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
153 lines
13 KiB
YAML
153 lines
13 KiB
YAML
# REQUIRES: asserts
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx908 -verify-misched -run-pass=machine-scheduler -debug-only=machine-scheduler -o - %s 2>&1 | FileCheck -check-prefix=GCN %s
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx908 -verify-misched -passes=machine-scheduler -debug-only=machine-scheduler -o - %s 2>&1 | FileCheck -check-prefix=GCN %s
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx908 -verify-misched -run-pass=machine-scheduler -amdgpu-use-amdgpu-trackers=1 -debug-only=machine-scheduler -o - %s 2>&1 | FileCheck -check-prefix=GCN-GCNTRACKER %s
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx908 -verify-misched -passes=machine-scheduler -amdgpu-use-amdgpu-trackers=1 -debug-only=machine-scheduler -o - %s 2>&1 | FileCheck -check-prefix=GCN-GCNTRACKER %s
|
|
|
|
--- |
|
|
define amdgpu_kernel void @high-RP-reschedule() { ret void }
|
|
...
|
|
|
|
# GCN: Unclustered High Register Pressure Reschedule stage successfully increased occupancy to 4
|
|
|
|
# GCN-GCNTRACKER: Occupancy before scheduling: 3, after 4.
|
|
# GCN-GCNTRACKER-NEXT: Ending scheduling stage: Max Occupancy Initial Schedule
|
|
|
|
# When using the GCN Trackers, the scheduler is able to acieve desired occupancy without running high-RP-reschedule stage.
|
|
|
|
---
|
|
name: high-RP-reschedule
|
|
tracksRegLiveness: true
|
|
machineFunctionInfo:
|
|
occupancy: 4
|
|
body: |
|
|
bb.0:
|
|
%0:vreg_128 = IMPLICIT_DEF
|
|
%1:vreg_128 = IMPLICIT_DEF
|
|
%2:vgpr_32 = IMPLICIT_DEF
|
|
%3:vreg_128 = IMPLICIT_DEF
|
|
%4:vreg_128 = IMPLICIT_DEF
|
|
%5:vreg_128 = IMPLICIT_DEF
|
|
%6:vreg_128 = IMPLICIT_DEF
|
|
%7:vreg_128 = IMPLICIT_DEF
|
|
%8:vreg_128 = IMPLICIT_DEF
|
|
%9:vreg_128 = IMPLICIT_DEF
|
|
%10:vreg_128 = IMPLICIT_DEF
|
|
%11:sreg_64_xexec_xnull = IMPLICIT_DEF
|
|
%12:vreg_64 = IMPLICIT_DEF
|
|
|
|
bb.1:
|
|
%13:vgpr_32 = V_LSHRREV_B16_e32 1, %12.sub0, implicit $exec
|
|
%14:vgpr_32 = V_AND_B32_e32 127, %13, implicit $exec
|
|
%15:vgpr_32 = V_MUL_LO_U16_e32 49, %14, implicit $exec
|
|
%16:vgpr_32 = V_LSHRREV_B16_e32 10, %15, implicit $exec
|
|
%17:vgpr_32 = V_MUL_LO_U16_e32 42, %16, implicit $exec
|
|
%18:vgpr_32 = V_SUB_U16_e32 %12.sub0, %17, implicit $exec
|
|
%19:vgpr_32 = V_MOV_B32_e32 7, implicit $exec
|
|
%20:vgpr_32 = V_MUL_U32_U24_sdwa 0, %18, 0, %19, 0, 6, 0, 0, 6, implicit $exec
|
|
%21:vgpr_32 = V_LSHLREV_B32_e32 4, %20, implicit $exec
|
|
%22:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 608, 0, implicit $exec :: (load (s128))
|
|
%23:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 576, 0, implicit $exec :: (load (s128))
|
|
%24:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 592, 0, implicit $exec :: (load (s128))
|
|
%25:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 624, 0, implicit $exec :: (load (s128))
|
|
%26:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 672, 0, implicit $exec :: (load (s128))
|
|
%27:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 640, 0, implicit $exec :: (load (s128))
|
|
%28:vreg_128 = GLOBAL_LOAD_DWORDX4_SADDR %11, %21, 656, 0, implicit $exec :: (load (s128))
|
|
%29:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %7.sub2_sub3, 0, %25.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%30:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %7.sub2_sub3, 0, %25.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%31:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %7.sub0_sub1, 0, %25.sub0_sub1, 1, %29, 0, 0, implicit $mode, implicit $exec
|
|
%32:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %7.sub0_sub1, 0, %25.sub2_sub3, 0, %30, 0, 0, implicit $mode, implicit $exec
|
|
%33:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %5.sub2_sub3, 0, %24.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%34:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %5.sub2_sub3, 0, %24.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%35:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %5.sub0_sub1, 0, %24.sub0_sub1, 1, %33, 0, 0, implicit $mode, implicit $exec
|
|
%36:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %5.sub0_sub1, 0, %24.sub2_sub3, 0, %34, 0, 0, implicit $mode, implicit $exec
|
|
%37:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %9.sub2_sub3, 0, %28.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%38:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %9.sub2_sub3, 0, %28.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%39:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %9.sub0_sub1, 0, %28.sub0_sub1, 1, %37, 0, 0, implicit $mode, implicit $exec
|
|
%40:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %9.sub0_sub1, 0, %28.sub2_sub3, 0, %38, 0, 0, implicit $mode, implicit $exec
|
|
%41:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %4.sub2_sub3, 0, %23.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%42:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %4.sub2_sub3, 0, %23.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%43:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %4.sub0_sub1, 0, %23.sub0_sub1, 1, %41, 0, 0, implicit $mode, implicit $exec
|
|
%44:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %4.sub0_sub1, 0, %23.sub2_sub3, 0, %42, 0, 0, implicit $mode, implicit $exec
|
|
%45:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %8.sub2_sub3, 0, %27.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%46:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %8.sub2_sub3, 0, %27.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%47:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %6.sub2_sub3, 0, %22.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%48:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %8.sub0_sub1, 0, %27.sub0_sub1, 1, %45, 0, 0, implicit $mode, implicit $exec
|
|
%49:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %8.sub0_sub1, 0, %27.sub2_sub3, 0, %46, 0, 0, implicit $mode, implicit $exec
|
|
%50:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %10.sub2_sub3, 0, %26.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%51:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %6.sub2_sub3, 0, %22.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%52:vreg_64 = contract nofpexcept V_MUL_F64_e64 0, %10.sub2_sub3, 0, %26.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%53:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %6.sub0_sub1, 0, %22.sub0_sub1, 1, %47, 0, 0, implicit $mode, implicit $exec
|
|
%54:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %10.sub0_sub1, 0, %26.sub0_sub1, 1, %50, 0, 0, implicit $mode, implicit $exec
|
|
%55:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %6.sub0_sub1, 0, %22.sub2_sub3, 0, %51, 0, 0, implicit $mode, implicit $exec
|
|
%56:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %10.sub0_sub1, 0, %26.sub2_sub3, 0, %52, 0, 0, implicit $mode, implicit $exec
|
|
%57:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %3.sub2_sub3, 1, %32, 0, 0, implicit $mode, implicit $exec
|
|
%58:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %35, 1, %39, 0, 0, implicit $mode, implicit $exec
|
|
%59:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %44, 1, %49, 0, 0, implicit $mode, implicit $exec
|
|
%60:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %53, 1, %54, 0, 0, implicit $mode, implicit $exec
|
|
%61:sreg_64 = S_MOV_B64_IMM_PSEUDO 4604544271217802189
|
|
%62:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %3.sub0_sub1, 1, %31, 0, 0, implicit $mode, implicit $exec
|
|
undef %63.sub1:sreg_64 = S_MOV_B32 -1075404642
|
|
%64:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %36, 1, %40, 0, 0, implicit $mode, implicit $exec
|
|
%65:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %43, 1, %48, 0, 0, implicit $mode, implicit $exec
|
|
%66:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %55, 1, %56, 0, 0, implicit $mode, implicit $exec
|
|
%67:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %57, 0, %58, 0, 0, implicit $mode, implicit $exec
|
|
%68:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %59, 0, %60, 0, 0, implicit $mode, implicit $exec
|
|
%69:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %3.sub2_sub3, 0, 4611686018427387904, 1, %57, 0, 0, implicit $mode, implicit $exec
|
|
%70:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %3.sub0_sub1, 0, 4611686018427387904, 1, %62, 0, 0, implicit $mode, implicit $exec
|
|
%71:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %35, 0, 4611686018427387904, 1, %58, 0, 0, implicit $mode, implicit $exec
|
|
%72:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %36, 0, 4611686018427387904, 1, %64, 0, 0, implicit $mode, implicit $exec
|
|
%73:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %43, 0, 4611686018427387904, 1, %65, 0, 0, implicit $mode, implicit $exec
|
|
%74:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %44, 0, 4611686018427387904, 1, %59, 0, 0, implicit $mode, implicit $exec
|
|
%75:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %55, 0, 4611686018427387904, 1, %66, 0, 0, implicit $mode, implicit $exec
|
|
%76:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %53, 0, 4611686018427387904, 1, %60, 0, 0, implicit $mode, implicit $exec
|
|
%77:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %62, 1, %64, 0, 0, implicit $mode, implicit $exec
|
|
%78:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %65, 1, %66, 0, 0, implicit $mode, implicit $exec
|
|
%79:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %70, 1, %71, 0, 0, implicit $mode, implicit $exec
|
|
%80:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %57, 0, 4611686018427387904, 1, %67, 0, 0, implicit $mode, implicit $exec
|
|
%81:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %59, 0, 4611686018427387904, 1, %68, 0, 0, implicit $mode, implicit $exec
|
|
%82:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %69, 1, %72, 0, 0, implicit $mode, implicit $exec
|
|
%83:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %74, 1, %75, 0, 0, implicit $mode, implicit $exec
|
|
%84:vreg_64 = contract nofpexcept V_ADD_F64_e64 0, %73, 1, %76, 0, 0, implicit $mode, implicit $exec
|
|
%85:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %62, 0, 4611686018427387904, 1, %77, 0, 0, implicit $mode, implicit $exec
|
|
%86:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %65, 0, 4611686018427387904, 1, %78, 0, 0, implicit $mode, implicit $exec
|
|
%63.sub0:sreg_64 = COPY %61.sub0
|
|
%87:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %68, 0, %61, 0, %67, 0, 0, implicit $mode, implicit $exec
|
|
%88:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %81, 0, %63, 0, %80, 0, 0, implicit $mode, implicit $exec
|
|
%89:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %70, 0, 4611686018427387904, 1, %79, 0, 0, implicit $mode, implicit $exec
|
|
%90:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %69, 0, 4611686018427387904, 1, %82, 0, 0, implicit $mode, implicit $exec
|
|
%91:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %74, 0, 4611686018427387904, 1, %83, 0, 0, implicit $mode, implicit $exec
|
|
%92:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %73, 0, 4611686018427387904, 1, %84, 0, 0, implicit $mode, implicit $exec
|
|
%93:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %86, 0, %63, 0, %85, 0, 0, implicit $mode, implicit $exec
|
|
%94:vreg_64 = contract nofpexcept V_FMA_F64_e64 0, %78, 0, %61, 0, %77, 0, 0, implicit $mode, implicit $exec
|
|
undef %95.sub2_sub3:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %78, 0, %61, 0, %87, 0, 0, implicit $mode, implicit $exec
|
|
undef %96.sub2_sub3:vreg_128 = contract nofpexcept V_ADD_F64_e64 0, %82, 0, %84, 0, 0, implicit $mode, implicit $exec
|
|
undef %97.sub2_sub3:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %86, 0, %61, 0, %88, 0, 0, implicit $mode, implicit $exec
|
|
undef %98.sub2_sub3:vreg_128 = contract nofpexcept V_ADD_F64_e64 0, %90, 1, %91, 0, 0, implicit $mode, implicit $exec
|
|
%98.sub0_sub1:vreg_128 = contract nofpexcept V_ADD_F64_e64 0, %89, 1, %92, 0, 0, implicit $mode, implicit $exec
|
|
%97.sub0_sub1:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %81, 0, %63, 0, %93, 0, 0, implicit $mode, implicit $exec
|
|
%96.sub0_sub1:vreg_128 = contract nofpexcept V_ADD_F64_e64 0, %79, 1, %83, 0, 0, implicit $mode, implicit $exec
|
|
%95.sub0_sub1:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %68, 0, %63, 0, %94, 0, 0, implicit $mode, implicit $exec
|
|
undef %99.sub2_sub3:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %67, 0, 4611686018427387904, 1, %95.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
undef %100.sub2_sub3:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %82, 0, 4611686018427387904, 1, %96.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
undef %101.sub2_sub3:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %80, 0, 4611686018427387904, 1, %97.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
undef %102.sub2_sub3:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %90, 0, 4611686018427387904, 1, %98.sub2_sub3, 0, 0, implicit $mode, implicit $exec
|
|
%102.sub0_sub1:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %89, 0, 4611686018427387904, 1, %98.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%101.sub0_sub1:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %85, 0, 4611686018427387904, 1, %97.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%100.sub0_sub1:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %79, 0, 4611686018427387904, 1, %96.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%99.sub0_sub1:vreg_128 = contract nofpexcept V_FMA_F64_e64 0, %77, 0, 4611686018427387904, 1, %95.sub0_sub1, 0, 0, implicit $mode, implicit $exec
|
|
%103:vgpr_32 = V_ADD_U32_sdwa 0, %2, 0, %18, 0, 6, 0, 6, 0, implicit $exec
|
|
%104:vgpr_32 = V_LSHL_ADD_U32_e64 %103, 4, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %102, 0, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %101, 672, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %100, 1344, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %99, 2016, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %98, 2688, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %97, 3360, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %96, 4032, 0, implicit $exec
|
|
DS_WRITE_B128_gfx9 %104, %95, 4704, 0, implicit $exec
|
|
|
|
bb.2:
|
|
S_ENDPGM 0, implicit %0, implicit %1
|
|
...
|