Platform: AMD Accelerated Parallel Processing
  Device: gfx1036
    Driver version  : 3635.0 (HSA1.1,LC) (Linux x64)
    Compute units   : 1
    Clock frequency : 2200 MHz

    Global memory bandwidth (GBPS)
      float   : 38.45
      float2  : 47.42
      float4  : 49.18
      float8  : 43.41
      float16 : 35.93

    Single-precision compute (GFLOPS)
      float   : 520.20
      float2  : 518.37
      float4  : 514.24
      float8  : 505.54
      float16 : 499.48

    Half-precision compute (GFLOPS)
      half   : 511.37
      half2  : 1050.10
      half4  : 1043.48
      half8  : 1031.78
      half16 : 1024.55

    Double-precision compute (GFLOPS)
      double   : 34.50
      double2  : 34.27
      double4  : 34.26
      double8  : 33.91
      double16 : 33.95

    Integer compute (GIOPS)
      int   : 107.25
      int2  : 108.30
      int4  : 107.50
      int8  : 107.25
      int16 : 107.89

    Integer compute Fast 24bit (GIOPS)
      int   : 520.75
      int2  : 518.55
      int4  : 521.63
      int8  : 516.54
      int16 : 467.75

    Integer char (8bit) compute (GIOPS)
      char   : 521.00
      char2  : 287.65
      char4  : 278.02
      char8  : 263.67
      char16 : 253.14

    Integer short (16bit) compute (GIOPS)
      short   : 519.08
      short2  : 523.19
      short4  : 522.27
      short8  : 516.10
      short16 : 506.43

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 18.40
      enqueueReadBuffer               : 4.62
      enqueueWriteBuffer non-blocking : 18.51
      enqueueReadBuffer non-blocking  : 4.50
      enqueueMapBuffer(for read)      : 766958.44
        memcpy from mapped ptr        : 4.52
      enqueueUnmap(after write)       : 1867377.12
        memcpy to mapped ptr          : 18.38

    Kernel launch latency : 6.43 us