Platform: NVIDIA CUDA
  Device: Quadro GV100
    Driver version  : 455.23.05 (Linux x64)
    Compute units   : 80
    Clock frequency : 1627 MHz

    Global memory bandwidth (GBPS)
      float   : 554.70
      float2  : 575.69
      float4  : 258.14
      float8  : 537.39
      float16 : 552.97

    Single-precision compute (GFLOPS)
      float   : 6216.30
      float2  : 11449.58
      float4  : 14290.00
      float8  : 7261.11
      float16 : 7262.70

    No half precision support! Skipped

    Double-precision compute (GFLOPS)
      double   : 7212.01
      double2  : 7191.88
      double4  : 7168.91
      double8  : 4505.54
      double16 : 3124.18

    Integer compute (GIOPS)
      int   : 6219.71
      int2  : 9340.54
      int4  : 14371.73
      int8  : 14373.41
      int16 : 14342.93

    Integer compute Fast 24bit (GIOPS)
      int   : 9037.88
      int2  : 6214.77
      int4  : 6216.47
      int8  : 9337.87
      int16 : 14352.44

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 9.75
      enqueueReadBuffer               : 12.11
      enqueueWriteBuffer non-blocking : 9.32
      enqueueReadBuffer non-blocking  : 11.31
      enqueueMapBuffer(for read)      : 11.25
        memcpy from mapped ptr        : 9.54
      enqueueUnmap(after write)       : 11.26
        memcpy to mapped ptr          : 9.84

    Kernel launch latency : 19.29 us