Research Article
GPU-Accelerated Parallel FDTD on Distributed Heterogeneous Platform
Table 1
The performances before and after optimization.
| | Loop time of E5-2670 single core (s) | Loop time of K20m single card (s) |
| Before optimization | 2.2740 | 0.1320 | After optimization | 1.3587 | 0.1021 | Performance improvement | 40% | 23% |
|
|