Platform: NVIDIA CUDA
  Device: TITAN V
    Driver version  : 430.14 (Linux x64)
    Compute units   : 80
    Clock frequency : 1455 MHz

    Global memory bandwidth (GBPS)
      float   : 561.37
      float2  : 591.37
      float4  : 607.71
      float8  : 516.49
      float16 : 466.27

    Single-precision compute (GFLOPS)
      float   : 13651.32
      float2  : 13688.23
      float4  : 13648.46
      float8  : 13606.27
      float16 : 13502.08

    No half precision support! Skipped

    Double-precision compute (GFLOPS)
      double   : 6858.92
      double2  : 6846.90
      double4  : 6822.64
      double8  : 6797.12
      double16 : 6737.34

    Integer compute (GIOPS)
      int   : 13622.13
      int2  : 13661.56
      int4  : 13666.12
      int8  : 13663.23
      int16 : 13640.81

    Integer compute Fast 24bit (GIOPS)
      int   : 13622.35
      int2  : 13662.14
      int4  : 13666.63
      int8  : 13658.38
      int16 : 13647.09

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 6.09
      enqueueReadBuffer               : 6.45
      enqueueWriteBuffer non-blocking : 4.58
      enqueueReadBuffer non-blocking  : 4.93
      enqueueMapBuffer(for read)      : 6.05
        memcpy from mapped ptr        : 9.09
      enqueueUnmap(after write)       : 6.26
        memcpy to mapped ptr          : 9.42

    Kernel launch latency : 6.51 us