Platform: NVIDIA CUDA
  Device: Tesla V100-PCIE-32GB
    Driver version  : 455.23.05 (Linux x64)
    Compute units   : 80
    Clock frequency : 1380 MHz

    Global memory bandwidth (GBPS)
      float   : 716.38
      float2  : 765.67
      float4  : 810.35
      float8  : 723.85
      float16 : 750.17

    Single-precision compute (GFLOPS)
      float   : 14098.15
      float2  : 14135.97
      float4  : 14095.57
      float8  : 14049.00
      float16 : 13934.45

    No half precision support! Skipped

    Double-precision compute (GFLOPS)
      double   : 7075.81
      double2  : 7065.56
      double4  : 7046.01
      double8  : 7013.68
      double16 : 6951.51

    Integer compute (GIOPS)
      int   : 14069.94
      int2  : 14118.04
      int4  : 14121.60
      int8  : 14124.16
      int16 : 14099.04

    Integer compute Fast 24bit (GIOPS)
      int   : 14077.32
      int2  : 14119.12
      int4  : 14122.14
      int8  : 14113.63
      int16 : 14104.60

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 12.06
      enqueueReadBuffer               : 10.64
      enqueueWriteBuffer non-blocking : 10.72
      enqueueReadBuffer non-blocking  : 8.13
      enqueueMapBuffer(for read)      : 10.25
        memcpy from mapped ptr        : 17.55
      enqueueUnmap(after write)       : 12.59
        memcpy to mapped ptr          : 18.20

    Kernel launch latency : 7.88 us