TestParameters: marks: [emulation5] GlobalParameters: MinimumRequiredVersion: 4.2.0 NumElementsToValidate: -1 BoundsCheck: True KernelTime: True BenchmarkProblems: - # hgemm TT - # ProblemType OperationType: GEMM DataType: h DestDataType: h TransposeA: True TransposeB: True UseBeta: True Batched: True - # BenchmarkProblemSizeGroup - Assembly InitialSolutionParameters: BenchmarkCommonParameters: - LoopTail: [True] - EdgeType: ["ShiftPtr"] - KernelLanguage: ["Assembly"] ForkParameters: - PrefetchLocalRead: [False] - PrefetchGlobalRead: [False] - ThreadTile: - [ 16, 4 ] - [ 16, 8 ] - [ 8, 8 ] - WorkGroup: - [ 16, 16, 1 ] - [ 8, 8, 1 ] - DepthU: [32] - VectorWidth: [-1] - AssertSummationElementMultiple: [1,2] - AssertFree0ElementMultiple: [1,2] BenchmarkForkParameters: JoinParameters: BenchmarkJoinParameters: BenchmarkFinalParameters: - ProblemSizes: - Range: [ [126,1,130], 0, [2], [62,1,66] ] # - # BenchmarkProblemSizeGroup - Assembly # InitialSolutionParameters: # BenchmarkCommonParameters: # - LoopTail: [True] # - EdgeType: ["ShiftPtr"] # ForkParameters: # - KernelLanguage: ["Assembly"] # - PrefetchLocalRead: [False] # - PrefetchGlobalRead: [False] # - ThreadTile: # - [ 8, 2 ] # - [ 2, 2 ] # - [ 4, 2 ] # - [ 8, 4 ] # - WorkGroup: # - [ 16, 16, 1 ] # - [ 8, 8, 1 ] # - DepthU: [16] # - VectorWidth: [-1] # - AssertSummationElementMultiple: [1,2] # - AssertFree0ElementMultiple: [1,2] # BenchmarkForkParameters: # JoinParameters: # BenchmarkJoinParameters: # BenchmarkFinalParameters: # - ProblemSizes: # - Range: [ [126,1,130], 0, [2], [62,1,66] ]