Parallel bug related to domain decomposition
I'm wondering if anyone has noticed varying results related to domain decomposition. For example, my results are not equal if I decompose by (4 1 1) or (1 4 1). The difference is small, yet still it should not be there considering this is practically the same case.
There is a bug report with a similar observation (link), but I'm curious to hear more experiences.
|All times are GMT -4. The time now is 14:22.|