232 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 52; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx950 -mattr=+half-rate-64-ops < %s | FileCheck -check-prefix=GFX950-FASTF64 %s3; 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 %s4; 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 %s5; 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 %s6; RUN: opt -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx950 -mattr=+half-rate-64-ops < %s | FileCheck -check-prefixes=SIZE,GFX950-SIZE %s7; 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 %s8; 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 %s9; 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 %s10 11define void @maximum_f16() {12; GFX950-FASTF64-LABEL: 'maximum_f16'13; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)14; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)15; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)16; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)17; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)18; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)19; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void20;21; GFX9-LABEL: 'maximum_f16'22; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)23; GFX9-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)24; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)25; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)26; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)27; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)28; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void29;30; SLOWF64-LABEL: 'maximum_f16'31; SLOWF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)32; SLOWF64-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)33; SLOWF64-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)34; SLOWF64-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)35; SLOWF64-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)36; SLOWF64-NEXT: Cost Model: Found an estimated cost of 176 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)37; SLOWF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void38;39; GFX950-SIZE-LABEL: 'maximum_f16'40; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)41; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)42; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)43; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)44; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)45; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)46; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void47;48; GFX9-SIZE-LABEL: 'maximum_f16'49; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)50; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)51; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)52; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)53; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)54; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)55; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void56;57; SLOW-SIZE-LABEL: 'maximum_f16'58; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)59; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)60; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)61; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)62; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)63; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)64; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void65;66; SLOW-LABEL: 'maximum_f16'67; SLOW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %f16 = call half @llvm.maximum.f16(half undef, half undef)68; SLOW-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)69; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3f16 = call <3 x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)70; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)71; SLOW-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)72; SLOW-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)73; SLOW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void74 %f16 = call half @llvm.maximum.f16(half undef, half undef)75 %v2f16 = call <2 x half> @llvm.maximum.v2f16(<2 x half> undef, <2 x half> undef)76 %v3f16 = call <3x half> @llvm.maximum.v3f16(<3 x half> undef, <3 x half> undef)77 %v4f16 = call <4 x half> @llvm.maximum.v4f16(<4 x half> undef, <4 x half> undef)78 %v8f16 = call <8 x half> @llvm.maximum.v8f16(<8 x half> undef, <8 x half> undef)79 %v16f16 = call <16 x half> @llvm.maximum.v16f16(<16 x half> undef, <16 x half> undef)80 ret void81}82 83define void @maximum_bf16() {84; GFX950-FASTF64-LABEL: 'maximum_bf16'85; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)86; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)87; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)88; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)89; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)90; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)91; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void92;93; GFX9-LABEL: 'maximum_bf16'94; GFX9-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)95; GFX9-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)96; GFX9-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)97; GFX9-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)98; GFX9-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)99; GFX9-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)100; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void101;102; SLOWF64-LABEL: 'maximum_bf16'103; SLOWF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)104; SLOWF64-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)105; SLOWF64-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)106; SLOWF64-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)107; SLOWF64-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)108; SLOWF64-NEXT: Cost Model: Found an estimated cost of 176 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)109; SLOWF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void110;111; GFX950-SIZE-LABEL: 'maximum_bf16'112; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)113; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)114; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)115; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)116; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)117; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)118; GFX950-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void119;120; GFX9-SIZE-LABEL: 'maximum_bf16'121; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)122; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)123; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)124; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)125; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)126; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)127; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void128;129; SLOW-SIZE-LABEL: 'maximum_bf16'130; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)131; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)132; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)133; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)134; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)135; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)136; SLOW-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void137;138; SLOW-LABEL: 'maximum_bf16'139; SLOW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)140; SLOW-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)141; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v3bf16 = call <3 x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)142; SLOW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)143; SLOW-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)144; SLOW-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)145; SLOW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void146 %bf16 = call bfloat @llvm.maximum.bf16(bfloat undef, bfloat undef)147 %v2bf16 = call <2 x bfloat> @llvm.maximum.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)148 %v3bf16 = call <3x bfloat> @llvm.maximum.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)149 %v4bf16 = call <4 x bfloat> @llvm.maximum.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)150 %v8bf16 = call <8 x bfloat> @llvm.maximum.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)151 %v16bf16 = call <16 x bfloat> @llvm.maximum.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)152 ret void153}154 155define void @maximum_f32() {156; GFX950-FASTF64-LABEL: 'maximum_f32'157; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f32 = call float @llvm.maximum.f32(float undef, float undef)158; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f32 = call <2 x float> @llvm.maximum.v2f32(<2 x float> undef, <2 x float> undef)159; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f32 = call <3 x float> @llvm.maximum.v3f32(<3 x float> undef, <3 x float> undef)160; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f32 = call <4 x float> @llvm.maximum.v4f32(<4 x float> undef, <4 x float> undef)161; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f32 = call <8 x float> @llvm.maximum.v8f32(<8 x float> undef, <8 x float> undef)162; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f32 = call <16 x float> @llvm.maximum.v16f32(<16 x float> undef, <16 x float> undef)163; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void164;165; ALL-LABEL: 'maximum_f32'166; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %f32 = call float @llvm.maximum.f32(float undef, float undef)167; ALL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v2f32 = call <2 x float> @llvm.maximum.v2f32(<2 x float> undef, <2 x float> undef)168; ALL-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %v3f32 = call <3 x float> @llvm.maximum.v3f32(<3 x float> undef, <3 x float> undef)169; ALL-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v4f32 = call <4 x float> @llvm.maximum.v4f32(<4 x float> undef, <4 x float> undef)170; ALL-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %v8f32 = call <8 x float> @llvm.maximum.v8f32(<8 x float> undef, <8 x float> undef)171; ALL-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %v16f32 = call <16 x float> @llvm.maximum.v16f32(<16 x float> undef, <16 x float> undef)172; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void173;174; SIZE-LABEL: 'maximum_f32'175; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f32 = call float @llvm.maximum.f32(float undef, float undef)176; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f32 = call <2 x float> @llvm.maximum.v2f32(<2 x float> undef, <2 x float> undef)177; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f32 = call <3 x float> @llvm.maximum.v3f32(<3 x float> undef, <3 x float> undef)178; SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f32 = call <4 x float> @llvm.maximum.v4f32(<4 x float> undef, <4 x float> undef)179; SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f32 = call <8 x float> @llvm.maximum.v8f32(<8 x float> undef, <8 x float> undef)180; SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f32 = call <16 x float> @llvm.maximum.v16f32(<16 x float> undef, <16 x float> undef)181; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void182;183 %f32 = call float @llvm.maximum.f32(float undef, float undef)184 %v2f32 = call <2 x float> @llvm.maximum.v2f32(<2 x float> undef, <2 x float> undef)185 %v3f32 = call <3x float> @llvm.maximum.v3f32(<3 x float> undef, <3 x float> undef)186 %v4f32 = call <4 x float> @llvm.maximum.v4f32(<4 x float> undef, <4 x float> undef)187 %v8f32 = call <8 x float> @llvm.maximum.v8f32(<8 x float> undef, <8 x float> undef)188 %v16f32 = call <16 x float> @llvm.maximum.v16f32(<16 x float> undef, <16 x float> undef)189 ret void190}191 192define void @maximum_f64() {193; GFX950-FASTF64-LABEL: 'maximum_f64'194; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maximum.f64(double undef, double undef)195; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v2f64 = call <2 x double> @llvm.maximum.v2f64(<2 x double> undef, <2 x double> undef)196; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %v3f64 = call <3 x double> @llvm.maximum.v3f64(<3 x double> undef, <3 x double> undef)197; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v4f64 = call <4 x double> @llvm.maximum.v4f64(<4 x double> undef, <4 x double> undef)198; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %v8f64 = call <8 x double> @llvm.maximum.v8f64(<8 x double> undef, <8 x double> undef)199; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %v16f64 = call <16 x double> @llvm.maximum.v16f64(<16 x double> undef, <16 x double> undef)200; GFX950-FASTF64-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void201;202; ALL-LABEL: 'maximum_f64'203; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %f64 = call double @llvm.maximum.f64(double undef, double undef)204; ALL-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v2f64 = call <2 x double> @llvm.maximum.v2f64(<2 x double> undef, <2 x double> undef)205; ALL-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %v3f64 = call <3 x double> @llvm.maximum.v3f64(<3 x double> undef, <3 x double> undef)206; ALL-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %v4f64 = call <4 x double> @llvm.maximum.v4f64(<4 x double> undef, <4 x double> undef)207; ALL-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %v8f64 = call <8 x double> @llvm.maximum.v8f64(<8 x double> undef, <8 x double> undef)208; ALL-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %v16f64 = call <16 x double> @llvm.maximum.v16f64(<16 x double> undef, <16 x double> undef)209; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void210;211; SIZE-LABEL: 'maximum_f64'212; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f64 = call double @llvm.maximum.f64(double undef, double undef)213; SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f64 = call <2 x double> @llvm.maximum.v2f64(<2 x double> undef, <2 x double> undef)214; SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f64 = call <3 x double> @llvm.maximum.v3f64(<3 x double> undef, <3 x double> undef)215; SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f64 = call <4 x double> @llvm.maximum.v4f64(<4 x double> undef, <4 x double> undef)216; SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f64 = call <8 x double> @llvm.maximum.v8f64(<8 x double> undef, <8 x double> undef)217; SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f64 = call <16 x double> @llvm.maximum.v16f64(<16 x double> undef, <16 x double> undef)218; SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void219;220 %f64 = call double @llvm.maximum.f64(double undef, double undef)221 %v2f64 = call <2 x double> @llvm.maximum.v2f64(<2 x double> undef, <2 x double> undef)222 %v3f64 = call <3x double> @llvm.maximum.v3f64(<3 x double> undef, <3 x double> undef)223 %v4f64 = call <4 x double> @llvm.maximum.v4f64(<4 x double> undef, <4 x double> undef)224 %v8f64 = call <8 x double> @llvm.maximum.v8f64(<8 x double> undef, <8 x double> undef)225 %v16f64 = call <16 x double> @llvm.maximum.v16f64(<16 x double> undef, <16 x double> undef)226 ret void227}228;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:229; FASTF64: {{.*}}230; GFX90A-FASTF64: {{.*}}231; GFX90A-SIZE: {{.*}}232