1 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
2 ; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx90a -mattr=+half-rate-64-ops < %s | FileCheck -check-prefixes=ALL,GFX9,GFX90A-FASTF64 %s
3 ; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 -mattr=+half-rate-64-ops < %s | FileCheck -check-prefixes=ALL,GFX9,FASTF64 %s
4 ; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mattr=-half-rate-64-ops < %s | FileCheck -check-prefixes=ALL,SLOWF64 %s
5 ; RUN: opt -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx90a -mattr=+half-rate-64-ops < %s | FileCheck -check-prefixes=SIZE,GFX9-SIZE,GFX90A-SIZE %s
6 ; RUN: opt -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 -mattr=+half-rate-64-ops < %s | FileCheck -check-prefixes=SIZE,GFX9-SIZE %s
7 ; RUN: opt -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mattr=-half-rate-64-ops < %s | FileCheck -check-prefixes=SIZE,SLOW-SIZE %s
9 define void @minnum_f16() {
10 ; GFX9-LABEL: 'minnum_f16'
11 ; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.minnum.f16(half undef, half undef)
12 ; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
13 ; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.minnum.v3f16(<3 x half> undef, <3 x half> undef)
14 ; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
15 ; GFX9-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %v8f16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
16 ; GFX9-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v16f16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
17 ; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
19 ; SLOWF64-LABEL: 'minnum_f16'
20 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.minnum.f16(half undef, half undef)
21 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
22 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.minnum.v3f16(<3 x half> undef, <3 x half> undef)
23 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
24 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
25 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
26 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
28 ; GFX9-SIZE-LABEL: 'minnum_f16'
29 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.minnum.f16(half undef, half undef)
30 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
31 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.minnum.v3f16(<3 x half> undef, <3 x half> undef)
32 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
33 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %v8f16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
34 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v16f16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
35 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
37 ; SLOW-SIZE-LABEL: 'minnum_f16'
38 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.minnum.f16(half undef, half undef)
39 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
40 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.minnum.v3f16(<3 x half> undef, <3 x half> undef)
41 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
42 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
43 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
44 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
46 ; SLOW-LABEL: 'minnum_f16'
47 ; SLOW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.minnum.f16(half undef, half undef)
48 ; SLOW-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
49 ; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.minnum.v3f16(<3 x half> undef, <3 x half> undef)
50 ; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
51 ; SLOW-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
52 ; SLOW-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
53 ; SLOW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
54 %f16 = call half @llvm.minnum.f16(half undef, half undef)
55 %v2f16 = call <2 x half> @llvm.minnum.v2f16(<2 x half> undef, <2 x half> undef)
56 %v3f16 = call <3x half> @llvm.minnum.v3f16(<3 x half> undef, <3 x half> undef)
57 %v4f16 = call <4 x half> @llvm.minnum.v4f16(<4 x half> undef, <4 x half> undef)
58 %v8f16 = call <8 x half> @llvm.minnum.v8f16(<8 x half> undef, <8 x half> undef)
59 %v16f16 = call <16 x half> @llvm.minnum.v16f16(<16 x half> undef, <16 x half> undef)
63 define void @minnum_bf16() {
64 ; GFX9-LABEL: 'minnum_bf16'
65 ; GFX9-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.minnum.bf16(bfloat undef, bfloat undef)
66 ; GFX9-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2bf16 = call <2 x bfloat> @llvm.minnum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)
67 ; GFX9-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3bf16 = call <3 x bfloat> @llvm.minnum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)
68 ; GFX9-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4bf16 = call <4 x bfloat> @llvm.minnum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)
69 ; GFX9-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8bf16 = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)
70 ; GFX9-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16bf16 = call <16 x bfloat> @llvm.minnum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)
71 ; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
73 ; SLOWF64-LABEL: 'minnum_bf16'
74 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %bf16 = call bfloat @llvm.minnum.bf16(bfloat undef, bfloat undef)
75 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2bf16 = call <2 x bfloat> @llvm.minnum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)
76 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3bf16 = call <3 x bfloat> @llvm.minnum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)
77 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4bf16 = call <4 x bfloat> @llvm.minnum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)
78 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8bf16 = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)
79 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16bf16 = call <16 x bfloat> @llvm.minnum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)
80 ; SLOWF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
82 ; GFX9-SIZE-LABEL: 'minnum_bf16'
83 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.minnum.bf16(bfloat undef, bfloat undef)
84 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2bf16 = call <2 x bfloat> @llvm.minnum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)
85 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3bf16 = call <3 x bfloat> @llvm.minnum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)
86 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4bf16 = call <4 x bfloat> @llvm.minnum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)
87 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8bf16 = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)
88 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16bf16 = call <16 x bfloat> @llvm.minnum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)
89 ; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
91 ; SLOW-SIZE-LABEL: 'minnum_bf16'
92 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %bf16 = call bfloat @llvm.minnum.bf16(bfloat undef, bfloat undef)
93 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2bf16 = call <2 x bfloat> @llvm.minnum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)
94 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3bf16 = call <3 x bfloat> @llvm.minnum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)
95 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4bf16 = call <4 x bfloat> @llvm.minnum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)
96 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8bf16 = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)
97 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16bf16 = call <16 x bfloat> @llvm.minnum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)
98 ; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
100 ; SLOW-LABEL: 'minnum_bf16'
101 ; SLOW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %bf16 = call bfloat @llvm.minnum.bf16(bfloat undef, bfloat undef)
102 ; SLOW-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2bf16 = call <2 x bfloat> @llvm.minnum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)
103 ; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3bf16 = call <3 x bfloat> @llvm.minnum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)
104 ; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4bf16 = call <4 x bfloat> @llvm.minnum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)
105 ; SLOW-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8bf16 = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)
106 ; SLOW-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16bf16 = call <16 x bfloat> @llvm.minnum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)
107 ; SLOW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
108 %bf16 = call bfloat @llvm.minnum.bf16(bfloat undef, bfloat undef)
109 %v2bf16 = call <2 x bfloat> @llvm.minnum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)
110 %v3bf16 = call <3x bfloat> @llvm.minnum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)
111 %v4bf16 = call <4 x bfloat> @llvm.minnum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)
112 %v8bf16 = call <8 x bfloat> @llvm.minnum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)
113 %v16bf16 = call <16 x bfloat> @llvm.minnum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)
117 define void @minnum_f32() {
118 ; ALL-LABEL: 'minnum_f32'
119 ; ALL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f32 = call float @llvm.minnum.f32(float undef, float undef)
120 ; ALL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
121 ; ALL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f32 = call <3 x float> @llvm.minnum.v3f32(<3 x float> undef, <3 x float> undef)
122 ; ALL-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
123 ; ALL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
124 ; ALL-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
125 ; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
127 ; SIZE-LABEL: 'minnum_f32'
128 ; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f32 = call float @llvm.minnum.f32(float undef, float undef)
129 ; SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
130 ; SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f32 = call <3 x float> @llvm.minnum.v3f32(<3 x float> undef, <3 x float> undef)
131 ; SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
132 ; SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
133 ; SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
134 ; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
136 %f32 = call float @llvm.minnum.f32(float undef, float undef)
137 %v2f32 = call <2 x float> @llvm.minnum.v2f32(<2 x float> undef, <2 x float> undef)
138 %v3f32 = call <3x float> @llvm.minnum.v3f32(<3 x float> undef, <3 x float> undef)
139 %v4f32 = call <4 x float> @llvm.minnum.v4f32(<4 x float> undef, <4 x float> undef)
140 %v8f32 = call <8 x float> @llvm.minnum.v8f32(<8 x float> undef, <8 x float> undef)
141 %v16f32 = call <16 x float> @llvm.minnum.v16f32(<16 x float> undef, <16 x float> undef)
145 define void @minnum_f64() {
146 ; ALL-LABEL: 'minnum_f64'
147 ; ALL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
148 ; ALL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
149 ; ALL-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f64 = call <3 x double> @llvm.minnum.v3f64(<3 x double> undef, <3 x double> undef)
150 ; ALL-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
151 ; ALL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
152 ; ALL-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
153 ; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void
155 ; SIZE-LABEL: 'minnum_f64'
156 ; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f64 = call double @llvm.minnum.f64(double undef, double undef)
157 ; SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
158 ; SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f64 = call <3 x double> @llvm.minnum.v3f64(<3 x double> undef, <3 x double> undef)
159 ; SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
160 ; SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
161 ; SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
162 ; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void
164 %f64 = call double @llvm.minnum.f64(double undef, double undef)
165 %v2f64 = call <2 x double> @llvm.minnum.v2f64(<2 x double> undef, <2 x double> undef)
166 %v3f64 = call <3x double> @llvm.minnum.v3f64(<3 x double> undef, <3 x double> undef)
167 %v4f64 = call <4 x double> @llvm.minnum.v4f64(<4 x double> undef, <4 x double> undef)
168 %v8f64 = call <8 x double> @llvm.minnum.v8f64(<8 x double> undef, <8 x double> undef)
169 %v16f64 = call <16 x double> @llvm.minnum.v16f64(<16 x double> undef, <16 x double> undef)
172 ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
174 ; GFX90A-FASTF64: {{.*}}
175 ; GFX90A-SIZE: {{.*}}