fixedSchedule {mspath} | R Documentation |
Return indices to cut input work into roughly equal chunks
fixedSchedule(work, ngroups)
work |
vector of work effort, sorted in descending order |
ngroups |
desired number of work chunks.
If some of the individual work items are big, you may
get less than ngroups back. |
Does simple fixed scheduling, dividing up cases into batches with roughly equal work effort. Dynamic scheduling would adjust to the run-time conditions; since we have uniform slave processors and expect run-time to be fairly repeatable, fixed scheduling should suffice.
However, this may initially be used with a proxy of actual run time in
work
; in that case, it may help to collect the actual run-times
and repartition the work.
The indices in work to divide it up. These are the right-hand endpoints, so [1] is the index of the last work item to include in the first group.
Ross Boylan