Rob Latham <[email protected]> writes:

ON blue gene... well you maybe can guess where I'm going with this. MPICH will use quite a bit of memory to describe 16 million points: about 200 MiB (4 bytes for each index, 8 bytes for each offset).

there isn't 200 MiB of extra memory on some configurations of Blue Gene.

What options does one have when trying to efficiently describe unstructured data, though?

How large is the data that this unstructured representation is describing? Usually there will be element connectivity and several elements are adjacent to each vertex, so the relative overhead of the MPICH datatypes may not be high. The index/offset arrays can probably be compressed, though that only reduces peak memory usage if you specify and free the uncompressed arrays before other allocations.