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
475 lines
16 KiB
YAML
475 lines
16 KiB
YAML
TestParameters:
|
|
marks: [skip-gfx900, skip-gfx906, skip-gfx908, skip-gfx90a, skip-gfx950, skip-gfx1010, skip-gfx1011, skip-gfx1012, skip-gfx1030, skip-gfx1100, skip-gfx1101, skip-gfx1102, skip-gfx1200, skip-gfx1201, skip-gfx1250] # not supported by arch
|
|
|
|
GlobalParameters:
|
|
SyncsPerBenchmark: 0
|
|
NumElementsToValidate: -1
|
|
MinimumRequiredVersion: 5.0.0
|
|
PrintLevel: 1
|
|
#PrintSolutionRejectionReason: True
|
|
Device: 0
|
|
CMakeBuildType: Release
|
|
KernelTime: True
|
|
MaxWorkspaceSize: 13421772800
|
|
DataInitTypeA: 13
|
|
DataInitTypeB: 12
|
|
DataInitTypeAlpha: 1
|
|
DataInitTypeBeta: 1
|
|
DataInitTypeBias: 13
|
|
DataInitTypeScaleAlphaVec: 12
|
|
BoundsCheck: 2
|
|
|
|
BenchmarkProblems:
|
|
########################################
|
|
# HHS TN SWIZZLED_B + BIAS + Activation + SAV
|
|
########################################
|
|
-
|
|
- # ProblemType
|
|
OperationType: GEMM
|
|
DataType: h
|
|
DestDataType: h
|
|
ComputeDataType: s
|
|
HighPrecisionAccumulate: True
|
|
TransposeA: True
|
|
TransposeB: False
|
|
SwizzleTensorB: True
|
|
UseBeta: True
|
|
Batched: True
|
|
UseBias: 1
|
|
Activation: True
|
|
BiasDataTypeList: ['h']
|
|
UseScaleAlphaVec: 1
|
|
- # BenchmarkProblemSizeGroup - Standard - All problem
|
|
InitialSolutionParameters:
|
|
BenchmarkCommonParameters:
|
|
- KernelLanguage: ["Assembly"]
|
|
ForkParameters:
|
|
- MatrixInstruction:
|
|
- [16, 16, 16, 1, 1, 8, 2, 1,4 ] # MT = 128x128
|
|
- [16, 16, 16, 1, 1, 16, 2, 1,4 ] # MT = 256x128
|
|
- [16, 16, 16, 1, 1, 8, 4, 1,4 ] # MT = 128x256
|
|
- [16, 16, 16, 1, 1, 16, 4, 1,4 ] # MT = 256x256
|
|
- [16, 16, 16, 1, 1, 8, 8, 1,4 ] # MT = 128x512
|
|
- [16, 16, 16, 1, 1, 16, 8, 1,4 ] # MT = 256x512
|
|
- [16, 16, 16, 1, 1, 8, 5, 1,2 ] # MT = 128x160
|
|
|
|
- [16, 16, 16, 1, 1, 4, 2, 2,2 ] # MT = 128x64
|
|
- [16, 16, 16, 1, 1, 8, 2, 2,2 ] # MT = 256x64
|
|
- [16, 16, 16, 1, 1, 4, 4, 2,2 ] # MT = 128x128
|
|
- [16, 16, 16, 1, 1, 8, 4, 2,2 ] # MT = 256x128
|
|
- [16, 16, 16, 1, 1, 4, 8, 2,2 ] # MT = 128x256
|
|
- [16, 16, 16, 1, 1, 8, 8, 2,2 ] # MT = 256x256
|
|
- [16, 16, 16, 1, 1, 4, 16, 2,2 ] # MT = 128x512
|
|
- [16, 16, 16, 1, 1, 8, 16, 2,2 ] # MT = 256x512
|
|
- [16, 16, 16, 1, 1, 4, 5, 2,2 ] # MT = 128x160
|
|
- WorkGroup:
|
|
- [16, 16, 1] # LSU=1, minDU=32
|
|
- [16, 16, 2] # LSU=2, minDU=64
|
|
- [16, 16, 4] # LSU=4, minDU=128
|
|
- GlobalReadVectorWidthA: [-1]
|
|
- GlobalReadVectorWidthB: [8]
|
|
- PrefetchGlobalRead: [1,2]
|
|
- PrefetchLocalRead: [1,4]
|
|
- ClusterLocalRead: [1]
|
|
- NumElementsPerBatchStore: [0]
|
|
- DepthU: [32,64,128]
|
|
- VectorWidthA: [-1]
|
|
- VectorWidthB: [1]
|
|
- LocalWritePerMfma: [-1]
|
|
- StaggerU: [4]
|
|
- StaggerUStride: [256]
|
|
- StaggerUMapping: [0]
|
|
- WorkGroupMappingXCC: [8]
|
|
- ScheduleIterAlg: [3]
|
|
- LdsBlockSizePerPadA: [-1]
|
|
- LdsBlockSizePerPadB: [-1]
|
|
- StorePriorityOpt: [false]
|
|
- VectorStore: [-1]
|
|
- StoreSyncOpt: [0]
|
|
- LdsPadA: [-1]
|
|
- LdsPadB: [-1]
|
|
- 1LDSBuffer: [1]
|
|
- GlobalSplitU: [1,3]
|
|
- GlobalSplitUAlgorithm: ["MultipleBuffer", "MultipleBufferSingleKernel"]
|
|
- LocalReadVectorWidth: [2,4,8]
|
|
- DirectToVgprB: [true]
|
|
- UseSgprForGRO: [0,1]
|
|
BenchmarkJoinParameters:
|
|
BenchmarkFinalParameters:
|
|
- ProblemSizes:
|
|
- Exact: [256, 160, 1, 224]
|
|
- Exact: [256, 160, 1, 256]
|
|
- Exact: [256, 160, 1, 288]
|
|
- Exact: [512, 1600, 1, 992]
|
|
- Exact: [512, 1600, 1, 1024]
|
|
- Exact: [512, 1600, 1, 1056]
|
|
- Exact: [256, 512, 1, 224]
|
|
- Exact: [256, 512, 1, 256]
|
|
- Exact: [256, 512, 1, 288]
|
|
- Exact: [127, 127, 1, 127]
|
|
- Exact: [128, 127, 1, 127]
|
|
- Exact: [127, 128, 1, 127]
|
|
- Exact: [1, 127, 1, 127]
|
|
- Exact: [127, 1, 1, 127]
|
|
- BiasTypeArgs: ['h']
|
|
- ActivationArgs:
|
|
- [Enum: relu]
|
|
|
|
########################################
|
|
# HSS TN SWIZZLED_B
|
|
########################################
|
|
-
|
|
- # ProblemType
|
|
OperationType: GEMM
|
|
DataType: h
|
|
DestDataType: s
|
|
ComputeDataType: s
|
|
HighPrecisionAccumulate: True
|
|
TransposeA: True
|
|
TransposeB: False
|
|
SwizzleTensorB: True
|
|
UseBeta: True
|
|
Batched: True
|
|
- # BenchmarkProblemSizeGroup - Standard - All problem
|
|
InitialSolutionParameters:
|
|
BenchmarkCommonParameters:
|
|
- KernelLanguage: ["Assembly"]
|
|
ForkParameters:
|
|
- MatrixInstruction:
|
|
- [16, 16, 16, 1, 1, 8, 1, 1,4 ] # MT = 128x64
|
|
- [16, 16, 16, 1, 1, 16, 1, 1,4 ] # MT = 256x64
|
|
- [16, 16, 16, 1, 1, 8, 2, 1,4 ] # MT = 128x128
|
|
- [16, 16, 16, 1, 1, 16, 2, 1,4 ] # MT = 256x128
|
|
|
|
- [16, 16, 16, 1, 1, 4, 2, 2,2 ] # MT = 128x64
|
|
- [16, 16, 16, 1, 1, 8, 2, 2,2 ] # MT = 256x64
|
|
- [16, 16, 16, 1, 1, 4, 4, 2,2 ] # MT = 128x128
|
|
- WorkGroup:
|
|
- [16, 16, 1] # LSU=1, minDU=32
|
|
- [16, 16, 2] # LSU=2, minDU=64
|
|
- GlobalReadVectorWidthA: [-1]
|
|
- GlobalReadVectorWidthB: [8]
|
|
- PrefetchGlobalRead: [2]
|
|
- PrefetchLocalRead: [1]
|
|
- ClusterLocalRead: [1]
|
|
- NumElementsPerBatchStore: [0]
|
|
- DepthU: [32,64]
|
|
- VectorWidthA: [-1]
|
|
- VectorWidthB: [1,2]
|
|
- LocalWritePerMfma: [-1]
|
|
- StaggerU: [4]
|
|
- StaggerUStride: [256]
|
|
- StaggerUMapping: [0]
|
|
- WorkGroupMappingXCC: [8]
|
|
- ScheduleIterAlg: [3]
|
|
- LdsBlockSizePerPadA: [-1]
|
|
- LdsBlockSizePerPadB: [-1]
|
|
- StorePriorityOpt: [false]
|
|
- VectorStore: [-1]
|
|
- StoreSyncOpt: [0]
|
|
- LdsPadA: [-1]
|
|
- LdsPadB: [-1]
|
|
- 1LDSBuffer: [1]
|
|
- GlobalSplitU: [1,3]
|
|
- GlobalSplitUAlgorithm: ["MultipleBuffer"]
|
|
- LocalReadVectorWidth: [2,4,8]
|
|
- DirectToVgprB: [true]
|
|
- UseSgprForGRO: [0,1]
|
|
BenchmarkJoinParameters:
|
|
BenchmarkFinalParameters:
|
|
- ProblemSizes:
|
|
- Exact: [256, 160, 1, 224]
|
|
- Exact: [256, 160, 1, 256]
|
|
- Exact: [256, 160, 1, 288]
|
|
- Exact: [127, 127, 1, 127]
|
|
- Exact: [127, 128, 1, 127]
|
|
- Exact: [1, 127, 1, 127]
|
|
- Exact: [127, 1, 1, 127]
|
|
|
|
########################################
|
|
# BBS TN SWIZZLED_B
|
|
########################################
|
|
-
|
|
- # ProblemType
|
|
OperationType: GEMM
|
|
DataType: b
|
|
DestDataType: b
|
|
ComputeDataType: s
|
|
HighPrecisionAccumulate: True
|
|
TransposeA: True
|
|
TransposeB: False
|
|
SwizzleTensorB: True
|
|
UseBeta: True
|
|
Batched: True
|
|
- # BenchmarkProblemSizeGroup - Standard - All problem
|
|
InitialSolutionParameters:
|
|
BenchmarkCommonParameters:
|
|
- KernelLanguage: ["Assembly"]
|
|
ForkParameters:
|
|
- MatrixInstruction:
|
|
- [16, 16, 16, 1, 1, 8, 1, 1,4 ] # MT = 128x64
|
|
- [16, 16, 16, 1, 1, 16, 1, 1,4 ] # MT = 256x64
|
|
- [16, 16, 16, 1, 1, 8, 2, 1,4 ] # MT = 128x128
|
|
- [16, 16, 16, 1, 1, 16, 2, 1,4 ] # MT = 256x128
|
|
|
|
- [16, 16, 16, 1, 1, 4, 2, 2,2 ] # MT = 128x64
|
|
- [16, 16, 16, 1, 1, 8, 2, 2,2 ] # MT = 256x64
|
|
- [16, 16, 16, 1, 1, 4, 4, 2,2 ] # MT = 128x128
|
|
- WorkGroup:
|
|
- [16, 16, 1] # LSU=1, minDU=32
|
|
- [16, 16, 2] # LSU=2, minDU=64
|
|
- GlobalReadVectorWidthA: [-1]
|
|
- GlobalReadVectorWidthB: [8]
|
|
- PrefetchGlobalRead: [2]
|
|
- PrefetchLocalRead: [1]
|
|
- ClusterLocalRead: [1]
|
|
- NumElementsPerBatchStore: [0]
|
|
- DepthU: [32,64]
|
|
- VectorWidthA: [-1]
|
|
- VectorWidthB: [1,2]
|
|
- LocalWritePerMfma: [-1]
|
|
- StaggerU: [4]
|
|
- StaggerUStride: [256]
|
|
- StaggerUMapping: [0]
|
|
- WorkGroupMappingXCC: [8]
|
|
- ScheduleIterAlg: [3]
|
|
- LdsBlockSizePerPadA: [-1]
|
|
- LdsBlockSizePerPadB: [-1]
|
|
- StorePriorityOpt: [false]
|
|
- VectorStore: [-1]
|
|
- StoreSyncOpt: [0]
|
|
- LdsPadA: [-1]
|
|
- LdsPadB: [-1]
|
|
- 1LDSBuffer: [1]
|
|
- GlobalSplitU: [1,3]
|
|
- GlobalSplitUAlgorithm: ["MultipleBuffer"]
|
|
- LocalReadVectorWidth: [2,4,8]
|
|
- DirectToVgprB: [true]
|
|
- UseSgprForGRO: [0,1]
|
|
BenchmarkJoinParameters:
|
|
BenchmarkFinalParameters:
|
|
- ProblemSizes:
|
|
- Exact: [256, 160, 1, 224]
|
|
- Exact: [256, 160, 1, 256]
|
|
- Exact: [256, 160, 1, 288]
|
|
- Exact: [127, 127, 1, 127]
|
|
- Exact: [127, 128, 1, 127]
|
|
- Exact: [1, 127, 1, 127]
|
|
- Exact: [127, 1, 1, 127]
|
|
|
|
########################################
|
|
# F8B8NHS TN SWIZZLED_B
|
|
########################################
|
|
-
|
|
- # ProblemType
|
|
OperationType: GEMM
|
|
DataType: F8B8N
|
|
DestDataType: h
|
|
ComputeDataType: s
|
|
HighPrecisionAccumulate: True
|
|
TransposeA: True
|
|
TransposeB: False
|
|
SwizzleTensorB: True
|
|
UseBeta: True
|
|
Batched: True
|
|
- # BenchmarkProblemSizeGroup - Standard - All problem
|
|
InitialSolutionParameters:
|
|
BenchmarkCommonParameters:
|
|
- KernelLanguage: ["Assembly"]
|
|
ForkParameters:
|
|
- MatrixInstruction:
|
|
- [16, 16, 32, 1, 1, 8, 1, 1,4 ] # MT = 128x64
|
|
- [16, 16, 32, 1, 1, 16, 1, 1,4 ] # MT = 256x64
|
|
- [16, 16, 32, 1, 1, 8, 2, 1,4 ] # MT = 128x128
|
|
- [16, 16, 32, 1, 1, 16, 2, 1,4 ] # MT = 256x128
|
|
|
|
- [16, 16, 32, 1, 1, 4, 16, 2,2 ] # MT = 128x512
|
|
- [16, 16, 32, 1, 1, 8, 16, 2,2 ] # MT = 256x512
|
|
- [16, 16, 32, 1, 1, 4, 5, 2,2 ] # MT = 128x160
|
|
- WorkGroup:
|
|
- [16, 16, 1] # LSU=1, minDU=64
|
|
- [16, 16, 2] # LSU=2, minDU=128
|
|
- GlobalReadVectorWidthA: [-1]
|
|
- GlobalReadVectorWidthB: [16]
|
|
- PrefetchGlobalRead: [2]
|
|
- PrefetchLocalRead: [1]
|
|
- ClusterLocalRead: [1]
|
|
- NumElementsPerBatchStore: [0]
|
|
- DepthU: [64,128]
|
|
- VectorWidthA: [-1]
|
|
- VectorWidthB: [1,4]
|
|
- LocalWritePerMfma: [-1]
|
|
- StaggerU: [4]
|
|
- StaggerUStride: [256]
|
|
- StaggerUMapping: [0]
|
|
- WorkGroupMappingXCC: [8]
|
|
- ScheduleIterAlg: [3]
|
|
- LdsBlockSizePerPadA: [-1]
|
|
- LdsBlockSizePerPadB: [-1]
|
|
- StorePriorityOpt: [false]
|
|
- VectorStore: [-1]
|
|
- StoreSyncOpt: [0]
|
|
- LdsPadA: [-1]
|
|
- LdsPadB: [-1]
|
|
- 1LDSBuffer: [1]
|
|
- GlobalSplitU: [1,3]
|
|
- GlobalSplitUAlgorithm: ["MultipleBuffer"]
|
|
- LocalReadVectorWidth: [16]
|
|
- DirectToVgprB: [true]
|
|
- UseSgprForGRO: [0,1]
|
|
BenchmarkJoinParameters:
|
|
BenchmarkFinalParameters:
|
|
- ProblemSizes:
|
|
- Exact: [256, 160, 1, 224]
|
|
- Exact: [256, 160, 1, 256]
|
|
- Exact: [256, 160, 1, 288]
|
|
- Exact: [127, 127, 1, 127]
|
|
- Exact: [127, 128, 1, 127]
|
|
- Exact: [1, 127, 1, 127]
|
|
- Exact: [127, 1, 1, 127]
|
|
|
|
########################################
|
|
# XFloat TN SWIZZLED_B
|
|
########################################
|
|
-
|
|
- # ProblemType
|
|
OperationType: GEMM
|
|
DataType: s
|
|
DestDataType: s
|
|
ComputeDataType: s
|
|
F32XdlMathOp: X
|
|
HighPrecisionAccumulate: False
|
|
TransposeA: True
|
|
TransposeB: False
|
|
SwizzleTensorB: True
|
|
UseBeta: True
|
|
Batched: True
|
|
- # BenchmarkProblemSizeGroup - Standard - All problem
|
|
InitialSolutionParameters:
|
|
BenchmarkCommonParameters:
|
|
- KernelLanguage: ["Assembly"]
|
|
ForkParameters:
|
|
- MatrixInstruction:
|
|
- [16, 16, 8, 1, 1, 16, 4, 1,4 ] # MT = 256x256
|
|
- [16, 16, 8, 1, 1, 8, 8, 1,4 ] # MT = 128x512
|
|
- [16, 16, 8, 1, 1, 16, 8, 1,4 ] # MT = 256x512
|
|
- [16, 16, 8, 1, 1, 8, 5, 1,2 ] # MT = 128x160
|
|
|
|
- [16, 16, 8, 1, 1, 8, 4, 2,2 ] # MT = 256x128
|
|
- [16, 16, 8, 1, 1, 4, 8, 2,2 ] # MT = 128x256
|
|
- [16, 16, 8, 1, 1, 8, 8, 2,2 ] # MT = 256x256
|
|
- WorkGroup:
|
|
- [16, 16, 1] # LSU=1, minDU=16
|
|
- [16, 16, 2] # LSU=2, minDU=32
|
|
- [16, 16, 4] # LSU=4, minDU=64
|
|
- GlobalReadVectorWidthA: [-1]
|
|
- GlobalReadVectorWidthB: [4]
|
|
- PrefetchGlobalRead: [2]
|
|
- PrefetchLocalRead: [1]
|
|
- ClusterLocalRead: [1]
|
|
- NumElementsPerBatchStore: [0]
|
|
- DepthU: [32,64]
|
|
- VectorWidthA: [-1]
|
|
- VectorWidthB: [1,8]
|
|
- LocalWritePerMfma: [-1]
|
|
- StaggerU: [4]
|
|
- StaggerUStride: [256]
|
|
- StaggerUMapping: [0]
|
|
- WorkGroupMappingXCC: [8]
|
|
- ScheduleIterAlg: [3]
|
|
- LdsBlockSizePerPadA: [-1]
|
|
- LdsBlockSizePerPadB: [-1]
|
|
- StorePriorityOpt: [false]
|
|
- VectorStore: [-1]
|
|
- StoreSyncOpt: [0]
|
|
- LdsPadA: [-1]
|
|
- LdsPadB: [-1]
|
|
- 1LDSBuffer: [1]
|
|
- GlobalSplitU: [1,3]
|
|
- GlobalSplitUAlgorithm: ["MultipleBuffer"]
|
|
- LocalReadVectorWidth: [2,4,8]
|
|
- DirectToVgprB: [true]
|
|
- UseSgprForGRO: [0,1]
|
|
BenchmarkJoinParameters:
|
|
BenchmarkFinalParameters:
|
|
- ProblemSizes:
|
|
- Exact: [256, 160, 1, 224]
|
|
- Exact: [256, 160, 1, 256]
|
|
- Exact: [256, 160, 1, 288]
|
|
- Exact: [127, 127, 1, 127]
|
|
- Exact: [127, 128, 1, 127]
|
|
- Exact: [1, 127, 1, 127]
|
|
- Exact: [127, 1, 1, 127]
|
|
|
|
########################################
|
|
# SSS TN SWIZZLED_B
|
|
########################################
|
|
-
|
|
- # ProblemType
|
|
OperationType: GEMM
|
|
DataType: s
|
|
DestDataType: s
|
|
ComputeDataType: s
|
|
HighPrecisionAccumulate: False
|
|
TransposeA: True
|
|
TransposeB: False
|
|
SwizzleTensorB: True
|
|
UseBeta: True
|
|
Batched: True
|
|
- # BenchmarkProblemSizeGroup - Standard - All problem
|
|
InitialSolutionParameters:
|
|
BenchmarkCommonParameters:
|
|
- KernelLanguage: ["Assembly"]
|
|
ForkParameters:
|
|
- MatrixInstruction:
|
|
- [16, 16, 4, 1, 1, 8, 2, 1,4 ] # MT = 128x128
|
|
- [16, 16, 4, 1, 1, 16, 2, 1,4 ] # MT = 256x128
|
|
- [16, 16, 4, 1, 1, 8, 4, 1,4 ] # MT = 128x256
|
|
- [16, 16, 4, 1, 1, 16, 4, 1,4 ] # MT = 256x256
|
|
|
|
- [16, 16, 4, 1, 1, 4, 16, 2,2 ] # MT = 128x512
|
|
- [16, 16, 4, 1, 1, 8, 16, 2,2 ] # MT = 256x512
|
|
- [16, 16, 4, 1, 1, 4, 5, 2,2 ] # MT = 128x160
|
|
- WorkGroup:
|
|
- [16, 16, 1] # LSU=1, minDU=16
|
|
- [16, 16, 2] # LSU=2, minDU=32
|
|
- [16, 16, 4] # LSU=4, minDU=64
|
|
- GlobalReadVectorWidthA: [-1]
|
|
- GlobalReadVectorWidthB: [4]
|
|
- PrefetchGlobalRead: [2]
|
|
- PrefetchLocalRead: [1]
|
|
- ClusterLocalRead: [1]
|
|
- NumElementsPerBatchStore: [0]
|
|
- DepthU: [32,64]
|
|
- VectorWidthA: [-1]
|
|
- VectorWidthB: [1,2]
|
|
- LocalWritePerMfma: [-1]
|
|
- StaggerU: [4]
|
|
- StaggerUStride: [256]
|
|
- StaggerUMapping: [0]
|
|
- WorkGroupMappingXCC: [8]
|
|
- ScheduleIterAlg: [3]
|
|
- LdsBlockSizePerPadA: [-1]
|
|
- LdsBlockSizePerPadB: [-1]
|
|
- StorePriorityOpt: [false]
|
|
- VectorStore: [-1]
|
|
- StoreSyncOpt: [0]
|
|
- LdsPadA: [-1]
|
|
- LdsPadB: [-1]
|
|
- 1LDSBuffer: [1]
|
|
- GlobalSplitU: [1,3]
|
|
- GlobalSplitUAlgorithm: ["MultipleBuffer"]
|
|
- LocalReadVectorWidth: [2,4,8]
|
|
- DirectToVgprB: [true]
|
|
- UseSgprForGRO: [0,1]
|
|
BenchmarkJoinParameters:
|
|
BenchmarkFinalParameters:
|
|
- ProblemSizes:
|
|
- Exact: [256, 160, 1, 224]
|
|
- Exact: [256, 160, 1, 256]
|
|
- Exact: [256, 160, 1, 288]
|
|
- Exact: [127, 127, 1, 127]
|
|
- Exact: [127, 128, 1, 127]
|
|
- Exact: [1, 127, 1, 127]
|
|
- Exact: [127, 1, 1, 127]
|