Table 4.
Speedup and efficiency of parallelized SMILES parsing
| Number of Threads | Time [ms] | Speedup | Efficiency |
|---|---|---|---|
| 1 |
9493 |
1.00 |
1.00 |
| 2 |
5309 |
1.79 |
0.89 |
| 3 |
4134 |
2.30 |
0.77 |
| 4 |
2988 |
3.18 |
0.79 |
| 5 |
2785 |
3.41 |
0.68 |
| 6 |
2647 |
3.59 |
0.60 |
| 7 |
2273 |
4.18 |
0.60 |
| 8 |
2091 |
4.54 |
0.57 |
| 9 |
2037 |
4.66 |
0.52 |
| 10 |
2061 |
4.61 |
0.46 |
| 11 |
2061 |
4.61 |
0.42 |
| 12 | 2096 | 4.53 | 0.38 |
The test runs were performed on a hexacore processor supporting hyperthreading. Again our testing excerpt of the ZINC database was parsed using Scala’s parallel collections.