279 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 52; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output | FileCheck -check-prefixes=ALL,BASE %s3; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx803 -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output | FileCheck -check-prefixes=ALL,GFX8 %s4; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output | FileCheck -check-prefixes=ALL,GFX9 %s5; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx1030 -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output | FileCheck -check-prefixes=ALL,GFX10 %s6 7; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output | FileCheck -check-prefixes=ALL-SIZE,BASE-SIZE %s8; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx803 -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output | FileCheck -check-prefixes=ALL-SIZE,GFX8-SIZE %s9; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output | FileCheck -check-prefixes=ALL-SIZE,GFX9-SIZE %s10; RUN: opt < %s -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx1030 -passes="print<cost-model>" -cost-kind=code-size 2>&1 -disable-output | FileCheck -check-prefixes=ALL-SIZE,GFX10-SIZE %s11 12define void @copysign_f16() {13; BASE-LABEL: 'copysign_f16'14; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)15; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)16; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)17; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)18; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)19; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)20; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)21; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)22; BASE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void23;24; GFX8-LABEL: 'copysign_f16'25; GFX8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)26; GFX8-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)27; GFX8-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)28; GFX8-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)29; GFX8-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)30; GFX8-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)31; GFX8-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)32; GFX8-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)33; GFX8-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void34;35; GFX9-LABEL: 'copysign_f16'36; GFX9-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)37; GFX9-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)38; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)39; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)40; GFX9-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)41; GFX9-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)42; GFX9-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)43; GFX9-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)44; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void45;46; GFX10-LABEL: 'copysign_f16'47; GFX10-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)48; GFX10-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)49; GFX10-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)50; GFX10-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)51; GFX10-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)52; GFX10-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)53; GFX10-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)54; GFX10-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)55; GFX10-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void56;57; BASE-SIZE-LABEL: 'copysign_f16'58; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)59; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)60; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)61; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)62; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)63; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)64; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)65; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)66; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void67;68; GFX8-SIZE-LABEL: 'copysign_f16'69; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)70; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)71; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)72; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)73; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)74; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)75; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)76; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)77; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void78;79; GFX9-SIZE-LABEL: 'copysign_f16'80; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)81; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)82; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)83; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)84; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)85; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)86; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)87; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)88; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void89;90; GFX10-SIZE-LABEL: 'copysign_f16'91; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f16 = call half @llvm.copysign.f16(half undef, half undef)92; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)93; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)94; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)95; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)96; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)97; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)98; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)99; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void100;101 %f16 = call half @llvm.copysign.f16(half undef, half undef)102 %v2f16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef)103 %v3f16 = call <3 x half> @llvm.copysign.v3f16(<3 x half> undef, <3 x half> undef)104 %v4f16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef)105 %v5f16 = call <5 x half> @llvm.copysign.v5f16(<5 x half> undef, <5 x half> undef)106 %v8f16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef)107 %v9f16 = call <9 x half> @llvm.copysign.v9f16(<9 x half> undef, <9 x half> undef)108 %v16f16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef)109 ret void110}111 112define void @copysign_f32() {113; ALL-LABEL: 'copysign_f32'114; ALL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f32 = call float @llvm.copysign.f32(float undef, float undef)115; ALL-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f32 = call <2 x float> @llvm.copysign.v2f32(<2 x float> undef, <2 x float> undef)116; ALL-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f32 = call <3 x float> @llvm.copysign.v3f32(<3 x float> undef, <3 x float> undef)117; ALL-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)118; ALL-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v5f32 = call <5 x float> @llvm.copysign.v5f32(<5 x float> undef, <5 x float> undef)119; ALL-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)120; ALL-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %v9f32 = call <9 x float> @llvm.copysign.v9f32(<9 x float> undef, <9 x float> undef)121; ALL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)122; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void123;124; ALL-SIZE-LABEL: 'copysign_f32'125; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f32 = call float @llvm.copysign.f32(float undef, float undef)126; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2f32 = call <2 x float> @llvm.copysign.v2f32(<2 x float> undef, <2 x float> undef)127; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f32 = call <3 x float> @llvm.copysign.v3f32(<3 x float> undef, <3 x float> undef)128; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4f32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)129; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v5f32 = call <5 x float> @llvm.copysign.v5f32(<5 x float> undef, <5 x float> undef)130; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8f32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)131; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %v9f32 = call <9 x float> @llvm.copysign.v9f32(<9 x float> undef, <9 x float> undef)132; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)133; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void134;135 %f32 = call float @llvm.copysign.f32(float undef, float undef)136 %v2f32 = call <2 x float> @llvm.copysign.v2f32(<2 x float> undef, <2 x float> undef)137 %v3f32 = call <3 x float> @llvm.copysign.v3f32(<3 x float> undef, <3 x float> undef)138 %v4f32 = call <4 x float> @llvm.copysign.v4f32(<4 x float> undef, <4 x float> undef)139 %v5f32 = call <5 x float> @llvm.copysign.v5f32(<5 x float> undef, <5 x float> undef)140 %v8f32 = call <8 x float> @llvm.copysign.v8f32(<8 x float> undef, <8 x float> undef)141 %v9f32 = call <9 x float> @llvm.copysign.v9f32(<9 x float> undef, <9 x float> undef)142 %v16f32 = call <16 x float> @llvm.copysign.v16f32(<16 x float> undef, <16 x float> undef)143 ret void144}145 146define void @copysign_bf16() {147; BASE-LABEL: 'copysign_bf16'148; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)149; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)150; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)151; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)152; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)153; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)154; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)155; BASE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)156; BASE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void157;158; GFX8-LABEL: 'copysign_bf16'159; GFX8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)160; GFX8-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)161; GFX8-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)162; GFX8-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)163; GFX8-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)164; GFX8-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)165; GFX8-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)166; GFX8-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)167; GFX8-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void168;169; GFX9-LABEL: 'copysign_bf16'170; GFX9-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)171; GFX9-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)172; GFX9-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)173; GFX9-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)174; GFX9-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)175; GFX9-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)176; GFX9-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)177; GFX9-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)178; GFX9-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void179;180; GFX10-LABEL: 'copysign_bf16'181; GFX10-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)182; GFX10-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)183; GFX10-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)184; GFX10-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)185; GFX10-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)186; GFX10-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)187; GFX10-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)188; GFX10-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)189; GFX10-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void190;191; BASE-SIZE-LABEL: 'copysign_bf16'192; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)193; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)194; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)195; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)196; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)197; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)198; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)199; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)200; BASE-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void201;202; GFX8-SIZE-LABEL: 'copysign_bf16'203; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)204; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)205; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)206; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)207; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)208; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)209; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)210; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)211; GFX8-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void212;213; GFX9-SIZE-LABEL: 'copysign_bf16'214; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)215; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)216; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)217; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)218; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)219; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)220; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)221; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)222; GFX9-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void223;224; GFX10-SIZE-LABEL: 'copysign_bf16'225; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)226; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)227; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)228; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)229; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)230; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)231; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)232; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)233; GFX10-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void234;235 %bf16 = call bfloat @llvm.copysign.bf16(bfloat undef, bfloat undef)236 %v2bf16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef)237 %v3bf16 = call <3 x bfloat> @llvm.copysign.v3bf16(<3 x bfloat> undef, <3 x bfloat> undef)238 %v4bf16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef)239 %v5bf16 = call <5 x bfloat> @llvm.copysign.v5bf16(<5 x bfloat> undef, <5 x bfloat> undef)240 %v8bf16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef)241 %v9bf16 = call <9 x bfloat> @llvm.copysign.v9bf16(<9 x bfloat> undef, <9 x bfloat> undef)242 %v16bf16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef)243 ret void244}245 246define void @copysign_f64() {247; ALL-LABEL: 'copysign_f64'248; ALL-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f64 = call double @llvm.copysign.f64(double undef, double undef)249; ALL-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)250; ALL-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f64 = call <3 x double> @llvm.copysign.v3f64(<3 x double> undef, <3 x double> undef)251; ALL-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)252; ALL-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5f64 = call <5 x double> @llvm.copysign.v5f64(<5 x double> undef, <5 x double> undef)253; ALL-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)254; ALL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9f64 = call <9 x double> @llvm.copysign.v9f64(<9 x double> undef, <9 x double> undef)255; ALL-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f64 = call <16 x double> @llvm.copysign.v16f64(<16 x double> undef, <16 x double> undef)256; ALL-NEXT: Cost Model: Found an estimated cost of 10 for instruction: ret void257;258; ALL-SIZE-LABEL: 'copysign_f64'259; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %f64 = call double @llvm.copysign.f64(double undef, double undef)260; ALL-SIZE-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)261; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v3f64 = call <3 x double> @llvm.copysign.v3f64(<3 x double> undef, <3 x double> undef)262; ALL-SIZE-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)263; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v5f64 = call <5 x double> @llvm.copysign.v5f64(<5 x double> undef, <5 x double> undef)264; ALL-SIZE-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)265; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v9f64 = call <9 x double> @llvm.copysign.v9f64(<9 x double> undef, <9 x double> undef)266; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %v16f64 = call <16 x double> @llvm.copysign.v16f64(<16 x double> undef, <16 x double> undef)267; ALL-SIZE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret void268;269 %f64 = call double @llvm.copysign.f64(double undef, double undef)270 %v2f64 = call <2 x double> @llvm.copysign.v2f64(<2 x double> undef, <2 x double> undef)271 %v3f64 = call <3 x double> @llvm.copysign.v3f64(<3 x double> undef, <3 x double> undef)272 %v4f64 = call <4 x double> @llvm.copysign.v4f64(<4 x double> undef, <4 x double> undef)273 %v5f64 = call <5 x double> @llvm.copysign.v5f64(<5 x double> undef, <5 x double> undef)274 %v8f64 = call <8 x double> @llvm.copysign.v8f64(<8 x double> undef, <8 x double> undef)275 %v9f64 = call <9 x double> @llvm.copysign.v9f64(<9 x double> undef, <9 x double> undef)276 %v16f64 = call <16 x double> @llvm.copysign.v16f64(<16 x double> undef, <16 x double> undef)277 ret void278}279