Platform: Portable Computing Language
  Device: Orin
    Driver version  : 5.0 (Linux ARM64)
    Compute units   : 8
    Clock frequency : 624 MHz

    Global memory bandwidth (GBPS)
      float   : 62.87
      float2  : 63.71
      float4  : 63.71
      float8  : 63.68
      float16 : 40.08

    Single-precision compute (GFLOPS)
      float   : 1245.91
      float2  : 1237.44
      float4  : 1217.39
      float8  : 1254.84
      float16 : 1245.56

    Half-precision compute (GFLOPS)
      half   : 635.04
      half2  : 2397.03
      half4  : 1801.24
      half8  : 1961.29
      half16 : 1727.49

    Double-precision compute (GFLOPS)
      double   : 19.97
      double2  : 19.95
      double4  : 19.91
      double8  : 19.81
      double16 : 19.61

    Integer compute (GIOPS)
      int   : 425.48
      int2  : 427.67
      int4  : 437.13
      int8  : 426.58
      int16 : 428.96

    Integer compute Fast 24bit (GIOPS)
      int   : 425.48
      int2  : 427.67
      int4  : 437.13
      int8  : 426.58
      int16 : 429.16

    Integer char (8bit) compute (GIOPS)
      char   : 210.67
      char2  : 280.69
      char4  : 323.71
      char8  : 359.63
      char16 : 337.26

    Integer short (16bit) compute (GIOPS)
      short   : 210.61
      short2  : 273.66
      short4  : 322.43
      short8  : 351.30
      short16 : 336.00

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 6.67
      enqueueReadBuffer               : 6.65
      enqueueWriteBuffer non-blocking : 6.65
      enqueueReadBuffer non-blocking  : 6.65
      enqueueMapBuffer(for read)      : 47989.21
        memcpy from mapped ptr        : 6.70
      enqueueUnmap(after write)       : 8.04
        memcpy to mapped ptr          : 6.70

    Kernel launch latency : -530.93 us