Platform: NVIDIA CUDA Device: GeForce GTX TITAN Driver version : 352.21 (Linux x64) Compute units : 14 Clock frequency : 928 MHz Global memory bandwidth (GBPS) float : 227.55 float2 : 235.48 float4 : 244.46 float8 : 203.42 float16 : 150.25 Single-precision compute (GFLOPS) float : 3205.36 float2 : 4043.81 float4 : 4002.08 float8 : 3948.40 float16 : 3705.41 Double-precision compute (GFLOPS) double : 1629.84 double2 : 1628.86 double4 : 1625.99 double8 : 1619.69 double16 : 1606.94 Integer compute (GIOPS) int : 815.49 int2 : 814.82 int4 : 814.34 int8 : 812.88 int16 : 814.81 Transfer bandwidth (GBPS) enqueueWriteBuffer : 4.08 enqueueReadBuffer : 3.71 enqueueMapBuffer(for read) : 6.04 memcpy from mapped ptr : 6.47 enqueueUnmap(after write) : 6.35 memcpy to mapped ptr : 6.45 Kernel launch latency : 6.58 us