Barry Smith <bsmith at mcs.anl.gov> writes: > Fine, data is order M*N*P, but what about order M*N + N*P + M*P > data? Some of the scatters are this size and hence ok to keep > around?
But they aren't really that size because we don't have an optimized representation for the local part. I wouldn't be at all worried about those scatters if the local part was optimized in this way. Optimizing the scatter representation is more attractive, but the VecScatter code isn't so easy to extend.
