Closed pancetta closed 7 years ago
ParallelAccelerator uses OpenMP as the parallel backend. You can change the number of threads using OMP_NUM_THREADS
environment variable. Changing it for a specific region of the function is not provided.
Thanks for your answer! Are there any plans to include distributed memory parallelism?
Not for ParallelAccelerator. HPAT provides distributed-memory parallelism for data analytics workloads.
I'm going through and cleaning up some old open issues. It looks like this was sufficiently answered, but if not, feel free to re-open. Thanks!
Is there a way to actually control the degree of parallelism? In particular, how do I control and/or change the number of processes used for e.g.
runStencil
?