Table 1 Factors at which parallel runtime was sped up using blocking. Measurements for all three datasets are included and the factor is calculated from the measured runtime of a default HarmonizR run divided by the measured runtimes of runs including blocking with block = 2 and block = 4, respectively (the latter are also shown in seconds). Computed on an Intel Xeon Gold 6226, 2.70 GHz, 2 × 12 compute cores, 96 GB RAM
Krug et al. 2020 | Petralia et al. 2020 | Nusinow et al. 2020 | |
|---|---|---|---|
Ratio: Default to block = 2 (1 core/sequential) | Runtime: 12 s Factor: 21.92 | Runtime: 61 s Factor: 7.44 | Runtime: 277 s Factor: 2.14 |
Ratio: Default to block = 2 (4 cores) | Runtime: 7 s Factor: 10.29 | Runtime: 19 s Factor: 6.68 | Runtime: 72 s Factor: 1.96 |
Ratio: Default to block = 2 (12 cores) | Runtime: 6 s Factor: 4.67 | Runtime: 10 s Factor: 6.80 | Runtime: 49 s Factor: 1.59 |
Ratio: Default to block = 4 (1 core/sequential) | Runtime: 5 s Factor: 52.60 | Runtime: 7 s Factor: 64.86 | Runtime: 70 s Factor: 6.93 |
Ratio: Default to block = 4 (4 cores) | Runtime: 4 s Factor: 18.00 | Runtime: 5 s Factor: 25.40 | Runtime: 31 s Factor: 4.55 |
Ratio: Default to block = 4 (12 cores) | Runtime: 3 s Factor: 9.34 | Runtime: 5 s Factor: 13.60 | Runtime: 23 s Factor: 3.39 |