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
149 lines
5.9 KiB
LLVM
149 lines
5.9 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
|
|
; RUN: -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr -mcpu=future < %s | FileCheck %s
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-aix \
|
|
; RUN: -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr -mcpu=future < %s | FileCheck %s --check-prefix=AIX
|
|
; RUN: llc -verify-machineinstrs -mtriple=powerpc-unknown-aix \
|
|
; RUN: -ppc-asm-full-reg-names \
|
|
; RUN: -ppc-vsr-nums-as-vr -mcpu=future < %s | FileCheck %s --check-prefix=AIX32
|
|
|
|
declare <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1>, <256 x i1>, <16 x i8>)
|
|
declare void @dummy_func()
|
|
|
|
define void @spillDMRreg(ptr %vop, ptr %vpp, ptr %vcp, ptr %resp) nounwind {
|
|
; CHECK-LABEL: spillDMRreg:
|
|
; CHECK: # %bb.0:
|
|
; CHECK-NEXT: mflr r0
|
|
; CHECK-NEXT: std r30, -16(r1) # 8-byte Folded Spill
|
|
; CHECK-NEXT: std r0, 16(r1)
|
|
; CHECK-NEXT: stdu r1, -176(r1)
|
|
; CHECK-NEXT: lxvp vsp34, 0(r3)
|
|
; CHECK-NEXT: lxvp vsp36, 32(r3)
|
|
; CHECK-NEXT: lxvp vsp32, 64(r3)
|
|
; CHECK-NEXT: lxvp vsp38, 96(r3)
|
|
; CHECK-NEXT: lxvp vsp40, 0(r4)
|
|
; CHECK-NEXT: lxv vs0, 0(r5)
|
|
; CHECK-NEXT: dmxxinstdmr512 wacc_hi0, vsp36, vsp34, 1
|
|
; CHECK-NEXT: dmxxinstdmr512 wacc0, vsp38, vsp32, 0
|
|
; CHECK-NEXT: mr r30, r6
|
|
; CHECK-NEXT: dmxvbf16gerx2pp dmr0, vsp40, vs0
|
|
; CHECK-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc0, 0
|
|
; CHECK-NEXT: stxvp vsp36, 128(r1)
|
|
; CHECK-NEXT: stxvp vsp34, 96(r1)
|
|
; CHECK-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc_hi0, 1
|
|
; CHECK-NEXT: stxvp vsp36, 64(r1)
|
|
; CHECK-NEXT: stxvp vsp34, 32(r1)
|
|
; CHECK-NEXT: bl dummy_func@notoc
|
|
; CHECK-NEXT: lxvp vsp34, 128(r1)
|
|
; CHECK-NEXT: lxvp vsp36, 96(r1)
|
|
; CHECK-NEXT: dmxxinstdmr512 wacc0, vsp34, vsp36, 0
|
|
; CHECK-NEXT: lxvp vsp34, 64(r1)
|
|
; CHECK-NEXT: lxvp vsp36, 32(r1)
|
|
; CHECK-NEXT: dmxxinstdmr512 wacc_hi0, vsp34, vsp36, 1
|
|
; CHECK-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc0, 0
|
|
; CHECK-NEXT: stxvp vsp34, 96(r30)
|
|
; CHECK-NEXT: stxvp vsp36, 64(r30)
|
|
; CHECK-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc_hi0, 1
|
|
; CHECK-NEXT: stxvp vsp34, 32(r30)
|
|
; CHECK-NEXT: stxvp vsp36, 0(r30)
|
|
; CHECK-NEXT: addi r1, r1, 176
|
|
; CHECK-NEXT: ld r0, 16(r1)
|
|
; CHECK-NEXT: ld r30, -16(r1) # 8-byte Folded Reload
|
|
; CHECK-NEXT: mtlr r0
|
|
; CHECK-NEXT: blr
|
|
;
|
|
; AIX-LABEL: spillDMRreg:
|
|
; AIX: # %bb.0:
|
|
; AIX-NEXT: mflr r0
|
|
; AIX-NEXT: std r0, 16(r1)
|
|
; AIX-NEXT: stdu r1, -256(r1)
|
|
; AIX-NEXT: std r31, 248(r1) # 8-byte Folded Spill
|
|
; AIX-NEXT: lxvp vsp34, 96(r3)
|
|
; AIX-NEXT: lxvp vsp36, 64(r3)
|
|
; AIX-NEXT: lxvp vsp32, 32(r3)
|
|
; AIX-NEXT: lxvp vsp38, 0(r3)
|
|
; AIX-NEXT: lxvp vsp40, 0(r4)
|
|
; AIX-NEXT: lxv vs0, 0(r5)
|
|
; AIX-NEXT: dmxxinstdmr512 wacc_hi0, vsp36, vsp34, 1
|
|
; AIX-NEXT: dmxxinstdmr512 wacc0, vsp38, vsp32, 0
|
|
; AIX-NEXT: mr r31, r6
|
|
; AIX-NEXT: dmxvbf16gerx2pp dmr0, vsp40, vs0
|
|
; AIX-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc0, 0
|
|
; AIX-NEXT: stxvp vsp36, 112(r1)
|
|
; AIX-NEXT: stxvp vsp34, 144(r1)
|
|
; AIX-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc_hi0, 1
|
|
; AIX-NEXT: stxvp vsp36, 176(r1)
|
|
; AIX-NEXT: stxvp vsp34, 208(r1)
|
|
; AIX-NEXT: bl .dummy_func[PR]
|
|
; AIX-NEXT: nop
|
|
; AIX-NEXT: lxvp vsp34, 112(r1)
|
|
; AIX-NEXT: lxvp vsp36, 144(r1)
|
|
; AIX-NEXT: dmxxinstdmr512 wacc0, vsp34, vsp36, 0
|
|
; AIX-NEXT: lxvp vsp34, 176(r1)
|
|
; AIX-NEXT: lxvp vsp36, 208(r1)
|
|
; AIX-NEXT: dmxxinstdmr512 wacc_hi0, vsp34, vsp36, 1
|
|
; AIX-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc_hi0, 1
|
|
; AIX-NEXT: stxvp vsp36, 96(r31)
|
|
; AIX-NEXT: stxvp vsp34, 64(r31)
|
|
; AIX-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc0, 0
|
|
; AIX-NEXT: stxvp vsp36, 32(r31)
|
|
; AIX-NEXT: stxvp vsp34, 0(r31)
|
|
; AIX-NEXT: ld r31, 248(r1) # 8-byte Folded Reload
|
|
; AIX-NEXT: addi r1, r1, 256
|
|
; AIX-NEXT: ld r0, 16(r1)
|
|
; AIX-NEXT: mtlr r0
|
|
; AIX-NEXT: blr
|
|
;
|
|
; AIX32-LABEL: spillDMRreg:
|
|
; AIX32: # %bb.0:
|
|
; AIX32-NEXT: mflr r0
|
|
; AIX32-NEXT: stw r0, 8(r1)
|
|
; AIX32-NEXT: stwu r1, -208(r1)
|
|
; AIX32-NEXT: stw r31, 204(r1) # 4-byte Folded Spill
|
|
; AIX32-NEXT: lxvp vsp34, 96(r3)
|
|
; AIX32-NEXT: lxvp vsp36, 64(r3)
|
|
; AIX32-NEXT: lxvp vsp32, 32(r3)
|
|
; AIX32-NEXT: lxvp vsp38, 0(r3)
|
|
; AIX32-NEXT: lxvp vsp40, 0(r4)
|
|
; AIX32-NEXT: lxv vs0, 0(r5)
|
|
; AIX32-NEXT: dmxxinstdmr512 wacc_hi0, vsp36, vsp34, 1
|
|
; AIX32-NEXT: dmxxinstdmr512 wacc0, vsp38, vsp32, 0
|
|
; AIX32-NEXT: mr r31, r6
|
|
; AIX32-NEXT: dmxvbf16gerx2pp dmr0, vsp40, vs0
|
|
; AIX32-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc0, 0
|
|
; AIX32-NEXT: stxvp vsp36, 64(r1)
|
|
; AIX32-NEXT: stxvp vsp34, 96(r1)
|
|
; AIX32-NEXT: dmxxextfdmr512 vsp36, vsp34, wacc_hi0, 1
|
|
; AIX32-NEXT: stxvp vsp36, 128(r1)
|
|
; AIX32-NEXT: stxvp vsp34, 160(r1)
|
|
; AIX32-NEXT: bl .dummy_func[PR]
|
|
; AIX32-NEXT: nop
|
|
; AIX32-NEXT: lxvp vsp34, 64(r1)
|
|
; AIX32-NEXT: lxvp vsp36, 96(r1)
|
|
; AIX32-NEXT: dmxxinstdmr512 wacc0, vsp34, vsp36, 0
|
|
; AIX32-NEXT: lxvp vsp34, 128(r1)
|
|
; AIX32-NEXT: lxvp vsp36, 160(r1)
|
|
; AIX32-NEXT: dmxxinstdmr512 wacc_hi0, vsp34, vsp36, 1
|
|
; AIX32-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc_hi0, 1
|
|
; AIX32-NEXT: stxvp vsp36, 96(r31)
|
|
; AIX32-NEXT: stxvp vsp34, 64(r31)
|
|
; AIX32-NEXT: dmxxextfdmr512 vsp34, vsp36, wacc0, 0
|
|
; AIX32-NEXT: stxvp vsp36, 32(r31)
|
|
; AIX32-NEXT: stxvp vsp34, 0(r31)
|
|
; AIX32-NEXT: lwz r31, 204(r1) # 4-byte Folded Reload
|
|
; AIX32-NEXT: addi r1, r1, 208
|
|
; AIX32-NEXT: lwz r0, 8(r1)
|
|
; AIX32-NEXT: mtlr r0
|
|
; AIX32-NEXT: blr
|
|
%v.dmr = load <1024 x i1>, ptr %vop, align 64
|
|
%v1 = load <256 x i1>, ptr %vpp, align 32
|
|
%v2 = load <16 x i8>, ptr %vcp, align 32
|
|
%call = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> %v.dmr, <256 x i1> %v1, <16 x i8> %v2)
|
|
tail call void @dummy_func()
|
|
%call2 = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> %v.dmr, <256 x i1> %v1, <16 x i8> %v2)
|
|
store <1024 x i1> %call, ptr %resp, align 64
|
|
ret void
|
|
}
|