Platform: NVIDIA CUDA
  Device: GeForce GTX 480
    Driver version  : 352.41 (Linux x64)
    Compute units   : 15
    Clock frequency : 1550 MHz

    Global memory bandwidth (GBPS)
      float   : 155.74
      float2  : 157.30
      float4  : 159.18
      float8  : 128.67
      float16 : 71.69

    Single-precision compute (GFLOPS)
      float   : 1475.08
      float2  : 1472.89
      float4  : 1471.08
      float8  : 1463.67
      float16 : 1450.65

    Double-precision compute (GFLOPS)
      double   : 185.74
      double2  : 185.88
      double4  : 185.43
      double8  : 184.61
      double16 : 183.21

    Integer compute (GIOPS)
      int   : 742.88
      int2  : 743.32
      int4  : 743.42
      int8  : 743.26
      int16 : 743.20

    Transfer bandwidth (GBPS)
      enqueueWriteBuffer         : 0.61
      enqueueReadBuffer          : 0.58
      enqueueMapBuffer(for read) : 1.67
        memcpy from mapped ptr   : 0.56
      enqueueUnmap(after write)  : 2.40
        memcpy to mapped ptr     : 0.56

    Kernel launch latency : 7.56 us