<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><br><div><div>On Jan 28, 2012, at 4:25 PM, Jed Brown wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div class="gmail_quote">On Fri, Jan 27, 2012 at 11:29, Mark F. Adams <span dir="ltr"><<a href="mailto:mark.adams@columbia.edu">mark.adams@columbia.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
GAMG does not repartition by default anymore -- it is very expensive. GAMG does now do simple process aggregation on coarser grids if repartitioning is not specified.</blockquote></div><br><div>Couldn't we at least do some cheap (e.g. greedy) repartitioning? </div></blockquote><div><br></div><div>Yes</div><br><blockquote type="cite"><div>Or at least squish out empty ranks so that the coarser levels tend to be nearby on the network? </div></blockquote><div><br></div><div>We do that now.</div><br><blockquote type="cite"><div>Would it make more sense to do this or to work on a "real" partitioner.</div>
</blockquote></div><div><br></div><div>I prefer to fold this into the new aggregation stuff. </div></body></html>