998; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
999; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1000; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1001;
1002; AVX512-LABEL: 'fcopysign'
1003; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1004; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
1005; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
1006; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
1007; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1008; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
1009; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
1010; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1011; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1012;
1013; SLM-LABEL: 'fcopysign'
1014; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1015; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
1016; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
1017; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
1018; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1019; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
1020; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
1021; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1022; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1023;
1024; GLM-LABEL: 'fcopysign'
1025; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1026; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
1027; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
1028; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
1029; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1030; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
1031; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
1032; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1033; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1034;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]1035; BTVER2-LABEL: 'fcopysign'
1036; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1037; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
1038; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
1039; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
1040; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1041; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
1042; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
1043; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1044; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1045;
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1046 %F32 = call float @llvm.copysign.f32(float undef, float undef)
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1047 %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1048 %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1049 %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
1050
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1051 %F64 = call double @llvm.copysign.f64(double undef, double undef)
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1052 %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1053 %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1054 %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1055
1056 ret i32 undef
1057}
1058
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1059define i32 @fma(i32 %arg) {
Simon Pilgrim71d61562019-01-04 16:37:01 +0000[diff] [blame]1060; SSE1-LABEL: 'fma'
1061; SSE1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1062; SSE1-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1063; SSE1-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1064; SSE1-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1065; SSE1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1066; SSE1-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1067; SSE1-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1068; SSE1-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1069; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1070;
1071; SSE2-LABEL: 'fma'
1072; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1073; SSE2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1074; SSE2-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1075; SSE2-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1076; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1077; SSE2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1078; SSE2-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1079; SSE2-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1080; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1081;
1082; SSE42-LABEL: 'fma'
1083; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1084; SSE42-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1085; SSE42-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1086; SSE42-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1087; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1088; SSE42-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1089; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1090; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1091; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]1092;
1093; AVX-LABEL: 'fma'
Simon Pilgrim3ede11b2018-04-13 19:12:32 +0000[diff] [blame]1094; AVX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1095; AVX-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1096; AVX-NEXT: Cost Model: Found an estimated cost of 87 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1097; AVX-NEXT: Cost Model: Found an estimated cost of 174 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1098; AVX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1099; AVX-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1100; AVX-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1101; AVX-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
Simon Pilgrim5334a2c2018-04-06 12:36:27 +0000[diff] [blame]1102; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1103;
1104; AVX512-LABEL: 'fma'
1105; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1106; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1107; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1108; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1109; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1110; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1111; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1112; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1113; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1114;
1115; SLM-LABEL: 'fma'
1116; SLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1117; SLM-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1118; SLM-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1119; SLM-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1120; SLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1121; SLM-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1122; SLM-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1123; SLM-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1124; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1125;
1126; GLM-LABEL: 'fma'
1127; GLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1128; GLM-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1129; GLM-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1130; GLM-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1131; GLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1132; GLM-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1133; GLM-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1134; GLM-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1135; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1136;
Simon Pilgrimcd9ccf82018-04-20 13:50:33 +0000[diff] [blame]1137; BTVER2-LABEL: 'fma'
1138; BTVER2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1139; BTVER2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1140; BTVER2-NEXT: Cost Model: Found an estimated cost of 87 for instruction: %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1141; BTVER2-NEXT: Cost Model: Found an estimated cost of 174 for instruction: %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1142; BTVER2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1143; BTVER2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1144; BTVER2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1145; BTVER2-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1146; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1147;
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1148 %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1149 %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1150 %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1151 %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1152
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1153 %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1154 %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1155 %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1156 %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1157
1158 ret i32 undef
1159}
1160
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1161declare float @llvm.sqrt.f32(float)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1162declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
1163declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
1164declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
1165
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1166declare double @llvm.sqrt.f64(double)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1167declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
1168declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)
1169declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
1170
Simon Pilgrim8b021c32016-10-01 16:30:13 +0000[diff] [blame]1171declare float @llvm.fabs.f32(float)
1172declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
1173declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
1174declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
1175
1176declare double @llvm.fabs.f64(double)
1177declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
1178declare <4 x double> @llvm.fabs.v4f64(<4 x double>)
1179declare <8 x double> @llvm.fabs.v8f64(<8 x double>)
1180
Simon Pilgrime0ec5c12016-10-01 16:41:52 +0000[diff] [blame]1181declare float @llvm.copysign.f32(float, float)
1182declare <4 x float> @llvm.copysign.v4f32(<4 x float>, <4 x float>)
1183declare <8 x float> @llvm.copysign.v8f32(<8 x float>, <8 x float>)
1184declare <16 x float> @llvm.copysign.v16f32(<16 x float>, <16 x float>)
1185
1186declare double @llvm.copysign.f64(double, double)
1187declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>)
1188declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>)
1189declare <8 x double> @llvm.copysign.v8f64(<8 x double>, <8 x double>)
1190
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1191declare float @llvm.fma.f32(float, float, float)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1192declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
1193declare <8 x float> @llvm.fma.v8f32(<8 x float>, <8 x float>, <8 x float>)
1194declare <16 x float> @llvm.fma.v16f32(<16 x float>, <16 x float>, <16 x float>)
1195
Simon Pilgrim91780592016-09-18 21:01:20 +0000[diff] [blame]1196declare double @llvm.fma.f64(double, double, double)
Simon Pilgrim07d7a212016-08-21 19:06:25 +0000[diff] [blame]1197declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
1198declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
1199declare <8 x double> @llvm.fma.v8f64(<8 x double>, <8 x double>, <8 x double>)