Skip to main content
. Author manuscript; available in PMC: 2017 Nov 30.
Published in final edited form as: IBM J Res Dev. 2017 May 23;61(2-3):6.1–6.14. doi: 10.1147/JRD.2017.2656758

Table 2.

Speedup achieved by parallelization of the model and evolutionary optimization process for a population of 60 individuals (6 islands).

Description Cores required (network + arm) Time/generation (minutes) Speedup
Purely sequential 1 + 1 2,945.2 1
Parallel simulation (sequential individuals + islands) 16 + 1 260.6 11.3
Parallel simulation + individuals (sequential islands) 160 + 10 44.9 11.3 × 5.8 = 65.6
Parallel simulation + individuals + islands 960 + 60 7.5 11.3 × 5.8 × 6.0 = 393.2
Parallel simulation + individuals (single population, no islands) 960 + 60 13.0 11.3 × 20.0 = 226.6