<div dir="ltr">Hi,<br><br><div class="gmail_quote"><div dir="ltr">On Mon, Oct 3, 2016 at 12:01 AM Justin Lemkul <<a href="mailto:jalemkul@vt.edu">jalemkul@vt.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br class="gmail_msg">
<br class="gmail_msg">
On 10/2/16 5:42 PM, Boris Timofeev wrote:<br class="gmail_msg">
><br class="gmail_msg">
> Good afternoon to all!<br class="gmail_msg">
><br class="gmail_msg">
> I did a great job on compilation of the "portable" Gromacs-2016 version on the<br class="gmail_msg">
> Win64 platform without cygwin with OpenCL support and AVX_256/AVX2_256<br class="gmail_msg">
> expansions. The project is builded on VS-2015. By small changes in CMakeLists, I<br class="gmail_msg">
> managed to achieve that both the "embedded" tests, and regressiontests, are<br class="gmail_msg">
> executed after assembly immediately from VS IDE. It was tested on Windows7/<br class="gmail_msg">
> Windows10 with IntelCore i3, i5, i7 processors and video cards from Nvidia and<br class="gmail_msg">
> AMD. So far it was not succeeded to win - against the video card from Intel -<br class="gmail_msg">
> Intel OpenCL compiler preprocessor is left unfinished, does not recognize<br class="gmail_msg">
> directive "-I" and have buggy concatention (##) implementation.<br class="gmail_msg">
> It was necessary to realize a primitive preprocessor, became successful, but<br class="gmail_msg">
> all the same calculations (tests and regressiontests) are wrong as it was<br class="gmail_msg">
> already metioned here (<a href="https://bugs.freedesktop.org/show_bug.cgi" rel="noreferrer" class="gmail_msg" target="_blank">https://bugs.freedesktop.org/show_bug.cgi</a>?<br class="gmail_msg">
> id=94265#add_comment).<br class="gmail_msg"></blockquote><div><br></div><div>Intel isn't supported - see <a href="http://manual.gromacs.org/documentation/2016/user-guide/mdrun-performance.html#running-the-opencl-version-of-mdrun">http://manual.gromacs.org/documentation/2016/user-guide/mdrun-performance.html#running-the-opencl-version-of-mdrun</a></div><div><br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
> If it is interesting to community, I am ready to report about some necessary<br class="gmail_msg">
> changes in progect and to provide the main CMake-file.<br class="gmail_msg"></blockquote><div><br></div><div>Suggestions are indeed welcome. :-) If you want to apply a fix to our git branch called release-2016, you can upload that to our code review site <a href="https://gerrit.gromacs.org/#/q/status:open">https://gerrit.gromacs.org/#/q/status:open</a>. Instructions at <a href="http://www.gromacs.org/Developer_Zone/Git/Gerrit">http://www.gromacs.org/Developer_Zone/Git/Gerrit</a>. Or if it's a simple thing, just describe it here.</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
> There are several questions to developers.<br class="gmail_msg">
><br class="gmail_msg">
> 1. The nbnxn_ocl_kernel_nvidia.clh, nbnxn_ocl_kernel_nowarp.clh and<br class="gmail_msg">
> nbnxn_ocl_kernel_amd.clh files, if to compare their with kdiff, differ only in<br class="gmail_msg">
> comments and lack of unroll pragma for Nvidia. Why not to unite them in one?<br class="gmail_msg">
> Where the promised optimization?<br class="gmail_msg"></blockquote><div><br></div><div>Nobody has had the resources to look into that more closely, but with GROMACS being an open source project, such contributions are welcome! NVIDIA isn't a priority because there the native CUDA port does better (and NVIDIA's OpenCL seems to have a bug that means the non-blocking offloading calls that GROMACS needs run blocking style).</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
> 2. Why in the Gromacs'a code so many paths to files are embedded up? To start<br class="gmail_msg">
> tests without development environment on other computer it is necessary to copy<br class="gmail_msg">
> practically all project.<br class="gmail_msg"></blockquote><div><br></div><div>I'm not sure what you're referring to. The JIT compilation that is required for OpenCL requires that some source code is available. The regression tests run a lot of simulations, and those require to be able to read the databases of how to describe simulation systems. Nor have we targeted being able to run the tests with neither a development environment nor a full installation available, since we've never encountered a need for that.</div><div><br></div><div>Mark</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
> 3. A question "on science". When performing an example of<br class="gmail_msg">
> <a href="http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/lysozyme/01_pdb2gmx.html" rel="noreferrer" class="gmail_msg" target="_blank">http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/lysozyme/01_pdb2gmx.html</a>,<br class="gmail_msg">
> pdb2gmx for the aminoacid residue HIS ("HIZE-branch" on stdout ), unlike all<br class="gmail_msg">
> others residue, gives a nonintegral charge 0.883, inexplicable with round-off<br class="gmail_msg">
> errors.<br class="gmail_msg">
> Whether there is no program mistake here?<br class="gmail_msg">
><br class="gmail_msg">
<br class="gmail_msg">
I can't speak to points 1 and 2, but here: what is HIZE-branch? I've never<br class="gmail_msg">
heard of that. Note, too, that if you're working with a single amino acid with<br class="gmail_msg">
OPLS-AA, you can't rely on the default terminus selection. You need to choose<br class="gmail_msg">
the zwitterion termini, otherwise the charges will be junk because OPLS-AA makes<br class="gmail_msg">
changes to CA depending on whether the residue is a zwitterion or in a polypeptide.<br class="gmail_msg">
<br class="gmail_msg">
-Justin<br class="gmail_msg">
<br class="gmail_msg">
--<br class="gmail_msg">
==================================================<br class="gmail_msg">
<br class="gmail_msg">
Justin A. Lemkul, Ph.D.<br class="gmail_msg">
Ruth L. Kirschstein NRSA Postdoctoral Fellow<br class="gmail_msg">
<br class="gmail_msg">
Department of Pharmaceutical Sciences<br class="gmail_msg">
School of Pharmacy<br class="gmail_msg">
Health Sciences Facility II, Room 629<br class="gmail_msg">
University of Maryland, Baltimore<br class="gmail_msg">
20 Penn St.<br class="gmail_msg">
Baltimore, MD 21201<br class="gmail_msg">
<br class="gmail_msg">
<a href="mailto:jalemkul@outerbanks.umaryland.edu" class="gmail_msg" target="_blank">jalemkul@outerbanks.umaryland.edu</a> | (410) 706-7441<br class="gmail_msg">
<a href="http://mackerell.umaryland.edu/~jalemkul" rel="noreferrer" class="gmail_msg" target="_blank">http://mackerell.umaryland.edu/~jalemkul</a><br class="gmail_msg">
<br class="gmail_msg">
==================================================<br class="gmail_msg">
--<br class="gmail_msg">
Gromacs Developers mailing list<br class="gmail_msg">
<br class="gmail_msg">
* Please search the archive at <a href="http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List" rel="noreferrer" class="gmail_msg" target="_blank">http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List</a> before posting!<br class="gmail_msg">
<br class="gmail_msg">
* Can't post? Read <a href="http://www.gromacs.org/Support/Mailing_Lists" rel="noreferrer" class="gmail_msg" target="_blank">http://www.gromacs.org/Support/Mailing_Lists</a><br class="gmail_msg">
<br class="gmail_msg">
* For (un)subscribe requests visit<br class="gmail_msg">
<a href="https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers" rel="noreferrer" class="gmail_msg" 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" class="gmail_msg" target="_blank">gmx-developers-request@gromacs.org</a>.<br class="gmail_msg">
</blockquote></div></div>