Closed HiroIshida closed 1 year ago
use cached one, because pool generation cannot be parallelizable of it's trial-and-error nature.
bottleneck is solve() rather than create mesh, when using distributed. So just use normal problem pool.
single problem pool
use cached one, because pool generation cannot be parallelizable of it's trial-and-error nature.
multiple pool
bottleneck is solve() rather than create mesh, when using distributed. So just use normal problem pool.