<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body dir="auto"><div><div style="direction: inherit;">All changes are available in the sense that they are public, but there is no guarantee whatsoever that they produce correct results or are representative of the performance that will be in the release.</div><div style="direction: inherit;"><br></div><div style="direction: inherit;">Not that we think it's bad, but the main reason things haven't been committed is that it's not completely ready yet. </div><div style="direction: inherit;"><br></div><div style="direction: inherit;">Feel free to play with them, but we want to spend the efforts on finishing the changes rather than supporting unfinished ones :-)</div><div style="direction: inherit;"><br></div><div style="direction: inherit;">Cheers,</div><div style="direction: inherit;"><br></div><div style="direction: inherit;">Erik</div><br>Erik Lindahl <<a href="mailto:erik.lindahl@scilifelab.se">erik.lindahl@scilifelab.se</a>><div>Professor of Biophysics</div><div>Science for Life Laboratory</div><div>Stockholm University & KTH</div><div>Office (SciLifeLab): +46 8 524 81567</div><div>Cell (Sweden): +46 73 4618050</div><div>Cell (US): +1 267 3078746</div><div><br></div></div><div><br>On 19 Jan 2017, at 11:52, Jochen Hub <<a href="mailto:jhub@gwdg.de">jhub@gwdg.de</a>> wrote:<br><br></div><blockquote type="cite"><div><span>Hi Berk,</span><br><span></span><br><span>many thanks for the quick reply. We are btw targeting purely throughput.</span><br><span></span><br><span>Is the development tree somewhere available, so we can get a rough feeling on the performace on our GPUs?</span><br><span></span><br><span>Many thanks,</span><br><span>Jochen</span><br><span></span><br><span>Am 19/01/2017 um 09:30 schrieb Berk Hess:</span><br><blockquote type="cite"><span>Hi,</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>A basic version of PME on CUDA GPUs will be shipped with the 2017</span><br></blockquote><blockquote type="cite"><span>release. Most likely this will only be single GPU, possibly with the</span><br></blockquote><blockquote type="cite"><span>option to run pair interactions on one GPU and PME on another. The</span><br></blockquote><blockquote type="cite"><span>changes up in gerrit are all working, but not complete yet. Aleksei has</span><br></blockquote><blockquote type="cite"><span>a development tree with a complete implementation. The main question is</span><br></blockquote><blockquote type="cite"><span>how much performance optimization and feature completion we can do</span><br></blockquote><blockquote type="cite"><span>before the 2017 release.</span><br></blockquote><blockquote type="cite"><span>PME on GPU will allow you to buy a CPU+GPU cluster with cheaper CPUs if</span><br></blockquote><blockquote type="cite"><span>you are targeting throughput. It is less clear what the best setup will</span><br></blockquote><blockquote type="cite"><span>be for the highest ns/day, since PME requires a lot of different kernels</span><br></blockquote><blockquote type="cite"><span>which results in higher overheads.</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>Cheers,</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>Berk</span><br></blockquote><blockquote type="cite"><span></span><br></blockquote><blockquote type="cite"><span>On 01/19/2017 08:55 AM, Jochen Hub wrote:</span><br></blockquote><blockquote type="cite"><blockquote type="cite"><span>Hi developers,</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>I noticed that Aleksei has uploaded a whole bunch of commits for</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>implenting PME for GPUs. Since we about to buy a new cluster, I was</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>wondering if there is a rough time line when PME/CUDA code goes into</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>master and realease branches. If PME runs smoothly on GPUs, the</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>hardware with the best price/performance ratio obviously changes.</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>Also, are the commits in the branch "master (pme)" already suitable</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>for some preliminary benchmarking?</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>Many thanks,</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span>Jochen</span><br></blockquote></blockquote><blockquote type="cite"><blockquote type="cite"><span></span><br></blockquote></blockquote><blockquote type="cite"><span></span><br></blockquote><span></span><br><span>-- </span><br><span>---------------------------------------------------</span><br><span>Dr. Jochen Hub</span><br><span>Computational Molecular Biophysics Group</span><br><span>Institute for Microbiology and Genetics</span><br><span>Georg-August-University of Göttingen</span><br><span>Justus-von-Liebig-Weg 11, 37077 Göttingen, Germany.</span><br><span>Phone: +49-551-39-14189</span><br><span><a href="http://cmb.bio.uni-goettingen.de/">http://cmb.bio.uni-goettingen.de/</a></span><br><span>---------------------------------------------------</span><br><span>-- </span><br><span>Gromacs Developers mailing list</span><br><span></span><br><span>* Please search the archive at <a href="http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List">http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List</a> before posting!</span><br><span></span><br><span>* Can't post? Read <a href="http://www.gromacs.org/Support/Mailing_Lists">http://www.gromacs.org/Support/Mailing_Lists</a></span><br><span></span><br><span>* For (un)subscribe requests visit</span><br><span><a href="https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers">https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers</a> or send a mail to <a href="mailto:gmx-developers-request@gromacs.org">gmx-developers-request@gromacs.org</a>.</span></div></blockquote></body></html>