Skip to main content
. 2018 Nov 13;12:102. doi: 10.3389/fncir.2018.00102

Figure 9.

Figure 9

Runtime of metric computation at different levels of parallelization. The line represents the optimal speedup for increasing the number of cores from the baseline 64 core implementation. The non-parallelized part of the framework represents the computation performed solely on the driver node and is indicated by the bottom two sections of each bar. This non-parallel time is around 129 and 187 s from the mushroom body and medulla respectively.