GlobalParameters:
  MinimumRequiredVersion: 4.4.0
  PrintLevel: 1
  ForceRedoBenchmarkProblems: True
  ForceRedoLibraryLogic: True
  ForceRedoLibraryClient: True
  CMakeBuildType: Release
  EnqueuesPerSync: 1
  SyncsPerBenchmark: 1
  LibraryPrintDebug: False
  NumElementsToValidate: -1
  ValidationMaxToPrint: 4
  ValidationPrintValids: False
  ShortNames: False
  MergeFiles: True
  Platform: 0
  Device: 0
  KernelTime: True
  PinClocks: True
  SleepPercent: 200
  DataInitTypeBeta : 0
  PrintSolutionRejectionReason: True
  DataInitTypeA: 2
  DataInitTypeB: 5
  #DataInitTypeC: 0
  #DataInitTypeD: 0
  DataInitTypeBeta: 0
  DataInitTypeAlpha: 1
  PrintTensorA: 0x1
  PrintTensorB: 0x1
  PrintTensorD: 0x2

BenchmarkProblems:
  ########################################
  # NT - standard
  ########################################
  -
    - # ProblemType
      OperationType: GEMM
      DataType: s
      TransposeA: True
      TransposeB: False
      UseBeta: True
      Batched: True

    - # BenchmarkProblemSizeGroup - Standard
      InitialSolutionParameters:
      BenchmarkCommonParameters:
        - KernelLanguage: ["Assembly"]
        - EdgeType: ["ShiftPtr"]
        #- LoopTail: [True]
        - PrefetchLocalRead: [True]
      ForkParameters:
        - MatrixInstruction:
          - [32, 32, 1, 2]
        - PrefetchGlobalRead: [True]
        - ThreadTile:
          - [ 2, 32 ]
        - WorkGroup:
          - [ 16, 16, 1 ]
        - WorkGroupMapping: [8]
        - GlobalSplitU: [1]
        - DepthU: [ 8 ]
        - VectorWidth: [1]
      BenchmarkForkParameters:
      JoinParameters:
        - MacroTile
        - GlobalSplitU
      BenchmarkJoinParameters:
      BenchmarkFinalParameters:
        - ProblemSizes:
          - Range: [ [64], [256], [1], [64] ]

  ########################################