Platform: Portable Computing Language
  Device: Orin
    Driver version  : 5.0 (Linux ARM64)
    Compute units   : 8
    Clock frequency : 918 MHz

    Global memory bandwidth (GBPS)
      float   : 86.53
      float2  : 93.67
      float4  : 93.68
      float8  : 93.64
      float16 : 58.93

    Single-precision compute (GFLOPS)
      float   : 1830.74
      float2  : 1818.19
      float4  : 1788.92
      float8  : 1843.85
      float16 : 1830.18

    Half-precision compute (GFLOPS)
      half   : 933.07
      half2  : 3520.95
      half4  : 2646.06
      half8  : 2881.55
      half16 : 2537.46

    Double-precision compute (GFLOPS)
      double   : 29.35
      double2  : 29.32
      double4  : 29.25
      double8  : 29.11
      double16 : 28.82

    Integer compute (GIOPS)
      int   : 625.10
      int2  : 628.33
      int4  : 642.09
      int8  : 626.74
      int16 : 630.25

    Integer compute Fast 24bit (GIOPS)
      int   : 625.10
      int2  : 628.37
      int4  : 642.10
      int8  : 626.74
      int16 : 630.53

    Integer char (8bit) compute (GIOPS)
      char   : 309.53
      char2  : 412.41
      char4  : 475.61
      char8  : 528.38
      char16 : 495.53

    Integer short (16bit) compute (GIOPS)
      short   : 309.46
      short2  : 402.09
      short4  : 473.74
      short8  : 516.14
      short16 : 493.67

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 7.03
      enqueueReadBuffer               : 7.02
      enqueueWriteBuffer non-blocking : 7.02
      enqueueReadBuffer non-blocking  : 7.02
      enqueueMapBuffer(for read)      : 117609.66
        memcpy from mapped ptr        : 7.04
      enqueueUnmap(after write)       : 8.41
        memcpy to mapped ptr          : 7.05

    Kernel launch latency : -408.96 us