The whole partition per one kernel
Hello to everybody. A parallel run for 4 partitions (one partition per each kernel) is started. During the run CFX calls my CEL Fuinction having at input locations corresponding to a given partition. However, sometimes, when the mesh is rather dense, one and the same partition is NOT developed at once, but at least two calls occur for the partition (of course, each time different set of locations). This slows down the overall performance of my model dramatically. Is there any way to make ANSYS call my CEL Function having at the input ALL the locations of a partition at once? Thanks!
Re: The whole partition per one kernel
Remember the multiple calls are to smaller subsets of the partition not multiple calls over the whole partition.
|All times are GMT -4. The time now is 03:59.|