|
1 | 1 | ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
|
2 |
| -; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfh,+zvfbfmin < %s | FileCheck %s --check-prefixes=CHECK,ZVFH |
3 |
| -; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfhmin,+zvfbfmin < %s | FileCheck %s --check-prefixes=CHECK,ZVFHMIN |
| 2 | +; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfh,+zvfbfmin --type-based-intrinsic-cost=true < %s | FileCheck %s --check-prefixes=CHECK,ZVFH |
| 3 | +; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64 -mattr=+v,+f,+d,+zfh,+zvfhmin,+zvfbfmin --type-based-intrinsic-cost=true < %s | FileCheck %s --check-prefixes=CHECK,ZVFHMIN |
4 | 4 | ; Check that we don't crash querying costs when vectors are not enabled.
|
5 | 5 | ; RUN: opt -passes="print<cost-model>" 2>&1 -disable-output -mtriple=riscv64
|
6 | 6 |
|
@@ -661,6 +661,36 @@ define void @fneg() {
|
661 | 661 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %NXV2F64 = fneg <vscale x 2 x double> undef
|
662 | 662 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %NXV4F64 = fneg <vscale x 4 x double> undef
|
663 | 663 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %NXV8F64 = fneg <vscale x 8 x double> undef
|
| 664 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1BF16_VP = call <1 x bfloat> @llvm.vp.fneg.v1bf16(<1 x bfloat> undef, <1 x i1> undef, i32 undef) |
| 665 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2BF16_VP = call <2 x bfloat> @llvm.vp.fneg.v2bf16(<2 x bfloat> undef, <2 x i1> undef, i32 undef) |
| 666 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4BF16_VP = call <4 x bfloat> @llvm.vp.fneg.v4bf16(<4 x bfloat> undef, <4 x i1> undef, i32 undef) |
| 667 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8BF16_VP = call <8 x bfloat> @llvm.vp.fneg.v8bf16(<8 x bfloat> undef, <8 x i1> undef, i32 undef) |
| 668 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16BF16_VP = call <16 x bfloat> @llvm.vp.fneg.v16bf16(<16 x bfloat> undef, <16 x i1> undef, i32 undef) |
| 669 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F32_VP = call <1 x float> @llvm.vp.fneg.v1f32(<1 x float> undef, <1 x i1> undef, i32 undef) |
| 670 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F32_VP = call <2 x float> @llvm.vp.fneg.v2f32(<2 x float> undef, <2 x i1> undef, i32 undef) |
| 671 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F32_VP = call <4 x float> @llvm.vp.fneg.v4f32(<4 x float> undef, <4 x i1> undef, i32 undef) |
| 672 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F32_VP = call <8 x float> @llvm.vp.fneg.v8f32(<8 x float> undef, <8 x i1> undef, i32 undef) |
| 673 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16F32_VP = call <16 x float> @llvm.vp.fneg.v16f32(<16 x float> undef, <16 x i1> undef, i32 undef) |
| 674 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V1F64_VP = call <1 x double> @llvm.vp.fneg.v1f64(<1 x double> undef, <1 x i1> undef, i32 undef) |
| 675 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2F64_VP = call <2 x double> @llvm.vp.fneg.v2f64(<2 x double> undef, <2 x i1> undef, i32 undef) |
| 676 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4F64_VP = call <4 x double> @llvm.vp.fneg.v4f64(<4 x double> undef, <4 x i1> undef, i32 undef) |
| 677 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8F64_VP = call <8 x double> @llvm.vp.fneg.v8f64(<8 x double> undef, <8 x i1> undef, i32 undef) |
| 678 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16F64_VP = call <16 x double> @llvm.vp.fneg.v16f64(<16 x double> undef, <16 x i1> undef, i32 undef) |
| 679 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %NXV1BF16_VP = call <vscale x 1 x bfloat> @llvm.vp.fneg.nxv1bf16(<vscale x 1 x bfloat> undef, <vscale x 1 x i1> undef, i32 undef) |
| 680 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %NXV2BF16_VP = call <vscale x 2 x bfloat> @llvm.vp.fneg.nxv2bf16(<vscale x 2 x bfloat> undef, <vscale x 2 x i1> undef, i32 undef) |
| 681 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %NXV4BF16_VP = call <vscale x 4 x bfloat> @llvm.vp.fneg.nxv4bf16(<vscale x 4 x bfloat> undef, <vscale x 4 x i1> undef, i32 undef) |
| 682 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %NXV8BF16_VP = call <vscale x 8 x bfloat> @llvm.vp.fneg.nxv8bf16(<vscale x 8 x bfloat> undef, <vscale x 8 x i1> undef, i32 undef) |
| 683 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %NXV16BF16_VP = call <vscale x 16 x bfloat> @llvm.vp.fneg.nxv16bf16(<vscale x 16 x bfloat> undef, <vscale x 16 x i1> undef, i32 undef) |
| 684 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %NXV1BF32_VP = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> undef, <vscale x 1 x i1> undef, i32 undef) |
| 685 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %NXV2BF32_VP = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> undef, i32 undef) |
| 686 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %NXV4BF32_VP = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> undef, i32 undef) |
| 687 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %NXV8BF32_VP = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> undef, <vscale x 8 x i1> undef, i32 undef) |
| 688 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %NXV16BF32_VP = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> undef, <vscale x 16 x i1> undef, i32 undef) |
| 689 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %NXV1BF64_VP = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> undef, <vscale x 1 x i1> undef, i32 undef) |
| 690 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %NXV2BF64_VP = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> undef, i32 undef) |
| 691 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %NXV4BF64_VP = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> undef, <vscale x 4 x i1> undef, i32 undef) |
| 692 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %NXV8BF64_VP = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> undef, <vscale x 8 x i1> undef, i32 undef) |
| 693 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %NXV16BF64_VP = call <vscale x 16 x double> @llvm.vp.fneg.nxv16f64(<vscale x 16 x double> undef, <vscale x 16 x i1> undef, i32 undef) |
664 | 694 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
|
665 | 695 | ;
|
666 | 696 | %BF16 = fneg half undef
|
@@ -701,6 +731,42 @@ define void @fneg() {
|
701 | 731 | %NXV4F64 = fneg <vscale x 4 x double> undef
|
702 | 732 | %NXV8F64 = fneg <vscale x 8 x double> undef
|
703 | 733 |
|
| 734 | + %V1BF16_VP = call <1 x bfloat> @llvm.vp.fneg.v1f16(<1 x bfloat> undef, <1 x i1> undef, i32 undef) |
| 735 | + %V2BF16_VP = call <2 x bfloat> @llvm.vp.fneg.v2f16(<2 x bfloat> undef, <2 x i1> undef, i32 undef) |
| 736 | + %V4BF16_VP = call <4 x bfloat> @llvm.vp.fneg.v4f16(<4 x bfloat> undef, <4 x i1> undef, i32 undef) |
| 737 | + %V8BF16_VP = call <8 x bfloat> @llvm.vp.fneg.v8f16(<8 x bfloat> undef, <8 x i1> undef, i32 undef) |
| 738 | + %V16BF16_VP = call <16 x bfloat> @llvm.vp.fneg.v16f16(<16 x bfloat> undef, <16 x i1> undef, i32 undef) |
| 739 | + |
| 740 | + %V1F32_VP = call <1 x float> @llvm.vp.fneg.v1f32(<1 x float> undef, <1 x i1> undef, i32 undef) |
| 741 | + %V2F32_VP = call <2 x float> @llvm.vp.fneg.v2f32(<2 x float> undef, <2 x i1> undef, i32 undef) |
| 742 | + %V4F32_VP = call <4 x float> @llvm.vp.fneg.v4f32(<4 x float> undef, <4 x i1> undef, i32 undef) |
| 743 | + %V8F32_VP = call <8 x float> @llvm.vp.fneg.v8f32(<8 x float> undef, <8 x i1> undef, i32 undef) |
| 744 | + %V16F32_VP = call <16 x float> @llvm.vp.fneg.v16f32(<16 x float> undef, <16 x i1> undef, i32 undef) |
| 745 | + |
| 746 | + %V1F64_VP = call <1 x double> @llvm.vp.fneg.v1f64(<1 x double> undef, <1 x i1> undef, i32 undef) |
| 747 | + %V2F64_VP = call <2 x double> @llvm.vp.fneg.v2f64(<2 x double> undef, <2 x i1> undef, i32 undef) |
| 748 | + %V4F64_VP = call <4 x double> @llvm.vp.fneg.v4f64(<4 x double> undef, <4 x i1> undef, i32 undef) |
| 749 | + %V8F64_VP = call <8 x double> @llvm.vp.fneg.v8f64(<8 x double> undef, <8 x i1> undef, i32 undef) |
| 750 | + %V16F64_VP = call <16 x double> @llvm.vp.fneg.v16f64(<16 x double> undef, <16 x i1> undef, i32 undef) |
| 751 | + |
| 752 | + %NXV1BF16_VP = call <vscale x 1 x bfloat> @llvm.vp.fneg.nxv1f16(<vscale x 1 x bfloat> undef, <vscale x 1 x i1> undef, i32 undef) |
| 753 | + %NXV2BF16_VP = call <vscale x 2 x bfloat> @llvm.vp.fneg.nxv2f16(<vscale x 2 x bfloat> undef, <vscale x 2 x i1> undef, i32 undef) |
| 754 | + %NXV4BF16_VP = call <vscale x 4 x bfloat> @llvm.vp.fneg.nxv4f16(<vscale x 4 x bfloat> undef, <vscale x 4 x i1> undef, i32 undef) |
| 755 | + %NXV8BF16_VP = call <vscale x 8 x bfloat> @llvm.vp.fneg.nxv8f16(<vscale x 8 x bfloat> undef, <vscale x 8 x i1> undef, i32 undef) |
| 756 | + %NXV16BF16_VP = call <vscale x 16 x bfloat> @llvm.vp.fneg.nxv16f16(<vscale x 16 x bfloat> undef, <vscale x 16 x i1> undef, i32 undef) |
| 757 | + |
| 758 | + %NXV1BF32_VP = call <vscale x 1 x float> @llvm.vp.fneg.nxv1f32(<vscale x 1 x float> undef, <vscale x 1 x i1> undef, i32 undef) |
| 759 | + %NXV2BF32_VP = call <vscale x 2 x float> @llvm.vp.fneg.nxv2f32(<vscale x 2 x float> undef, <vscale x 2 x i1> undef, i32 undef) |
| 760 | + %NXV4BF32_VP = call <vscale x 4 x float> @llvm.vp.fneg.nxv4f32(<vscale x 4 x float> undef, <vscale x 4 x i1> undef, i32 undef) |
| 761 | + %NXV8BF32_VP = call <vscale x 8 x float> @llvm.vp.fneg.nxv8f32(<vscale x 8 x float> undef, <vscale x 8 x i1> undef, i32 undef) |
| 762 | + %NXV16BF32_VP = call <vscale x 16 x float> @llvm.vp.fneg.nxv16f32(<vscale x 16 x float> undef, <vscale x 16 x i1> undef, i32 undef) |
| 763 | + |
| 764 | + %NXV1BF64_VP = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> undef, <vscale x 1 x i1> undef, i32 undef) |
| 765 | + %NXV2BF64_VP = call <vscale x 2 x double> @llvm.vp.fneg.nxv2f64(<vscale x 2 x double> undef, <vscale x 2 x i1> undef, i32 undef) |
| 766 | + %NXV4BF64_VP = call <vscale x 4 x double> @llvm.vp.fneg.nxv4f64(<vscale x 4 x double> undef, <vscale x 4 x i1> undef, i32 undef) |
| 767 | + %NXV8BF64_VP = call <vscale x 8 x double> @llvm.vp.fneg.nxv8f64(<vscale x 8 x double> undef, <vscale x 8 x i1> undef, i32 undef) |
| 768 | + %NXV16BF64_VP = call <vscale x 16 x double> @llvm.vp.fneg.nxv16f64(<vscale x 16 x double> undef, <vscale x 16 x i1> undef, i32 undef) |
| 769 | + |
704 | 770 | ret void
|
705 | 771 | }
|
706 | 772 |
|
@@ -746,10 +812,10 @@ define void @fcopysign() {
|
746 | 812 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call float @llvm.copysign.f32(float undef, float undef)
|
747 | 813 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call double @llvm.copysign.f64(double undef, double undef)
|
748 | 814 | ; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V1BF16 = call <1 x bfloat> @llvm.copysign.v1bf16(<1 x bfloat> undef, <1 x bfloat> undef)
|
749 |
| -; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2BF16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef) |
750 |
| -; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4BF16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef) |
751 |
| -; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8BF16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef) |
752 |
| -; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %V16BF16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef) |
| 815 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2BF16 = call <2 x bfloat> @llvm.copysign.v2bf16(<2 x bfloat> undef, <2 x bfloat> undef) |
| 816 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V4BF16 = call <4 x bfloat> @llvm.copysign.v4bf16(<4 x bfloat> undef, <4 x bfloat> undef) |
| 817 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V8BF16 = call <8 x bfloat> @llvm.copysign.v8bf16(<8 x bfloat> undef, <8 x bfloat> undef) |
| 818 | +; CHECK-NEXT: Cost Model: Found an estimated cost of 109 for instruction: %V16BF16 = call <16 x bfloat> @llvm.copysign.v16bf16(<16 x bfloat> undef, <16 x bfloat> undef) |
753 | 819 | ; CHECK-NEXT: Cost Model: Invalid cost for instruction: %NXV1BF16 = call <vscale x 1 x bfloat> @llvm.copysign.nxv1bf16(<vscale x 1 x bfloat> undef, <vscale x 1 x bfloat> undef)
|
754 | 820 | ; CHECK-NEXT: Cost Model: Invalid cost for instruction: %NXV2BF16 = call <vscale x 2 x bfloat> @llvm.copysign.nxv2bf16(<vscale x 2 x bfloat> undef, <vscale x 2 x bfloat> undef)
|
755 | 821 | ; CHECK-NEXT: Cost Model: Invalid cost for instruction: %NXV4BF16 = call <vscale x 4 x bfloat> @llvm.copysign.nxv4bf16(<vscale x 4 x bfloat> undef, <vscale x 4 x bfloat> undef)
|
@@ -836,11 +902,11 @@ define void @fcopysign_f16() {
|
836 | 902 | ; ZVFHMIN-LABEL: 'fcopysign_f16'
|
837 | 903 | ; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F16 = call half @llvm.copysign.f16(half undef, half undef)
|
838 | 904 | ; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V1F16 = call <1 x half> @llvm.copysign.v1f16(<1 x half> undef, <1 x half> undef)
|
839 |
| -; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef) |
840 |
| -; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef) |
841 |
| -; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef) |
842 |
| -; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %V16F16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef) |
843 |
| -; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 95 for instruction: %V32F16 = call <32 x half> @llvm.copysign.v32f16(<32 x half> undef, <32 x half> undef) |
| 905 | +; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2F16 = call <2 x half> @llvm.copysign.v2f16(<2 x half> undef, <2 x half> undef) |
| 906 | +; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V4F16 = call <4 x half> @llvm.copysign.v4f16(<4 x half> undef, <4 x half> undef) |
| 907 | +; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V8F16 = call <8 x half> @llvm.copysign.v8f16(<8 x half> undef, <8 x half> undef) |
| 908 | +; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 109 for instruction: %V16F16 = call <16 x half> @llvm.copysign.v16f16(<16 x half> undef, <16 x half> undef) |
| 909 | +; ZVFHMIN-NEXT: Cost Model: Found an estimated cost of 221 for instruction: %V32F16 = call <32 x half> @llvm.copysign.v32f16(<32 x half> undef, <32 x half> undef) |
844 | 910 | ; ZVFHMIN-NEXT: Cost Model: Invalid cost for instruction: %NXV1F16 = call <vscale x 1 x half> @llvm.copysign.nxv1f16(<vscale x 1 x half> undef, <vscale x 1 x half> undef)
|
845 | 911 | ; ZVFHMIN-NEXT: Cost Model: Invalid cost for instruction: %NXV2F16 = call <vscale x 2 x half> @llvm.copysign.nxv2f16(<vscale x 2 x half> undef, <vscale x 2 x half> undef)
|
846 | 912 | ; ZVFHMIN-NEXT: Cost Model: Invalid cost for instruction: %NXV4F16 = call <vscale x 4 x half> @llvm.copysign.nxv4f16(<vscale x 4 x half> undef, <vscale x 4 x half> undef)
|
|
0 commit comments