Runtime and memory usage comparison between serial and parallel multi-MUM computation across different partition schemes
| Serial | Parallel | |||||
|---|---|---|---|---|---|---|
| Data set | Partitions | Seqs per | Time (h) | Memory (GB) | Time (h) | Memory (GB) |
| Chr 19 HPRC (N = 474) | 1 | 474 | 4.77 | 47.30 | – | – |
| 5 | 96 | 5.00 | 14.99 | 1.11 | 71.71 | |
| 10 | 48 | 5.91 | 10.27 | 0.63 | 95.16 | |
| 20 | 24 | 7.21 | 7.69 | 0.40 | 134.53 | |
| 40 | 12 | 8.32 | 5.74 | 0.26 | 204.43 | |
| A. thaliana (N = 69) | 1 | 69 | 2.58 | 75.53 | – | – |
| 5 | 15 | 3.75 | 29.15 | 0.80 | 137.52 | |
| 10 | 7 | 4.92 | 20.18 | 0.57 | 184.68 | |
| 20 | 4 | 6.48 | 14.58 | 0.40 | 260.23 | |
[i] We report execution time (in hours) and peak memory usage (in GB) for two data sets, Chr 19 haplotypes from the Human Pangenome Reference Consortium (HPRC) (Liao et al. 2023) and the A. thaliana pangenome (Lian et al. 2024). The lowest peak memory footprint and the fastest running times are bolded.