<html><body><div style="color:#000; background-color:#fff; font-family:HelveticaNeue, Helvetica Neue, Helvetica, Arial, Lucida Grande, sans-serif;font-size:12pt"><div><span>I'm trying to use Petsc to solve a system with ksp preconditioned with jacobi. Is this algorithm effivient in a GPU cluster?</span></div><div style="color: rgb(0, 0, 0); font-size: 16px; font-family: HelveticaNeue, 'Helvetica Neue', Helvetica, Arial, 'Lucida Grande', sans-serif; background-color: transparent; font-style: normal;"><span><br><br></span></div><div class="yahoo_quoted" style="display: block;"> <div style="font-family: HelveticaNeue, 'Helvetica Neue', Helvetica, Arial, 'Lucida Grande', sans-serif; font-size: 12pt;"> <div style="font-family: HelveticaNeue, 'Helvetica Neue', Helvetica, Arial, 'Lucida Grande', sans-serif; font-size: 12pt;"> <div dir="ltr"> <font size="2" face="Arial"> Il Marted́ 13 Maggio 2014 17:32, Karl Rupp <rupp@iue.tuwien.ac.at> ha
scritto:<br> </font> </div> <div class="y_msg_container">Hi,<br clear="none"><br clear="none"> > I'm trying to configure PETSc in an cluster with multiple GPUs, one per<br clear="none">> node.<br clear="none">> I read it is possible to use PETSc with N processes, one per GPU, each<br clear="none">> process accesses to its own GPU. Isn't it?<br clear="none"><br clear="none">yes, this is possible and the recommended way of running it.<div class="yqt6823306745" id="yqtfd59168"><br clear="none"><br clear="none">> Can anyone help me in that configuration?<br clear="none">> should I change the code to realize communication and exchange between<br clear="none">> processes on different GPUs?</div><br clear="none"><br clear="none">No, this is all handled internally by PETSc. If your code handles <br clear="none">'standard PETSc with MPI on the CPU', then the use of GPUs is just a <br clear="none">matter of providing the right flags
(-vec_type XXX -mat_type YYY).<br clear="none"><br clear="none">Which algorithms do you intend to run? Many solver-related algorithms <br clear="none">don't map well to massively parallel architectures such as GPUs, so you <br clear="none">may better want to run a more efficient CPU-based algorithm.<br clear="none"><br clear="none">Best regards,<br clear="none">Karli<div class="yqt6823306745" id="yqtfd51452"><br clear="none"></div><br><br></div> </div> </div> </div> </div></body></html>