Platform: NVIDIA CUDA
  Device: Quadro P620
    Driver version  : 455.45.01 (Linux x64)
    Compute units   : 4
    Clock frequency : 1442 MHz

    Global memory bandwidth (GBPS)
      float   : 79.28
      float2  : 82.40
      float4  : 84.20
      float8  : 82.78
      float16 : 55.16

    Single-precision compute (GFLOPS)
      float   : 1357.53
      float2  : 1400.77
      float4  : 1397.70
      float8  : 1390.10
      float16 : 1385.35

    No half precision support! Skipped

    Double-precision compute (GFLOPS)
      double   : 45.19
      double2  : 45.03
      double4  : 44.97
      double8  : 44.82
      double16 : 44.36

    Integer compute (GIOPS)
      int   : 473.37
      int2  : 472.65
      int4  : 473.04
      int8  : 466.72
      int16 : 458.70

    Integer compute Fast 24bit (GIOPS)
      int   : 473.50
      int2  : 473.80
      int4  : 473.05
      int8  : 470.43
      int16 : 468.49

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer              : 11.44
      enqueueReadBuffer               : 10.75
      enqueueWriteBuffer non-blocking : 11.12
      enqueueReadBuffer non-blocking  : 10.50
      enqueueMapBuffer(for read)      : 11.75
        memcpy from mapped ptr        : 15.16
      enqueueUnmap(after write)       : 12.85
        memcpy to mapped ptr          : 15.27

    Kernel launch latency : 3.50 us