1 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
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
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
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
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
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
14 target 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"
15 target triple = "x86_64-apple-macosx10.8.0"
17 define i32 @fadd(i32 %arg) {
19 ; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fadd float undef, undef
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
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
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
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
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
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
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
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
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
117 %F32 = fadd float undef, undef
118 %V4F32 = fadd <4 x float> undef, undef
119 %V8F32 = fadd <8 x float> undef, undef
120 %V16F32 = fadd <16 x float> undef, undef
122 %F64 = fadd double undef, undef
123 %V2F64 = fadd <2 x double> undef, undef
124 %V4F64 = fadd <4 x double> undef, undef
125 %V8F64 = fadd <8 x double> undef, undef
130 define i32 @fsub(i32 %arg) {
132 ; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fsub float undef, undef
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
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
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
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
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
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
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
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
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
230 %F32 = fsub float undef, undef
231 %V4F32 = fsub <4 x float> undef, undef
232 %V8F32 = fsub <8 x float> undef, undef
233 %V16F32 = fsub <16 x float> undef, undef
235 %F64 = fsub double undef, undef
236 %V2F64 = fsub <2 x double> undef, undef
237 %V4F64 = fsub <4 x double> undef, undef
238 %V8F64 = fsub <8 x double> undef, undef
243 define 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
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
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
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
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
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
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
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
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
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
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
356 define i32 @fneg(i32 %arg) {
358 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
359 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
360 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F32 = fneg <8 x float> undef
361 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16F32 = fneg <16 x float> undef
362 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fneg double undef
363 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
364 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fneg <4 x double> undef
365 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fneg <8 x double> undef
366 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
369 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
370 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
371 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F32 = fneg <8 x float> undef
372 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16F32 = fneg <16 x float> undef
373 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
374 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
375 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fneg <4 x double> undef
376 ; SSE2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fneg <8 x double> undef
377 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
379 ; SSE42-LABEL: 'fneg'
380 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
381 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
382 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F32 = fneg <8 x float> undef
383 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16F32 = fneg <16 x float> undef
384 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
385 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
386 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fneg <4 x double> undef
387 ; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fneg <8 x double> undef
388 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
391 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
392 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
393 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fneg <8 x float> undef
394 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fneg <16 x float> undef
395 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
396 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
397 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fneg <4 x double> undef
398 ; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fneg <8 x double> undef
399 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
401 ; AVX512-LABEL: 'fneg'
402 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
403 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
404 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fneg <8 x float> undef
405 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fneg <16 x float> undef
406 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
407 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
408 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fneg <4 x double> undef
409 ; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fneg <8 x double> undef
410 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
413 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
414 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
415 ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F32 = fneg <8 x float> undef
416 ; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16F32 = fneg <16 x float> undef
417 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
418 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
419 ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fneg <4 x double> undef
420 ; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fneg <8 x double> undef
421 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
424 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
425 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
426 ; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F32 = fneg <8 x float> undef
427 ; GLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16F32 = fneg <16 x float> undef
428 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
429 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
430 ; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fneg <4 x double> undef
431 ; GLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fneg <8 x double> undef
432 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
434 ; BTVER2-LABEL: 'fneg'
435 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F32 = fneg float undef
436 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F32 = fneg <4 x float> undef
437 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fneg <8 x float> undef
438 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fneg <16 x float> undef
439 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %F64 = fneg double undef
440 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fneg <2 x double> undef
441 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fneg <4 x double> undef
442 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fneg <8 x double> undef
443 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
445 %F32 = fneg float undef
446 %V4F32 = fneg <4 x float> undef
447 %V8F32 = fneg <8 x float> undef
448 %V16F32 = fneg <16 x float> undef
450 %F64 = fneg double undef
451 %V2F64 = fneg <2 x double> undef
452 %V4F64 = fneg <4 x double> undef
453 %V8F64 = fneg <8 x double> undef
458 define i32 @fmul(i32 %arg) {
460 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fmul float undef, undef
461 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fmul <4 x float> undef, undef
462 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fmul <8 x float> undef, undef
463 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fmul <16 x float> undef, undef
464 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fmul double undef, undef
465 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fmul <2 x double> undef, undef
466 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fmul <4 x double> undef, undef
467 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fmul <8 x double> undef, undef
468 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
471 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = fmul float undef, undef
472 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fmul <4 x float> undef, undef
473 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fmul <8 x float> undef, undef
474 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fmul <16 x float> undef, undef
475 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fmul double undef, undef
476 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = fmul <2 x double> undef, undef
477 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = fmul <4 x double> undef, undef
478 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = fmul <8 x double> undef, undef
479 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
481 ; SSE42-LABEL: 'fmul'
482 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
483 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
484 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
485 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
486 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
487 ; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
488 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
489 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
490 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
493 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
494 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
495 ; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
496 ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
497 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
498 ; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
499 ; AVX1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
500 ; AVX1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
501 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
504 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
505 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
506 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
507 ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fmul <16 x float> undef, undef
508 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
509 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
510 ; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
511 ; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fmul <8 x double> undef, undef
512 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
514 ; AVX512-LABEL: 'fmul'
515 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
516 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
517 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = fmul <8 x float> undef, undef
518 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = fmul <16 x float> undef, undef
519 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
520 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
521 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = fmul <4 x double> undef, undef
522 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = fmul <8 x double> undef, undef
523 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
526 ; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
527 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = fmul <4 x float> undef, undef
528 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = fmul <8 x float> undef, undef
529 ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = fmul <16 x float> undef, undef
530 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fmul double undef, undef
531 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fmul <2 x double> undef, undef
532 ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fmul <4 x double> undef, undef
533 ; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fmul <8 x double> undef, undef
534 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
537 ; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
538 ; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
539 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
540 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
541 ; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
542 ; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
543 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
544 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
545 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
547 ; BTVER2-LABEL: 'fmul'
548 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = fmul float undef, undef
549 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = fmul <4 x float> undef, undef
550 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = fmul <8 x float> undef, undef
551 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = fmul <16 x float> undef, undef
552 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = fmul double undef, undef
553 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = fmul <2 x double> undef, undef
554 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = fmul <4 x double> undef, undef
555 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = fmul <8 x double> undef, undef
556 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
558 %F32 = fmul float undef, undef
559 %V4F32 = fmul <4 x float> undef, undef
560 %V8F32 = fmul <8 x float> undef, undef
561 %V16F32 = fmul <16 x float> undef, undef
563 %F64 = fmul double undef, undef
564 %V2F64 = fmul <2 x double> undef, undef
565 %V4F64 = fmul <4 x double> undef, undef
566 %V8F64 = fmul <8 x double> undef, undef
571 define i32 @fdiv(i32 %arg) {
573 ; SSE1-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %F32 = fdiv float undef, undef
574 ; SSE1-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V4F32 = fdiv <4 x float> undef, undef
575 ; SSE1-NEXT: Cost Model: Found an estimated cost of 68 for instruction: %V8F32 = fdiv <8 x float> undef, undef
576 ; SSE1-NEXT: Cost Model: Found an estimated cost of 136 for instruction: %V16F32 = fdiv <16 x float> undef, undef
577 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = fdiv double undef, undef
578 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = fdiv <2 x double> undef, undef
579 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = fdiv <4 x double> undef, undef
580 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = fdiv <8 x double> undef, undef
581 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
584 ; SSE2-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %F32 = fdiv float undef, undef
585 ; SSE2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4F32 = fdiv <4 x float> undef, undef
586 ; SSE2-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8F32 = fdiv <8 x float> undef, undef
587 ; SSE2-NEXT: Cost Model: Found an estimated cost of 156 for instruction: %V16F32 = fdiv <16 x float> undef, undef
588 ; SSE2-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %F64 = fdiv double undef, undef
589 ; SSE2-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %V2F64 = fdiv <2 x double> undef, undef
590 ; SSE2-NEXT: Cost Model: Found an estimated cost of 138 for instruction: %V4F64 = fdiv <4 x double> undef, undef
591 ; SSE2-NEXT: Cost Model: Found an estimated cost of 276 for instruction: %V8F64 = fdiv <8 x double> undef, undef
592 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
594 ; SSE42-LABEL: 'fdiv'
595 ; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
596 ; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
597 ; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
598 ; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
599 ; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
600 ; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
601 ; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
602 ; SSE42-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
603 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
606 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
607 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
608 ; AVX1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
609 ; AVX1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
610 ; AVX1-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
611 ; AVX1-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
612 ; AVX1-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
613 ; AVX1-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
614 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
617 ; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = fdiv float undef, undef
618 ; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = fdiv <4 x float> undef, undef
619 ; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = fdiv <8 x float> undef, undef
620 ; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = fdiv <16 x float> undef, undef
621 ; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = fdiv double undef, undef
622 ; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = fdiv <2 x double> undef, undef
623 ; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = fdiv <4 x double> undef, undef
624 ; AVX2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = fdiv <8 x double> undef, undef
625 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
627 ; AVX512-LABEL: 'fdiv'
628 ; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = fdiv float undef, undef
629 ; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = fdiv <4 x float> undef, undef
630 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = fdiv <8 x float> undef, undef
631 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = fdiv <16 x float> undef, undef
632 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = fdiv double undef, undef
633 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = fdiv <2 x double> undef, undef
634 ; AVX512-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = fdiv <4 x double> undef, undef
635 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = fdiv <8 x double> undef, undef
636 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
639 ; SLM-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %F32 = fdiv float undef, undef
640 ; SLM-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4F32 = fdiv <4 x float> undef, undef
641 ; SLM-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8F32 = fdiv <8 x float> undef, undef
642 ; SLM-NEXT: Cost Model: Found an estimated cost of 156 for instruction: %V16F32 = fdiv <16 x float> undef, undef
643 ; SLM-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %F64 = fdiv double undef, undef
644 ; SLM-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %V2F64 = fdiv <2 x double> undef, undef
645 ; SLM-NEXT: Cost Model: Found an estimated cost of 138 for instruction: %V4F64 = fdiv <4 x double> undef, undef
646 ; SLM-NEXT: Cost Model: Found an estimated cost of 276 for instruction: %V8F64 = fdiv <8 x double> undef, undef
647 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
650 ; GLM-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %F32 = fdiv float undef, undef
651 ; GLM-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %V4F32 = fdiv <4 x float> undef, undef
652 ; GLM-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %V8F32 = fdiv <8 x float> undef, undef
653 ; GLM-NEXT: Cost Model: Found an estimated cost of 140 for instruction: %V16F32 = fdiv <16 x float> undef, undef
654 ; GLM-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %F64 = fdiv double undef, undef
655 ; GLM-NEXT: Cost Model: Found an estimated cost of 65 for instruction: %V2F64 = fdiv <2 x double> undef, undef
656 ; GLM-NEXT: Cost Model: Found an estimated cost of 130 for instruction: %V4F64 = fdiv <4 x double> undef, undef
657 ; GLM-NEXT: Cost Model: Found an estimated cost of 260 for instruction: %V8F64 = fdiv <8 x double> undef, undef
658 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
660 ; BTVER2-LABEL: 'fdiv'
661 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = fdiv float undef, undef
662 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = fdiv <4 x float> undef, undef
663 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = fdiv <8 x float> undef, undef
664 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = fdiv <16 x float> undef, undef
665 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %F64 = fdiv double undef, undef
666 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V2F64 = fdiv <2 x double> undef, undef
667 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4F64 = fdiv <4 x double> undef, undef
668 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8F64 = fdiv <8 x double> undef, undef
669 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
671 %F32 = fdiv float undef, undef
672 %V4F32 = fdiv <4 x float> undef, undef
673 %V8F32 = fdiv <8 x float> undef, undef
674 %V16F32 = fdiv <16 x float> undef, undef
676 %F64 = fdiv double undef, undef
677 %V2F64 = fdiv <2 x double> undef, undef
678 %V4F64 = fdiv <4 x double> undef, undef
679 %V8F64 = fdiv <8 x double> undef, undef
684 define i32 @frem(i32 %arg) {
686 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
687 ; SSE1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
688 ; SSE1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
689 ; SSE1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
690 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
691 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = frem <2 x double> undef, undef
692 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = frem <4 x double> undef, undef
693 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = frem <8 x double> undef, undef
694 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
697 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
698 ; SSE2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
699 ; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
700 ; SSE2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
701 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
702 ; SSE2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
703 ; SSE2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
704 ; SSE2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
705 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
707 ; SSE42-LABEL: 'frem'
708 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
709 ; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
710 ; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
711 ; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
712 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
713 ; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
714 ; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
715 ; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
716 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
719 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
720 ; AVX-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
721 ; AVX-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
722 ; AVX-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = frem <16 x float> undef, undef
723 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
724 ; AVX-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
725 ; AVX-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
726 ; AVX-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = frem <8 x double> undef, undef
727 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
729 ; AVX512-LABEL: 'frem'
730 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
731 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
732 ; AVX512-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
733 ; AVX512-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16F32 = frem <16 x float> undef, undef
734 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
735 ; AVX512-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
736 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
737 ; AVX512-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F64 = frem <8 x double> undef, undef
738 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
741 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
742 ; SLM-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
743 ; SLM-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
744 ; SLM-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
745 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
746 ; SLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
747 ; SLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
748 ; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
749 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
752 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
753 ; GLM-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
754 ; GLM-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = frem <8 x float> undef, undef
755 ; GLM-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = frem <16 x float> undef, undef
756 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
757 ; GLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
758 ; GLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4F64 = frem <4 x double> undef, undef
759 ; GLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8F64 = frem <8 x double> undef, undef
760 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
762 ; BTVER2-LABEL: 'frem'
763 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = frem float undef, undef
764 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = frem <4 x float> undef, undef
765 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8F32 = frem <8 x float> undef, undef
766 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16F32 = frem <16 x float> undef, undef
767 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = frem double undef, undef
768 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2F64 = frem <2 x double> undef, undef
769 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F64 = frem <4 x double> undef, undef
770 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F64 = frem <8 x double> undef, undef
771 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
773 %F32 = frem float undef, undef
774 %V4F32 = frem <4 x float> undef, undef
775 %V8F32 = frem <8 x float> undef, undef
776 %V16F32 = frem <16 x float> undef, undef
778 %F64 = frem double undef, undef
779 %V2F64 = frem <2 x double> undef, undef
780 %V4F64 = frem <4 x double> undef, undef
781 %V8F64 = frem <8 x double> undef, undef
786 define i32 @fsqrt(i32 %arg) {
787 ; SSE1-LABEL: 'fsqrt'
788 ; SSE1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
789 ; SSE1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
790 ; SSE1-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
791 ; SSE1-NEXT: Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
792 ; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
793 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
794 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
795 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
796 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
798 ; SSE2-LABEL: 'fsqrt'
799 ; SSE2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
800 ; SSE2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
801 ; SSE2-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
802 ; SSE2-NEXT: Cost Model: Found an estimated cost of 224 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
803 ; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
804 ; SSE2-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
805 ; SSE2-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
806 ; SSE2-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
807 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
809 ; SSE42-LABEL: 'fsqrt'
810 ; SSE42-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
811 ; SSE42-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
812 ; SSE42-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
813 ; SSE42-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
814 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
815 ; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
816 ; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
817 ; SSE42-NEXT: Cost Model: Found an estimated cost of 128 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
818 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
820 ; AVX1-LABEL: 'fsqrt'
821 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
822 ; AVX1-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
823 ; AVX1-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
824 ; AVX1-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
825 ; AVX1-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
826 ; AVX1-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
827 ; AVX1-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
828 ; AVX1-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
829 ; AVX1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
831 ; AVX2-LABEL: 'fsqrt'
832 ; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
833 ; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
834 ; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
835 ; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
836 ; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
837 ; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
838 ; AVX2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
839 ; AVX2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
840 ; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
842 ; AVX512-LABEL: 'fsqrt'
843 ; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
844 ; AVX512-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
845 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
846 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
847 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
848 ; AVX512-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
849 ; AVX512-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
850 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
851 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
854 ; SLM-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
855 ; SLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
856 ; SLM-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
857 ; SLM-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
858 ; SLM-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
859 ; SLM-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
860 ; SLM-NEXT: Cost Model: Found an estimated cost of 140 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
861 ; SLM-NEXT: Cost Model: Found an estimated cost of 280 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
862 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
865 ; GLM-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
866 ; GLM-NEXT: Cost Model: Found an estimated cost of 37 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
867 ; GLM-NEXT: Cost Model: Found an estimated cost of 74 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
868 ; GLM-NEXT: Cost Model: Found an estimated cost of 148 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
869 ; GLM-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
870 ; GLM-NEXT: Cost Model: Found an estimated cost of 67 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
871 ; GLM-NEXT: Cost Model: Found an estimated cost of 134 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
872 ; GLM-NEXT: Cost Model: Found an estimated cost of 268 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
873 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
875 ; BTVER2-LABEL: 'fsqrt'
876 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %F32 = call float @llvm.sqrt.f32(float undef)
877 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
878 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
879 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
880 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %F64 = call double @llvm.sqrt.f64(double undef)
881 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
882 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
883 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
884 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
886 %F32 = call float @llvm.sqrt.f32(float undef)
887 %V4F32 = call <4 x float> @llvm.sqrt.v4f32(<4 x float> undef)
888 %V8F32 = call <8 x float> @llvm.sqrt.v8f32(<8 x float> undef)
889 %V16F32 = call <16 x float> @llvm.sqrt.v16f32(<16 x float> undef)
891 %F64 = call double @llvm.sqrt.f64(double undef)
892 %V2F64 = call <2 x double> @llvm.sqrt.v2f64(<2 x double> undef)
893 %V4F64 = call <4 x double> @llvm.sqrt.v4f64(<4 x double> undef)
894 %V8F64 = call <8 x double> @llvm.sqrt.v8f64(<8 x double> undef)
899 define i32 @fabs(i32 %arg) {
901 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
902 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
903 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
904 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
905 ; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
906 ; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
907 ; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
908 ; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
909 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
912 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
913 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
914 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
915 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
916 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
917 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
918 ; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
919 ; SSE2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
920 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
922 ; SSE42-LABEL: 'fabs'
923 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
924 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
925 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
926 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
927 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
928 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
929 ; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
930 ; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
931 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
934 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
935 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
936 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
937 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
938 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
939 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
940 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
941 ; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
942 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
944 ; AVX512-LABEL: 'fabs'
945 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
946 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
947 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
948 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
949 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
950 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
951 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
952 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
953 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
956 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
957 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
958 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
959 ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
960 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
961 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
962 ; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
963 ; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
964 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
967 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
968 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
969 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
970 ; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
971 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
972 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
973 ; GLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
974 ; GLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
975 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
977 ; BTVER2-LABEL: 'fabs'
978 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.fabs.f32(float undef)
979 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
980 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
981 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
982 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.fabs.f64(double undef)
983 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
984 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
985 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
986 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
988 %F32 = call float @llvm.fabs.f32(float undef)
989 %V4F32 = call <4 x float> @llvm.fabs.v4f32(<4 x float> undef)
990 %V8F32 = call <8 x float> @llvm.fabs.v8f32(<8 x float> undef)
991 %V16F32 = call <16 x float> @llvm.fabs.v16f32(<16 x float> undef)
993 %F64 = call double @llvm.fabs.f64(double undef)
994 %V2F64 = call <2 x double> @llvm.fabs.v2f64(<2 x double> undef)
995 %V4F64 = call <4 x double> @llvm.fabs.v4f64(<4 x double> undef)
996 %V8F64 = call <8 x double> @llvm.fabs.v8f64(<8 x double> undef)
1001 define i32 @fcopysign(i32 %arg) {
1002 ; SSE1-LABEL: 'fcopysign'
1003 ; SSE1-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1004 ; 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)
1005 ; 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)
1006 ; 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)
1007 ; SSE1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1008 ; 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)
1009 ; 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)
1010 ; 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)
1011 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1013 ; SSE2-LABEL: 'fcopysign'
1014 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1015 ; 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)
1016 ; 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)
1017 ; 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)
1018 ; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1019 ; 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)
1020 ; 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)
1021 ; 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)
1022 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1024 ; SSE42-LABEL: 'fcopysign'
1025 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1026 ; 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)
1027 ; 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)
1028 ; 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)
1029 ; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1030 ; 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)
1031 ; 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)
1032 ; 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)
1033 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1035 ; AVX-LABEL: 'fcopysign'
1036 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1037 ; 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)
1038 ; 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)
1039 ; 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)
1040 ; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1041 ; 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)
1042 ; 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)
1043 ; 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)
1044 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1046 ; AVX512-LABEL: 'fcopysign'
1047 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1048 ; 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)
1049 ; 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)
1050 ; 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)
1051 ; AVX512-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1052 ; 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)
1053 ; 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)
1054 ; 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)
1055 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1057 ; SLM-LABEL: 'fcopysign'
1058 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1059 ; 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)
1060 ; 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)
1061 ; 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)
1062 ; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1063 ; 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)
1064 ; 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)
1065 ; 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)
1066 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1068 ; GLM-LABEL: 'fcopysign'
1069 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1070 ; 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)
1071 ; 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)
1072 ; 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)
1073 ; GLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1074 ; 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)
1075 ; 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)
1076 ; 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)
1077 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1079 ; BTVER2-LABEL: 'fcopysign'
1080 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
1081 ; 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)
1082 ; 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)
1083 ; 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)
1084 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
1085 ; 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)
1086 ; 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)
1087 ; 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)
1088 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1090 %F32 = call float @llvm.copysign.f32(float undef, float undef)
1091 %V4F32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)
1092 %V8F32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)
1093 %V16F32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)
1095 %F64 = call double @llvm.copysign.f64(double undef, double undef)
1096 %V2F64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)
1097 %V4F64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)
1098 %V8F64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)
1103 define i32 @fma(i32 %arg) {
1105 ; SSE1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1106 ; 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)
1107 ; 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)
1108 ; 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)
1109 ; SSE1-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1110 ; 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)
1111 ; 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)
1112 ; 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)
1113 ; SSE1-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1116 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1117 ; 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)
1118 ; 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)
1119 ; 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)
1120 ; SSE2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1121 ; 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)
1122 ; 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)
1123 ; 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)
1124 ; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1126 ; SSE42-LABEL: 'fma'
1127 ; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1128 ; 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)
1129 ; 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)
1130 ; 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)
1131 ; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1132 ; 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)
1133 ; 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)
1134 ; 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)
1135 ; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1138 ; AVX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1139 ; 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)
1140 ; 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)
1141 ; 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)
1142 ; AVX-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1143 ; 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)
1144 ; 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)
1145 ; 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)
1146 ; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1148 ; AVX512-LABEL: 'fma'
1149 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1150 ; 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)
1151 ; 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)
1152 ; 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)
1153 ; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1154 ; 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)
1155 ; 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)
1156 ; 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)
1157 ; AVX512-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1160 ; SLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1161 ; 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)
1162 ; 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)
1163 ; 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)
1164 ; SLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1165 ; 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)
1166 ; 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)
1167 ; 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)
1168 ; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1171 ; GLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1172 ; 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)
1173 ; 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)
1174 ; 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)
1175 ; GLM-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1176 ; 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)
1177 ; 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)
1178 ; 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)
1179 ; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1181 ; BTVER2-LABEL: 'fma'
1182 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1183 ; 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)
1184 ; 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)
1185 ; 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)
1186 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1187 ; 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)
1188 ; 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)
1189 ; 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)
1190 ; BTVER2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
1192 %F32 = call float @llvm.fma.f32(float undef, float undef, float undef)
1193 %V4F32 = call <4 x float> @llvm.fma.v4f32(<4 x float> undef, <4 x float> undef, <4 x float> undef)
1194 %V8F32 = call <8 x float> @llvm.fma.v8f32(<8 x float> undef, <8 x float> undef, <8 x float> undef)
1195 %V16F32 = call <16 x float> @llvm.fma.v16f32(<16 x float> undef, <16 x float> undef, <16 x float> undef)
1197 %F64 = call double @llvm.fma.f64(double undef, double undef, double undef)
1198 %V2F64 = call <2 x double> @llvm.fma.v2f64(<2 x double> undef, <2 x double> undef, <2 x double> undef)
1199 %V4F64 = call <4 x double> @llvm.fma.v4f64(<4 x double> undef, <4 x double> undef, <4 x double> undef)
1200 %V8F64 = call <8 x double> @llvm.fma.v8f64(<8 x double> undef, <8 x double> undef, <8 x double> undef)
1205 declare float @llvm.sqrt.f32(float)
1206 declare <4 x float> @llvm.sqrt.v4f32(<4 x float>)
1207 declare <8 x float> @llvm.sqrt.v8f32(<8 x float>)
1208 declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
1210 declare double @llvm.sqrt.f64(double)
1211 declare <2 x double> @llvm.sqrt.v2f64(<2 x double>)
1212 declare <4 x double> @llvm.sqrt.v4f64(<4 x double>)
1213 declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
1215 declare float @llvm.fabs.f32(float)
1216 declare <4 x float> @llvm.fabs.v4f32(<4 x float>)
1217 declare <8 x float> @llvm.fabs.v8f32(<8 x float>)
1218 declare <16 x float> @llvm.fabs.v16f32(<16 x float>)
1220 declare double @llvm.fabs.f64(double)
1221 declare <2 x double> @llvm.fabs.v2f64(<2 x double>)
1222 declare <4 x double> @llvm.fabs.v4f64(<4 x double>)
1223 declare <8 x double> @llvm.fabs.v8f64(<8 x double>)
1225 declare float @llvm.copysign.f32(float, float)
1226 declare <4 x float> @llvm.copysign.v4f32(<4 x float>, <4 x float>)
1227 declare <8 x float> @llvm.copysign.v8f32(<8 x float>, <8 x float>)
1228 declare <16 x float> @llvm.copysign.v16f32(<16 x float>, <16 x float>)
1230 declare double @llvm.copysign.f64(double, double)
1231 declare <2 x double> @llvm.copysign.v2f64(<2 x double>, <2 x double>)
1232 declare <4 x double> @llvm.copysign.v4f64(<4 x double>, <4 x double>)
1233 declare <8 x double> @llvm.copysign.v8f64(<8 x double>, <8 x double>)
1235 declare float @llvm.fma.f32(float, float, float)
1236 declare <4 x float> @llvm.fma.v4f32(<4 x float>, <4 x float>, <4 x float>)
1237 declare <8 x float> @llvm.fma.v8f32(<8 x float>, <8 x float>, <8 x float>)
1238 declare <16 x float> @llvm.fma.v16f32(<16 x float>, <16 x float>, <16 x float>)
1240 declare double @llvm.fma.f64(double, double, double)
1241 declare <2 x double> @llvm.fma.v2f64(<2 x double>, <2 x double>, <2 x double>)
1242 declare <4 x double> @llvm.fma.v4f64(<4 x double>, <4 x double>, <4 x double>)
1243 declare <8 x double> @llvm.fma.v8f64(<8 x double>, <8 x double>, <8 x double>)