Platform: AMD Accelerated Parallel Processing
  Device: gfx1100
    Driver version  : 3635.0 (HSA1.1,LC) (Linux x64)
    Compute units   : 48
    Clock frequency : 2371 MHz

    Global memory bandwidth (GBPS)
      float   : 718.66
      float2  : 714.01
      float4  : 776.41
      float8  : 814.28
      float16 : 843.46

    Single-precision compute (GFLOPS)
      float   : 29842.28
      float2  : 32297.03
      float4  : 34005.70
      float8  : 33138.76
      float16 : 28972.20

    Half-precision compute (GFLOPS)
      half   : 31209.96
      half2  : 61740.10
      half4  : 61625.68
      half8  : 60854.38
      half16 : 61260.65

    Double-precision compute (GFLOPS)
      double   : 1125.76
      double2  : 1121.55
      double4  : 1120.61
      double8  : 1112.81
      double16 : 1102.92

    Integer compute (GIOPS)
      int   : 6727.44
      int2  : 6779.56
      int4  : 6684.11
      int8  : 6645.27
      int16 : 6712.87

    Integer compute Fast 24bit (GIOPS)
      int   : 28233.15
      int2  : 27638.64
      int4  : 27601.14
      int8  : 27523.02
      int16 : 23518.67

    Integer char (8bit) compute (GIOPS)
      char   : 30417.02
      char2  : 17098.56
      char4  : 16796.54
      char8  : 13885.71
      char16 : 13536.93

    Integer short (16bit) compute (GIOPS)
      short   : 30685.04
      short2  : 29478.72
      short4  : 28988.49
      short8  : 28415.79
      short16 : 27920.62

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 17.56
      enqueueReadBuffer               : 4.61
      enqueueWriteBuffer non-blocking : 17.50
      enqueueReadBuffer non-blocking  : 4.61
      enqueueMapBuffer(for read)      : 753503.06
        memcpy from mapped ptr        : 4.54
      enqueueUnmap(after write)       : 1342177.25
        memcpy to mapped ptr          : 17.48

    Kernel launch latency : 2913.32 us