| 1 |
|
| 2 |
Platform: NVIDIA CUDA |
| 3 |
Device: GeForce GTX 960 |
| 4 |
Driver version : 355.11 (Linux x64) |
| 5 |
Compute units : 8 |
| 6 |
Clock frequency : 1329 MHz |
| 7 |
|
| 8 |
Global memory bandwidth (GBPS) |
| 9 |
float : 82.67 |
| 10 |
float2 : 85.63 |
| 11 |
float4 : 87.22 |
| 12 |
float8 : 81.16 |
| 13 |
float16 : 83.39 |
| 14 |
|
| 15 |
Single-precision compute (GFLOPS) |
| 16 |
float : 2550.71 |
| 17 |
float2 : 2747.97 |
| 18 |
float4 : 2793.35 |
| 19 |
float8 : 2728.88 |
| 20 |
float16 : 2760.22 |
| 21 |
|
| 22 |
Double-precision compute (GFLOPS) |
| 23 |
double : 89.67 |
| 24 |
double2 : 89.63 |
| 25 |
double4 : 89.46 |
| 26 |
double8 : 89.10 |
| 27 |
double16 : 88.42 |
| 28 |
|
| 29 |
Integer compute (GIOPS) |
| 30 |
int : 761.99 |
| 31 |
int2 : 803.24 |
| 32 |
int4 : 816.24 |
| 33 |
int8 : 815.58 |
| 34 |
int16 : 826.16 |
| 35 |
|
| 36 |
Transfer bandwidth (GBPS) |
| 37 |
enqueueWriteBuffer : 6.58 |
| 38 |
enqueueReadBuffer : 6.56 |
| 39 |
enqueueMapBuffer(for read) : 6.27 |
| 40 |
memcpy from mapped ptr : 7.07 |
| 41 |
enqueueUnmap(after write) : 6.76 |
| 42 |
memcpy to mapped ptr : 7.12 |
| 43 |
|
| 44 |
Kernel launch latency : 5.16 us |
| 45 |
|