Fig. 9
From: Fast noisy long read alignment with multi-level parallelism
The speedup and parallel efficiency of ParaHAT. a Speedup ratios of ParaHAT on real and simulated datasets. The larger the dataset, the more linear the growth in speedup, peaking at 94.61x on 128 nodes. b Parallel efficiency of ParaHAT. The larger the dataset, the slower the reduction in parallel efficiency