Ceci est une ancienne révision du document !
Bnechmarks GPU
Matrix mulitply provided with CUDA SDK
Using square matruix (NxN)
| N | K20 time_ms (GFlop/s) | k40 time_ms (GFlop/s) |
|---|---|---|
| 1000 | 8.349 ( 239.55) | x |
| 2000 | 64.629 (247.57) | x |
| 4000 | 491.310 (260.53) | x |
CUBLAS
Peak performance
| k20 | k40 |
|---|---|
| 1210 Glops/s | x |