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   : 707.45
      float2  : 750.59
      float4  : 756.57
      float8  : 811.21
      float16 : 856.90

    Single-precision compute (GFLOPS)
      float   : 30893.80
      float2  : 33078.14
      float4  : 32527.02
      float8  : 33777.64
      float16 : 28855.14

    Half-precision compute (GFLOPS)
      half   : 31073.23
      half2  : 64274.54
      half4  : 63923.77
      half8  : 63553.30
      half16 : 63412.55

    Double-precision compute (GFLOPS)
      double   : 1141.44
      double2  : 1137.77
      double4  : 1134.46
      double8  : 1127.09
      double16 : 1119.44

    Integer compute (GIOPS)
      int   : 6710.68
      int2  : 6667.33
      int4  : 6689.00
      int8  : 6660.87
      int16 : 6622.93

    Integer compute Fast 24bit (GIOPS)
      int   : 29006.98
      int2  : 28720.87
      int4  : 28630.46
      int8  : 28535.36
      int16 : 24233.41

    Integer char (8bit) compute (GIOPS)
      char   : 31074.16
      char2  : 16723.87
      char4  : 16627.11
      char8  : 14027.69
      char16 : 13643.48

    Integer short (16bit) compute (GIOPS)
      short   : 31083.53
      short2  : 29873.99
      short4  : 29376.22
      short8  : 28949.41
      short16 : 28414.74

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 17.18
      enqueueReadBuffer               : 4.65
      enqueueWriteBuffer non-blocking : 17.06
      enqueueReadBuffer non-blocking  : 4.61
      enqueueMapBuffer(for read)      : 254140.08
        memcpy from mapped ptr        : 4.54
      enqueueUnmap(after write)       : 338186.41
        memcpy to mapped ptr          : 16.28

    Kernel launch latency : 1545461120.00 us