Matrix Factorization on Netflix (test)
3.3Training Time (s)CuMF_SGD-P
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CuMF_SGD-PPlatform=Pascal (Tesla P100 GPU)2016.10 | 3.3 | 7 | |
| CuMF_SGD-MPlatform=Maxwell (TITAN X GPU)2016.10 | 7.5 | 3.1 | |
| NOMADPlatform=32-node HPC cluster2016.10 | 9.6 | 2.4 | |
| LIBMFPlatform=Maxwell (CPU, 40 threads)2016.10 | 23 | — |