Table 3:

Execution time and speed-up of parSMURFⁿ with synth_1 and synth_2 datasets.

Aggregated threads	synth_1 time (s)	synth_1 speed-up	synth_2 time (s)	synth_2 speed-up
1	3,768.59		9,981.82
2	1,910.19	1.97	5,020.18	1.99
4	571.56	3.88	2,539.10	3.93
8	542.35	6.95	1,329.74	7.51
16	288.82	13.05	788.31	12.66
32	248.84	15.18	686.41	14.54

Threads are counted as “aggregated” in the sense that they are the product of the number of MPI processes with the number of threads for each process. All executions have been performed with a 10-fold cross-validation setting.