GlobalParameters: MinimumRequiredVersion: 4.4.0 PrintLevel: 1 ForceRedoBenchmarkProblems: True ForceRedoLibraryLogic: True ForceRedoLibraryClient: True CMakeBuildType: Release EnqueuesPerSync: 1 SyncsPerBenchmark: 2 LibraryPrintDebug: False NumElementsToValidate: 128 ValidationMaxToPrint: 4 ValidationPrintValids: False ShortNames: False MergeFiles: True Platform: 0 Device: 0 KernelTime: True PinClocks: True SleepPercent: 200 DataInitTypeBeta : 0 BenchmarkProblems: ######################################## # NN ######################################## - - # ProblemType OperationType: GEMM DataType: d TransposeA: False TransposeB: False UseBeta: True Batched: True - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [128] ] #source for k<=128 - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [256] ] - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [128] ] #source for k<=128 - Range: [ [1], [64], [1], [128] ] #source for k<=128 - Range: [ [64], [1], [1], [128] ] #source for k<=128 - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [256] ] - Range: [ [1], [64], [1], [256] ] - Range: [ [64], [1], [1], [256] ] ######################################## # NT ######################################## - - # ProblemType OperationType: GEMM DataType: d TransposeA: False TransposeB: True UseBeta: True Batched: True - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [128] ] # source for k<=128 - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [256] ] - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [128] ] # source for k<=128 - Range: [ [1], [64], [1], [128] ] # source for k<=128 - Range: [ [64], [1], [1], [128] ] # source for k<=128 - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [256] ] - Range: [ [1], [64], [1], [256] ] - Range: [ [64], [1], [1], [256] ] ######################################## # TN ######################################## - - # ProblemType OperationType: GEMM DataType: d TransposeA: True TransposeB: False UseBeta: True Batched: True - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [128] ] # source for k<=128 - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [256] ] - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [128] ] # source for k<=128 - Range: [ [1], [64], [1], [128] ] # source for k<=128 - Range: [ [64], [1], [1], [128] ] # source for k<=128 - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [256] ] - Range: [ [1], [64], [1], [256] ] - Range: [ [64], [1], [1], [256] ] ######################################## # TT ######################################## - - # ProblemType OperationType: GEMM DataType: d TransposeA: True TransposeB: True UseBeta: True Batched: True - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [128] ] # source for k<=128 - # BenchmarkProblemSizeGroup - Standard InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 8 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [64], [64], [1], [256] ] - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Source"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [128] ] # source for k<=128 - Range: [ [1], [64], [1], [128] ] # source for k<=128 - Range: [ [64], [1], [1], [128] ] # source for k<=128 - # BenchmarkProblemSizeGroup - M,N < VW InitialSolutionParameters: BenchmarkCommonParameters: - KernelLanguage: ["Assembly"] - EdgeType: ["ShiftPtr"] - LoopTail: [True] - WorkGroupMapping: [8] - PrefetchLocalRead: [True] - PrefetchGlobalRead: [True] - VectorWidth: [-1] ForkParameters: - ThreadTile: - [ 4, 4 ] - WorkGroup: - [ 16, 16, 1 ] - DepthU: [ 4 ] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [1], [1], [1], [256] ] - Range: [ [1], [64], [1], [256] ] - Range: [ [64], [1], [1], [256] ] LibraryLogic: # ScheduleName: "vega20" # DeviceNames: ["Device 66a0", "Device 66a1", "Device 66a7", "Device 66af", "Vega 20"] # ArchitectureName: "gfx906" ScheduleName: "vega10" DeviceNames: ["Device 6863", "Device 6862", "Device 687f", "Device 6860", "Device 6861", "Vega 10 XTX [Radeon Vega Frontier Edition]", "Vega [Radeon RX Vega]", Vega, Device 6864, Device 686c] ArchitectureName: "gfx900" # ScheduleName: "mi25" # DeviceNames: ["Device 6860"] # ArchitectureName: "gfx900" # ScheduleName: "r9nano" # DeviceNames: ["Device 7300"] # ArchitectureName: "gfx803" # ScheduleName: "hip" # DeviceNames: ["Device 0000"] # ArchitectureName: "fallback" LibraryClient: