Fig 1.
Schematic representation of sequential algorithms.
Fig 2.
Schematic representation of saCeSS2 algorithm.
Table 1.
Relation between functions B(x1, x2) and .
Table 2.
Case study 1 SSP: Performance analysis from a horizontal view.
Table 3.
Case study 1 SSP: Performance analysis from a vertical perspective.
Fig 3.
Case study 1 SSP: Bean plots of execution time for np-eSS vs saCeSS2 using 10, 20 and 40 MPI processors.
VTR = 10 and 20 independent runs for each experiment.
Fig 4.
Case study 1 SSP: Convergence curves.
Fig 5.
Convergence curves for saCeSS2 using 1, 10, 20 and 40 processors corresponding to the runs in the median values of the results distribution.
Table 4.
Case study 2 HepG2: Performance analysis from a horizontal view.
Table 5.
Case study 2 HePG2: Performance analysis from a vertical view.
Fig 6.
Case study 2 HePG2: Bean plots of execution time for NP-eSS vs saCeSS2 using 10, 20 and 40 MPI processors in the HePG2 problem.
VTR = 33 and the number of runs is equal to 20.
Fig 7.
Case study 2 HePG2: Convergence curves.
Fig 8.
Convergence curves for saCeSS2 using 10, 20 and 40 processors corresponding to the runs in the median values of the results distribution.
Fig 9.
Case study 3: HPN-DREAM convergence curves in the NEMO local cluster.
Convergence curves using 10, 20, 40 and 60 cores.
Fig 10.
Case study 3: HPN-DREAM convergence curves in the EBI cluster.
Convergence curves using 100, and 300 cores.
Table 6.
Performance of saCeSS2 for both SSP and HePG2 case studies in azure public cloud.
Fig 11.
Case study 1: Beanplots comparing results in terms of execution time in azure vs local cluster in the SSP problem.
Fig 12.
Case study 2: Beanplots comparing results in terms of execution time in azure vs local cluster in the HePG2 problem.