[Beowulf] New HPCC results and the Myri viewpoint

[Greg Lindahl]

On Thu, Jul 21, 2005 at 01:07:46PM -0600, Keith D. Underwood wrote:

> I don't know of anything that
> does a tight loop of "small amount of work, ping-pong with one other
> processor, repeat".

Right, it's typically "small amount of work, ping-pong with N other
processors, repeat", where N=2, 4, 8, 9, 27... Solidly in the middle
between ping-pong with 1 node and streaming. This describes pretty
much all time-explicit finite-difference codes decomposed in 1-3
dimensions.

-- greg