Platform: NVIDIA CUDA
  Device: GeForce RTX 2080 Ti
    Driver version  : 415.27 (Linux x64)
    Compute units   : 68
    Clock frequency : 1650 MHz

    Global memory bandwidth (GBPS)
      float   : 506.69
      float2  : 532.16
      float4  : 548.03
      float8  : 556.57
      float16 : 492.17

    Single-precision compute (GFLOPS)
      float   : 16909.53
      float2  : 16894.22
      float4  : 16866.23
      float8  : 16798.47
      float16 : 16672.67

    No half precision support! Skipped

    Double-precision compute (GFLOPS)
      double   : 529.92
      double2  : 529.30
      double4  : 527.99
      double8  : 525.44
      double16 : 519.98

    Integer compute (GIOPS)
      int   : 15480.90
      int2  : 15398.06
      int4  : 15411.76
      int8  : 15226.44
      int16 : 15304.72

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer         : 9.61
      enqueueReadBuffer          : 8.49
      enqueueMapBuffer(for read) : 10.79
        memcpy from mapped ptr   : 11.37
      enqueueUnmap(after write)  : 12.27
        memcpy to mapped ptr     : 11.84

    Kernel launch latency : 3.81 us