Platform: AMD Accelerated Parallel Processing
  Device: Ellesmere (RX 570)
    Driver version  : 2639.3 (Linux x64)
    Compute units   : 32
    Clock frequency : 1268 MHz

    Global memory bandwidth (GBPS)
      float   : 182.67
      float2  : 191.18
      float4  : 193.05
      float8  : 163.17
      float16 : 139.53

    Single-precision compute (GFLOPS)
      float   : 5133.96
      float2  : 5129.58
      float4  : 5109.60
      float8  : 5087.33
      float16 : 5018.78

    Half-precision compute (GFLOPS)
      half   : 5119.05
      half2  : 5115.60
      half4  : 5104.28
      half8  : 5089.02
      half16 : 5045.35

    Double-precision compute (GFLOPS)
      double   : 323.55
      double2  : 323.39
      double4  : 322.84
      double8  : 321.66
      double16 : 320.53

    Integer compute (GIOPS)
      int   : 1032.21
      int2  : 1032.00
      int4  : 1031.56
      int8  : 1030.70
      int16 : 1029.31

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer         : 38.47
      enqueueReadBuffer          : 14.87
      enqueueMapBuffer(for read) : 220429.70
        memcpy from mapped ptr   : 15.33
      enqueueUnmap(after write)  : 604140.69
        memcpy to mapped ptr     : 15.09

    Kernel launch latency : 42.83 us