Platform: AMD Accelerated Parallel Processing
  Device: gfx1103
    Driver version  : 3614.0 (HSA1.1,LC) (Linux x64)
    Compute units   : 6
    Clock frequency : 2700 MHz

    Global memory bandwidth (GBPS)
      float   : 71.93
      float2  : 76.21
      float4  : 79.33
      float8  : 79.91
      float16 : 82.53

    Single-precision compute (GFLOPS)
      float   : 2301.48
      float2  : 2522.03
      float4  : 2519.30
      float8  : 2503.28
      float16 : 2505.05

    Half-precision compute (GFLOPS)
      half   : 2513.05
      half2  : 4574.63
      half4  : 4561.89
      half8  : 4452.38
      half16 : 4418.84

    Double-precision compute (GFLOPS)
      double   : 85.92
      double2  : 85.84
      double4  : 85.59
      double8  : 85.34
      double16 : 84.68

    Integer compute (GIOPS)
      int   : 535.31
      int2  : 535.65
      int4  : 536.31
      int8  : 532.17
      int16 : 531.11

    Integer compute Fast 24bit (GIOPS)
      int   : 2283.13
      int2  : 2319.12
      int4  : 2299.29
      int8  : 2273.81
      int16 : 1905.24

    Integer char (8bit) compute (GIOPS)
      char   : 2296.83
      char2  : 1331.43
      char4  : 1301.48
      char8  : 1167.89
      char16 : 1162.60

    Integer short (16bit) compute (GIOPS)
      short   : 2288.83
      short2  : 2298.11
      short4  : 2290.89
      short8  : 2229.92
      short16 : 2216.77

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 16.81
      enqueueReadBuffer               : 5.23
      enqueueWriteBuffer non-blocking : 16.84
      enqueueReadBuffer non-blocking  : 5.23
      enqueueMapBuffer(for read)      : 49334.05
        memcpy from mapped ptr        : 5.17
      enqueueUnmap(after write)       : 285212.47
        memcpy to mapped ptr          : 16.86

    Kernel launch latency : 8.57 us