| 1 |
|
| 2 |
Platform: NVIDIA CUDA |
| 3 |
Device: Quadro P620 |
| 4 |
Driver version : 455.45.01 (Linux x64) |
| 5 |
Compute units : 4 |
| 6 |
Clock frequency : 1442 MHz |
| 7 |
|
| 8 |
Global memory bandwidth (GBPS) |
| 9 |
float : 79.28 |
| 10 |
float2 : 82.40 |
| 11 |
float4 : 84.20 |
| 12 |
float8 : 82.78 |
| 13 |
float16 : 55.16 |
| 14 |
|
| 15 |
Single-precision compute (GFLOPS) |
| 16 |
float : 1357.53 |
| 17 |
float2 : 1400.77 |
| 18 |
float4 : 1397.70 |
| 19 |
float8 : 1390.10 |
| 20 |
float16 : 1385.35 |
| 21 |
|
| 22 |
No half precision support! Skipped |
| 23 |
|
| 24 |
Double-precision compute (GFLOPS) |
| 25 |
double : 45.19 |
| 26 |
double2 : 45.03 |
| 27 |
double4 : 44.97 |
| 28 |
double8 : 44.82 |
| 29 |
double16 : 44.36 |
| 30 |
|
| 31 |
Integer compute (GIOPS) |
| 32 |
int : 473.37 |
| 33 |
int2 : 472.65 |
| 34 |
int4 : 473.04 |
| 35 |
int8 : 466.72 |
| 36 |
int16 : 458.70 |
| 37 |
|
| 38 |
Integer compute Fast 24bit (GIOPS) |
| 39 |
int : 473.50 |
| 40 |
int2 : 473.80 |
| 41 |
int4 : 473.05 |
| 42 |
int8 : 470.43 |
| 43 |
int16 : 468.49 |
| 44 |
|
| 45 |
Transfer bandwidth (GBPS) |
| 46 |
enqueueWriteBuffer : 11.44 |
| 47 |
enqueueReadBuffer : 10.75 |
| 48 |
enqueueWriteBuffer non-blocking : 11.12 |
| 49 |
enqueueReadBuffer non-blocking : 10.50 |
| 50 |
enqueueMapBuffer(for read) : 11.75 |
| 51 |
memcpy from mapped ptr : 15.16 |
| 52 |
enqueueUnmap(after write) : 12.85 |
| 53 |
memcpy to mapped ptr : 15.27 |
| 54 |
|
| 55 |
Kernel launch latency : 3.50 us |
| 56 |
|