Table 5.
Timing and speedup for 4 × 4 ROIs with single precision.
| Configurations | P100 | vs | Xeon E5-2620 | TITAN V | vs | Xeon E5-2620 | units |
|---|---|---|---|---|---|---|---|
| T1 | T2 | PD | T1 | T2 | PD | ||
| CPU time | 70.7±0.5 | 70.3±0.4 | 71.9±2.3 | 70.7±0.5 | 70.3±0.4 | 71.9±2.3 | ms |
| GPU time | 56.4±18.3 | 50.7±11.5 | 38.2±0.5 | 37.1±2.0 | 36.9±2.7 | 37.6±2.9 | ms |
| GLRLM Speedup | 1.3±0.4 | 1.4±0.3 | 1.9±0.1 | 1.9±0.1 | 1.5±0.3 | 1.9±0.2 | x |
| CPU time | 209.1±0.4 | 209.3±1.2 | 212.9±6.9 | 209.1±0.4 | 209.3±1.2 | 212.9±6.9 | ms |
| GPU time | 96.3±24.1 | 96.4±27.2 | 60.1±0.9 | 57.4±3.6 | 54.1±2.6 | 52.1±4.2 | ms |
| 0° feature Speedup | 2.2±0.6 | 2.2±0.7 | 3.6±0.1 | 3.7±0.2 | 2.4±0.7 | 4.1±0.3 | x |
| CPU time | 209.6±0.5 | 210.1±1.3 | 214.3±6.9 | 209.6±0.5 | 210.1±1.3 | 214.3±6.9 | ms |
| GPU time | 37.1±12.6 | 37.1±12.3 | 19.3±0.4 | 28.5±2.7 | 28.2±2.6 | 20.8±3.2 | ms |
| 45° feature Speedup | 5.7±1.7 | 5.7±1.7 | 11.1±0.4 | 7.4±0.7 | 7.5±0.6 | 10.5±1.3 | x |
| CPU time | 208.9±0.4 | 209.1±1.2 | 212.8±6.8 | 208.9±0.4 | 209.1±1.2 | 212.8±6.8 | ms |
| GPU time | 21.3±7.9 | 21.8±10.3 | 15.9±0.3 | 17.2±1.5 | 26.6±1.5 | 15.7±1.5 | ms |
| 90° feature Speedup | 9.8±2.4 | 9.6±3.3 | 13.4±0.6 | 12.3±1.1 | 13.1±1.1 | 13.7±1.3 | x |
| CPU time | 208.5±0.4 | 209.1±1.2 | 212.4±6.8 | 208.5±0.4 | 209.1±1.2 | 212.4±6.8 | ms |
| GPU time | 41.2±14.4 | 34.0±8.9 | 18.7±0.4 | 25.6±0.9 | 26.6±0.5 | 18.5±1.3 | ms |
| 135° feature Speedup | 5.1±1.9 | 6.1±1.4 | 11.4±0.5 | 8.2±0.3 | 7.9±0.2 | 11.7±0.8 | x |