Processes are currently forked as soon as an opportunity for parallelization is identified. When the process requires more memory than what is available, the additional paging causes a slowdown. The scheduler should wait until previous processes have completed before forking the next.
Processes are currently forked as soon as an opportunity for parallelization is identified. When the process requires more memory than what is available, the additional paging causes a slowdown. The scheduler should wait until previous processes have completed before forking the next.