<div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sun, Mar 29, 2020 at 6:20 PM Jed Brown <<a href="mailto:jed@jedbrown.org">jed@jedbrown.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">We are interested in assembling point-block diagonals for use in<br>
pbjacobi smoothing. This is much simpler than AIJ, </blockquote><div><br></div><div>Would the coloring/DD algo on the outside be the same, just with a different assembly kernel?</div><div><br></div><div>Or are you looking a fusing the element construction with assembly? That seems reasonable now that I think about it.</div><div><br></div><div>It's the same coloring, right?</div><div><br></div><div>And, do we have an appropriate coloring method now?</div><div><br></div><div>Thanks,</div><div>Mark</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">but lots of people<br>
will need AIJ so it's certainly worthwhile.<br>
<br>
Mark Adams <<a href="mailto:mfadams@lbl.gov" target="_blank">mfadams@lbl.gov</a>> writes:<br>
<br>
> Is there any plans/interest in assembling AIJ matrices on GPUs?<br>
><br>
> I have a somewhat strangely balanced FE operator and I have each thread<br>
> block create one element matrix, one thread per integration point with an<br>
> expensive computation at each integration point. Then I assemble the<br>
> matrices on the CPU.<br>
><br>
> I want to parallelize assembly on GPU machines, with coloring or DD I<br>
> assume. Does anyone else care about this?<br>
><br>
> Thanks,<br>
> Mark<br>
</blockquote></div></div>