Overview

Optimization with many design variables requires exploring a large design space, leading to increased computational cost. In these scenarios, leveraging distributed processing on multi-core systems is highly effective for cutting down calculation time.

Fig.1 Distribution efficiency for parameter optimization tasks
The chart shows CPU efficiency by number of simultaneous jobs, with error bars representing the range of each data point. Efficiency (unit: %) is defined as\( (T_{non-dist}/T_{dist})/N_{siml}\times100\), where \(T_{non-dist}\) is the non-distributed calculation time using one core under load, \(T_{dist}\) is the distributed calculation time using multiple cores, and \(N_{ siml }\) is the number of simultaneous jobs. Efficiency averages 85% at 100 jobs and 61% at 2,000 jobs.