[MPICH] MPI_REDUCE with MPI_IN_PLACE fails with memory error

[Ashley Pittman]

On Wed, 2007-03-14 at 12:27 +0100, Martin Kleinschmidt wrote:
> but - as said in some other post - we are very often running at the
> limit of the available memory. So I would have to break it down to
> really small pieces in a loop - say 1MB blocks to be on the safe side.
> Wouldn't that lead to a remarkable decrease in throughput?
> For my current problem, this is not an issue, because this point in the
> calculation is reached every 4 hours of calculating on these 600 MB,
> there could be other scenarios...

In our implementation we split it into pipelined 64k chunks and see a
improved throughput by doing so, at least in part due to cacheing
issues, remember that the data has to be processed when it arrives, I'd
pick a chunk size that fits in cache and gives you close to asymptotic
bandwidth for your network.

Ashley,