<div dir="ltr">Hi Mirco,<br><br>I've solved three correctness issues with the OpenCL implementation now, and NPT on AMD looks quite nice. Can you please try the code at <a href="https://gerrit.gromacs.org/#/c/4314/26">https://gerrit.gromacs.org/#/c/4314/26</a> and let us know if the situation is improved?<div><br></div><div>Thanks!</div><div><br></div><div>Mark</div></div><br><div class="gmail_quote"><div dir="ltr">On Tue, Jun 9, 2015 at 9:34 PM Szilárd Páll <<a href="mailto:pall.szilard@gmail.com">pall.szilard@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Mirco,<br>
<br>
Thanks for running the tests and sharing the logs. Performance looks<br>
reasonable and based on your log file it looks like the large OpenCL<br>
runtime overhead I observed on Linux is not present on Windows.<br>
<br>
On Sun, Jun 7, 2015 at 12:23 PM, Mirco Wahab<br>
<<a href="mailto:mirco.wahab@chemie.tu-freiberg.de" target="_blank">mirco.wahab@chemie.tu-freiberg.de</a>> wrote:<br>
> Hi Szilard,<br>
><br>
> On 31.05.2015 22:15, Szilárd Páll wrote:<br>
>><br>
>> [...]<br>
>> <a href="https://gerrit.gromacs.org/#/c/4314/" rel="noreferrer" target="_blank">https://gerrit.gromacs.org/#/c/4314/</a><br>
>> Use the version in review on gerrit I linked above, that's the most up<br>
>> to date code.<br>
>> [...]<br>
><br>
><br>
> I completed some tests using an AMD device (Pitcairn-GPU,<br>
> Radeon R9-270X), and used adh_cubic_vsites, rnase_cubic, and<br>
> villin_vsites from the "gromacs acceleration page".<br>
><br>
> So far, the simulations appear to run stable and deliver correct<br>
> results if *no pressure coupling* is used. As I noted already with<br>
> the NV opencl test, this affects the AMD opencl implementation too:<br>
> pressure coupling just don't work so far.<br>
><br>
> Compilation: there needs to be an additional include of<br>
> #include <algorithm> in src/gromacs/gmxana/gmx_wham.cpp<br>
> in order to provide std::min/max for VC. Another problem<br>
> is the use of binary numerals in src\gromacs\listed-forces\bonded.cpp,<br>
> which are C++14. Maybe these could be written in hexadecimal<br>
> numerals in order to avoid problems w/non C++14-compilers?<br>
<br>
I can't comment much on these, but I don't think we should be using C++14 code.<br>
<br>
> Pressure coupling: Depending on the simulation setup w/pressure<br>
> coupling, gmx will either crash immediately (when using pme) or<br>
> continuously enlarge the volume (eg. when using rf + berendsen).<br>
<br>
Yeah, that should not happen. I did very brief testing, and my guess<br>
is that something is wrong with the shift forces during non-energy<br>
calculating steps.<br>
<br>
> I put the log files an a host, these include log files and<br>
> input for a very simple spc water system w/pressure coupling<br>
> (which doesn't work w/opencl but works perfectly using -nb cpu).<br>
> <a href="http://spule.compch.tu-freiberg.de/~wahab/gromacs/gmx_opencl_2015-06_szilard.zip" rel="noreferrer" target="_blank">http://spule.compch.tu-freiberg.de/~wahab/gromacs/gmx_opencl_2015-06_szilard.zip</a><br>
><br>
> short overview:<br>
><br>
> adh_cubic_vsites<br>
> cpu/pme: 5.738, gpu/pme: 17.421<br>
> cpu/rf: 7.913, gpu/rf: 26.344<br>
><br>
> rnase_cubic<br>
> cpu/pme: 15.001, gpu/pme: 39.363<br>
> cpu/rf: 24.729, gpu/rf: 59.179<br>
><br>
> villin_vsites<br>
> cpu/pme: 124.86, gpu/pme: 284.213<br>
> cpu/rf: 151.05, gpu/rf: 310.794<br>
><br>
> (water_small_npt/nvt)<br>
> cpu/pme/npt: 20.00 gpu/pme/npt: n/a<br>
> cpu/pme/nvt: 21.65 gpu/pme/nvt: 52.05<br>
><br>
><br>
> How is the current status for the planned inclusion of<br>
> opencl into a gromacs release?<br>
<br>
I can't comment on that, there are a few issues (including this<br>
pressure bug) that need to be solved before merge can be considered.<br>
Hopefully somebody will have time to look at these.<br>
<br>
--<br>
Szilárd<br>
<br>
> Regards,<br>
><br>
><br>
> M.<br>
><br>
> --<br>
> Gromacs Developers mailing list<br>
><br>
> * Please search the archive at<br>
> <a href="http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List" rel="noreferrer" target="_blank">http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List</a> before<br>
> posting!<br>
><br>
> * Can't post? Read <a href="http://www.gromacs.org/Support/Mailing_Lists" rel="noreferrer" target="_blank">http://www.gromacs.org/Support/Mailing_Lists</a><br>
><br>
> * For (un)subscribe requests visit<br>
> <a href="https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers" rel="noreferrer" target="_blank">https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers</a> or<br>
> send a mail to <a href="mailto:gmx-developers-request@gromacs.org" target="_blank">gmx-developers-request@gromacs.org</a>.<br>
--<br>
Gromacs Developers mailing list<br>
<br>
* Please search the archive at <a href="http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List" rel="noreferrer" target="_blank">http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List</a> before posting!<br>
<br>
* Can't post? Read <a href="http://www.gromacs.org/Support/Mailing_Lists" rel="noreferrer" target="_blank">http://www.gromacs.org/Support/Mailing_Lists</a><br>
<br>
* For (un)subscribe requests visit<br>
<a href="https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers" rel="noreferrer" target="_blank">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" target="_blank">gmx-developers-request@gromacs.org</a>.</blockquote></div>