Platform: Portable Computing Language
  Device: NVIDIA Tegra X2
    Driver version  : 3.0-rc2 (Linux ARM64)
    Compute units   : 2
    Clock frequency : 1020 MHz

    Global memory bandwidth (GBPS)
      float   : 37.31
      float2  : 46.28
      float4  : 45.69
      float8  : 34.27
      float16 : 28.97

    Single-precision compute (GFLOPS)
      float   : 592.11
      float2  : 649.11
      float4  : 647.81
      float8  : 651.77
      float16 : 649.53

    No half precision support! Skipped

    Double-precision compute (GFLOPS)
      double   : 20.79
      double2  : 20.74
      double4  : 20.71
      double8  : 20.64
      double16 : 20.48

    Integer compute (GIOPS)
      int   : 217.46
      int2  : 219.72
      int4  : 218.27
      int8  : 216.66
      int16 : 215.70

    Integer compute Fast 24bit (GIOPS)
      int   : 217.46
      int2  : 219.69
      int4  : 218.28
      int8  : 216.60
      int16 : 215.16

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 5.79
      enqueueReadBuffer               : 0.31
      enqueueWriteBuffer non-blocking : 5.55
      enqueueReadBuffer non-blocking  : 0.31
      enqueueMapBuffer(for read)      : 14717.57
        memcpy from mapped ptr        : 0.31
      enqueueUnmap(after write)       : 0.53
        memcpy to mapped ptr          : 2.86

    Kernel launch latency : -947.51 us