Rather than assigning all jobs in a single chunk to each core this assigns 20 chunks per core used. This means that as chunks finish they can get another one from the central pool. This takes some more set up time but should mean all cores run for the full run time.
Rather than assigning all jobs in a single chunk to each core this assigns 20 chunks per core used. This means that as chunks finish they can get another one from the central pool. This takes some more set up time but should mean all cores run for the full run time.