| 1 |
|
| 2 |
Platform: NVIDIA CUDA |
| 3 |
Device: GeForce GT 630 |
| 4 |
Driver version : 337.88 (Win32) |
| 5 |
Compute units : 2 |
| 6 |
Clock frequency : 1620 MHz |
| 7 |
|
| 8 |
Global memory bandwidth (GBPS) |
| 9 |
float : 17.03 |
| 10 |
float2 : 17.43 |
| 11 |
float4 : 17.66 |
| 12 |
float8 : 8.83 |
| 13 |
float16 : 4.55 |
| 14 |
|
| 15 |
Single-precision compute (GFLOPS) |
| 16 |
float : 206.15 |
| 17 |
float2 : 308.23 |
| 18 |
float4 : 307.50 |
| 19 |
float8 : 294.83 |
| 20 |
float16 : 279.48 |
| 21 |
|
| 22 |
Double-precision compute (GFLOPS) |
| 23 |
double : 25.89 |
| 24 |
double2 : 25.87 |
| 25 |
double4 : 25.81 |
| 26 |
double8 : 25.72 |
| 27 |
double16 : 25.52 |
| 28 |
|
| 29 |
Integer compute (GIOPS) |
| 30 |
int : 103.40 |
| 31 |
int2 : 103.49 |
| 32 |
int4 : 103.40 |
| 33 |
int8 : 103.40 |
| 34 |
int16 : 103.40 |
| 35 |
|
| 36 |
Transfer bandwidth (GBPS) |
| 37 |
enqueueWriteBuffer : 3.11 |
| 38 |
enqueueReadBuffer : 2.99 |
| 39 |
enqueueMapBuffer(for read) : 2.91 |
| 40 |
memcpy from mapped ptr : 3.75 |
| 41 |
enqueueUnmap(after write) : 4.17 |
| 42 |
memcpy to mapped ptr : 3.78 |
| 43 |
|
| 44 |
Kernel launch latency : 23.75 us |
| 45 |
|
| 46 |
|