derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
228 lines
11 KiB
LLVM
228 lines
11 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
|
|
; RUN: -mcpu=pwr10 -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu \
|
|
; RUN: -mcpu=pwr10 -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=CHECK-BE
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
|
|
; RUN: -mcpu=future -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=CHECK-LE-WACC
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu \
|
|
; RUN: -mcpu=future -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=CHECK-BE-WACC
|
|
|
|
define void @testMultiply(ptr nocapture noundef readonly %a, ptr nocapture noundef readonly %b, ptr nocapture noundef writeonly %c) local_unnamed_addr #0 {
|
|
; CHECK-LABEL: testMultiply:
|
|
; CHECK: # %bb.0: # %entry
|
|
; CHECK-NEXT: mflr r0
|
|
; CHECK-NEXT: std r30, -16(r1)
|
|
; CHECK-NEXT: mr r30, r1
|
|
; CHECK-NEXT: std r0, 16(r1)
|
|
; CHECK-NEXT: clrldi r0, r1, 59
|
|
; CHECK-NEXT: subfic r0, r0, -128
|
|
; CHECK-NEXT: stdux r1, r1, r0
|
|
; CHECK-NEXT: stxv v30, -64(r30) # 16-byte Folded Spill
|
|
; CHECK-NEXT: stxv v31, -48(r30) # 16-byte Folded Spill
|
|
; CHECK-NEXT: lxv v31, 0(r3)
|
|
; CHECK-NEXT: lxv v30, 0(r4)
|
|
; CHECK-NEXT: addi r3, r1, 32
|
|
; CHECK-NEXT: vmr v2, v31
|
|
; CHECK-NEXT: vmr v3, v30
|
|
; CHECK-NEXT: std r29, -24(r30) # 8-byte Folded Spill
|
|
; CHECK-NEXT: mr r29, r5
|
|
; CHECK-NEXT: bl _Z15buildVectorPairPu13__vector_pairDv16_hS0_@notoc
|
|
; CHECK-NEXT: xxsetaccz acc1
|
|
; CHECK-NEXT: xvf32gerpp acc1, v31, v30
|
|
; CHECK-NEXT: lxv vs0, 48(r1)
|
|
; CHECK-NEXT: lxv vs1, 32(r1)
|
|
; CHECK-NEXT: xvf32gerpp acc1, vs1, vs0
|
|
; CHECK-NEXT: lxv v31, -48(r30) # 16-byte Folded Reload
|
|
; CHECK-NEXT: lxv v30, -64(r30) # 16-byte Folded Reload
|
|
; CHECK-NEXT: xxmfacc acc1
|
|
; CHECK-NEXT: xxlor vs0, vs4, vs4
|
|
; CHECK-NEXT: xxlor vs1, vs5, vs5
|
|
; CHECK-NEXT: xxlor vs2, vs6, vs6
|
|
; CHECK-NEXT: xxlor vs3, vs7, vs7
|
|
; CHECK-NEXT: stxv vs3, 0(r29)
|
|
; CHECK-NEXT: pstxv vs2, 8(r29), 0
|
|
; CHECK-NEXT: stxv vs1, 16(r29)
|
|
; CHECK-NEXT: pstxv vs0, 24(r29), 0
|
|
; CHECK-NEXT: ld r29, -24(r30) # 8-byte Folded Reload
|
|
; CHECK-NEXT: mr r1, r30
|
|
; CHECK-NEXT: ld r0, 16(r1)
|
|
; CHECK-NEXT: ld r30, -16(r1)
|
|
; CHECK-NEXT: mtlr r0
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; CHECK-BE-LABEL: testMultiply:
|
|
; CHECK-BE: # %bb.0: # %entry
|
|
; CHECK-BE-NEXT: mflr r0
|
|
; CHECK-BE-NEXT: std r30, -16(r1)
|
|
; CHECK-BE-NEXT: mr r30, r1
|
|
; CHECK-BE-NEXT: std r0, 16(r1)
|
|
; CHECK-BE-NEXT: clrldi r0, r1, 59
|
|
; CHECK-BE-NEXT: subfic r0, r0, -224
|
|
; CHECK-BE-NEXT: stdux r1, r1, r0
|
|
; CHECK-BE-NEXT: stxv v30, -64(r30) # 16-byte Folded Spill
|
|
; CHECK-BE-NEXT: stxv v31, -48(r30) # 16-byte Folded Spill
|
|
; CHECK-BE-NEXT: lxv v31, 0(r3)
|
|
; CHECK-BE-NEXT: lxv v30, 0(r4)
|
|
; CHECK-BE-NEXT: addi r3, r1, 128
|
|
; CHECK-BE-NEXT: vmr v2, v31
|
|
; CHECK-BE-NEXT: vmr v3, v30
|
|
; CHECK-BE-NEXT: std r29, -24(r30) # 8-byte Folded Spill
|
|
; CHECK-BE-NEXT: mr r29, r5
|
|
; CHECK-BE-NEXT: bl _Z15buildVectorPairPu13__vector_pairDv16_hS0_
|
|
; CHECK-BE-NEXT: nop
|
|
; CHECK-BE-NEXT: xxsetaccz acc1
|
|
; CHECK-BE-NEXT: xvf32gerpp acc1, v31, v30
|
|
; CHECK-BE-NEXT: lxv vs0, 128(r1)
|
|
; CHECK-BE-NEXT: lxv vs1, 144(r1)
|
|
; CHECK-BE-NEXT: xvf32gerpp acc1, vs0, vs1
|
|
; CHECK-BE-NEXT: lxv v31, -48(r30) # 16-byte Folded Reload
|
|
; CHECK-BE-NEXT: lxv v30, -64(r30) # 16-byte Folded Reload
|
|
; CHECK-BE-NEXT: xxmfacc acc1
|
|
; CHECK-BE-NEXT: xxlor vs3, vs4, vs4
|
|
; CHECK-BE-NEXT: xxlor vs2, vs5, vs5
|
|
; CHECK-BE-NEXT: xxlor vs1, vs6, vs6
|
|
; CHECK-BE-NEXT: xxlor vs0, vs7, vs7
|
|
; CHECK-BE-NEXT: stxv vs0, 0(r29)
|
|
; CHECK-BE-NEXT: pstxv vs1, 8(r29), 0
|
|
; CHECK-BE-NEXT: stxv vs2, 16(r29)
|
|
; CHECK-BE-NEXT: pstxv vs3, 24(r29), 0
|
|
; CHECK-BE-NEXT: ld r29, -24(r30) # 8-byte Folded Reload
|
|
; CHECK-BE-NEXT: mr r1, r30
|
|
; CHECK-BE-NEXT: ld r0, 16(r1)
|
|
; CHECK-BE-NEXT: ld r30, -16(r1)
|
|
; CHECK-BE-NEXT: mtlr r0
|
|
; CHECK-BE-NEXT: blr
|
|
;
|
|
; CHECK-LE-WACC-LABEL: testMultiply:
|
|
; CHECK-LE-WACC: # %bb.0: # %entry
|
|
; CHECK-LE-WACC-NEXT: mflr r0
|
|
; CHECK-LE-WACC-NEXT: std r30, -16(r1)
|
|
; CHECK-LE-WACC-NEXT: std r0, 16(r1)
|
|
; CHECK-LE-WACC-NEXT: clrldi r0, r1, 59
|
|
; CHECK-LE-WACC-NEXT: subfic r0, r0, -128
|
|
; CHECK-LE-WACC-NEXT: mr r30, r1
|
|
; CHECK-LE-WACC-NEXT: stdux r1, r1, r0
|
|
; CHECK-LE-WACC-NEXT: stxv v30, -64(r30) # 16-byte Folded Spill
|
|
; CHECK-LE-WACC-NEXT: stxv v31, -48(r30) # 16-byte Folded Spill
|
|
; CHECK-LE-WACC-NEXT: lxv v31, 0(r3)
|
|
; CHECK-LE-WACC-NEXT: lxv v30, 0(r4)
|
|
; CHECK-LE-WACC-NEXT: addi r3, r1, 32
|
|
; CHECK-LE-WACC-NEXT: std r29, -24(r30) # 8-byte Folded Spill
|
|
; CHECK-LE-WACC-NEXT: vmr v2, v31
|
|
; CHECK-LE-WACC-NEXT: vmr v3, v30
|
|
; CHECK-LE-WACC-NEXT: mr r29, r5
|
|
; CHECK-LE-WACC-NEXT: bl _Z15buildVectorPairPu13__vector_pairDv16_hS0_@notoc
|
|
; CHECK-LE-WACC-NEXT: dmxxsetaccz wacc0
|
|
; CHECK-LE-WACC-NEXT: xvf32gerpp wacc0, v31, v30
|
|
; CHECK-LE-WACC-NEXT: lxvp vsp34, 32(r1)
|
|
; CHECK-LE-WACC-NEXT: xvf32gerpp wacc0, v3, v2
|
|
; CHECK-LE-WACC-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc0, 0
|
|
; CHECK-LE-WACC-NEXT: dmxxinstdmr512 wacc0, vsp36, vsp34, 0
|
|
; CHECK-LE-WACC-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc0, 0
|
|
; CHECK-LE-WACC-NEXT: stxv v5, 0(r29)
|
|
; CHECK-LE-WACC-NEXT: pstxv v4, 8(r29), 0
|
|
; CHECK-LE-WACC-NEXT: stxv v3, 16(r29)
|
|
; CHECK-LE-WACC-NEXT: pstxv v2, 24(r29), 0
|
|
; CHECK-LE-WACC-NEXT: lxv v31, -48(r30) # 16-byte Folded Reload
|
|
; CHECK-LE-WACC-NEXT: lxv v30, -64(r30) # 16-byte Folded Reload
|
|
; CHECK-LE-WACC-NEXT: ld r29, -24(r30) # 8-byte Folded Reload
|
|
; CHECK-LE-WACC-NEXT: mr r1, r30
|
|
; CHECK-LE-WACC-NEXT: ld r0, 16(r1)
|
|
; CHECK-LE-WACC-NEXT: ld r30, -16(r1)
|
|
; CHECK-LE-WACC-NEXT: mtlr r0
|
|
; CHECK-LE-WACC-NEXT: blr
|
|
;
|
|
; CHECK-BE-WACC-LABEL: testMultiply:
|
|
; CHECK-BE-WACC: # %bb.0: # %entry
|
|
; CHECK-BE-WACC-NEXT: mflr r0
|
|
; CHECK-BE-WACC-NEXT: std r30, -16(r1)
|
|
; CHECK-BE-WACC-NEXT: std r0, 16(r1)
|
|
; CHECK-BE-WACC-NEXT: clrldi r0, r1, 59
|
|
; CHECK-BE-WACC-NEXT: subfic r0, r0, -224
|
|
; CHECK-BE-WACC-NEXT: mr r30, r1
|
|
; CHECK-BE-WACC-NEXT: stdux r1, r1, r0
|
|
; CHECK-BE-WACC-NEXT: stxv v30, -64(r30) # 16-byte Folded Spill
|
|
; CHECK-BE-WACC-NEXT: stxv v31, -48(r30) # 16-byte Folded Spill
|
|
; CHECK-BE-WACC-NEXT: lxv v31, 0(r3)
|
|
; CHECK-BE-WACC-NEXT: lxv v30, 0(r4)
|
|
; CHECK-BE-WACC-NEXT: addi r3, r1, 128
|
|
; CHECK-BE-WACC-NEXT: std r29, -24(r30) # 8-byte Folded Spill
|
|
; CHECK-BE-WACC-NEXT: vmr v2, v31
|
|
; CHECK-BE-WACC-NEXT: vmr v3, v30
|
|
; CHECK-BE-WACC-NEXT: mr r29, r5
|
|
; CHECK-BE-WACC-NEXT: bl _Z15buildVectorPairPu13__vector_pairDv16_hS0_
|
|
; CHECK-BE-WACC-NEXT: nop
|
|
; CHECK-BE-WACC-NEXT: dmxxsetaccz wacc0
|
|
; CHECK-BE-WACC-NEXT: xvf32gerpp wacc0, v31, v30
|
|
; CHECK-BE-WACC-NEXT: lxvp vsp34, 128(r1)
|
|
; CHECK-BE-WACC-NEXT: xvf32gerpp wacc0, v2, v3
|
|
; CHECK-BE-WACC-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc0, 0
|
|
; CHECK-BE-WACC-NEXT: vmr v1, v2
|
|
; CHECK-BE-WACC-NEXT: vmr v7, v4
|
|
; CHECK-BE-WACC-NEXT: vmr v0, v3
|
|
; CHECK-BE-WACC-NEXT: vmr v6, v5
|
|
; CHECK-BE-WACC-NEXT: dmxxinstdmr512 wacc0, vsp38, vsp32, 0
|
|
; CHECK-BE-WACC-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc0, 0
|
|
; CHECK-BE-WACC-NEXT: stxv v2, 0(r29)
|
|
; CHECK-BE-WACC-NEXT: pstxv v3, 8(r29), 0
|
|
; CHECK-BE-WACC-NEXT: stxv v4, 16(r29)
|
|
; CHECK-BE-WACC-NEXT: pstxv v5, 24(r29), 0
|
|
; CHECK-BE-WACC-NEXT: lxv v31, -48(r30) # 16-byte Folded Reload
|
|
; CHECK-BE-WACC-NEXT: lxv v30, -64(r30) # 16-byte Folded Reload
|
|
; CHECK-BE-WACC-NEXT: ld r29, -24(r30) # 8-byte Folded Reload
|
|
; CHECK-BE-WACC-NEXT: mr r1, r30
|
|
; CHECK-BE-WACC-NEXT: ld r0, 16(r1)
|
|
; CHECK-BE-WACC-NEXT: ld r30, -16(r1)
|
|
; CHECK-BE-WACC-NEXT: mtlr r0
|
|
; CHECK-BE-WACC-NEXT: blr
|
|
entry:
|
|
%vP = alloca <256 x i1>, align 32
|
|
call void @llvm.lifetime.start.p0(i64 32, ptr nonnull %vP)
|
|
%0 = tail call <512 x i1> @llvm.ppc.mma.xxsetaccz()
|
|
%1 = load <16 x i8>, ptr %a, align 16
|
|
%2 = load <16 x i8>, ptr %b, align 16
|
|
call void @_Z15buildVectorPairPu13__vector_pairDv16_hS0_(ptr noundef nonnull %vP, <16 x i8> noundef %1, <16 x i8> noundef %2)
|
|
%3 = call <512 x i1> @llvm.ppc.mma.xvf32gerpp(<512 x i1> %0, <16 x i8> %1, <16 x i8> %2)
|
|
%4 = load <256 x i1>, ptr %vP, align 32
|
|
%5 = call { <16 x i8>, <16 x i8> } @llvm.ppc.vsx.disassemble.pair(<256 x i1> %4)
|
|
%6 = extractvalue { <16 x i8>, <16 x i8> } %5, 0
|
|
%7 = extractvalue { <16 x i8>, <16 x i8> } %5, 1
|
|
%8 = call <512 x i1> @llvm.ppc.mma.xvf32gerpp(<512 x i1> %3, <16 x i8> %6, <16 x i8> %7)
|
|
%9 = call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.ppc.mma.disassemble.acc(<512 x i1> %8)
|
|
%10 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %9, 0
|
|
%11 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %9, 1
|
|
%12 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %9, 2
|
|
%13 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %9, 3
|
|
%14 = call <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8> %13, <16 x i8> %12, <16 x i8> %11, <16 x i8> %10)
|
|
%15 = call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.ppc.mma.disassemble.acc(<512 x i1> %14)
|
|
%16 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %15, 0
|
|
%17 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %15, 1
|
|
%18 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %15, 2
|
|
%19 = extractvalue { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %15, 3
|
|
store <16 x i8> %16, ptr %c, align 16
|
|
%add.ptr = getelementptr inbounds float, ptr %c, i64 2
|
|
store <16 x i8> %17, ptr %add.ptr, align 16
|
|
%add.ptr11 = getelementptr inbounds float, ptr %c, i64 4
|
|
store <16 x i8> %18, ptr %add.ptr11, align 16
|
|
%add.ptr13 = getelementptr inbounds float, ptr %c, i64 6
|
|
store <16 x i8> %19, ptr %add.ptr13, align 16
|
|
call void @llvm.lifetime.end.p0(i64 32, ptr nonnull %vP)
|
|
ret void
|
|
}
|
|
|
|
declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
|
|
declare <512 x i1> @llvm.ppc.mma.xxsetaccz()
|
|
declare void @_Z15buildVectorPairPu13__vector_pairDv16_hS0_(ptr noundef, <16 x i8> noundef, <16 x i8> noundef) local_unnamed_addr
|
|
declare <512 x i1> @llvm.ppc.mma.xvf32gerpp(<512 x i1>, <16 x i8>, <16 x i8>)
|
|
declare { <16 x i8>, <16 x i8> } @llvm.ppc.vsx.disassemble.pair(<256 x i1>)
|
|
declare { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.ppc.mma.disassemble.acc(<512 x i1>)
|
|
declare <512 x i1> @llvm.ppc.mma.assemble.acc(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>)
|
|
declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
|
|
|
|
attributes #0 = { nounwind }
|