Hi,<div><br></div><div>the CFLAGS are set correctly to -fastsse for PGI.</div><div><br></div><div>But while testing this I found a few problems with cmake (I haven't used it for a while - usually use automake).</div><div>
All not critical since automake is still available as alternative. Just writing this here to report what I find less convenient than with automake and what probably should be improved before automake is removed.</div><div>
<br></div><div>It doesn't automatically set Acceleration to SSE. This isn't implemented yet, right?</div>
<div><br></div><div>If one needs to specify the location of FFTW manually it is a little bit confusing. It would be good if after the error about the missing fftw is shown, that than one is able to enter that location to FFTW without going to the advanced settings. Also the description for the FFTW configuration fields for include/lib (FFTW3_INCLUDE_DIR, FFTW3_LIBRARIES) says "Path to .." . But it wants the -I/-L with the argument. Thus the full argument not only the path.</div>
<div><br></div><div>On Cray it is difficult to compile without their mpicc wrapper. And cmake doesn't detect all the libraries automatically (which I didn't expect). It would be much easier if it would just use the mpicc wrapper instead trying to find the libraries. If there is some advantage on other systems to not use the wrapper than at least it would be helpful if one could select that one doesn't want any MPI_INCLUDE_PATH oder MPI_LIBRARY. If one sets those to empty it tries to auto-detect again in the next step and then one gets again the wrong values.</div>
<div><br></div><div>I can't test the xlc flags because the one machine I have access to with xlc doesn't have cmake installed.</div><div><br></div><div>Roland<br><br><div class="gmail_quote">On Thu, Jul 1, 2010 at 11:43 AM, Sander Pronk <span dir="ltr"><<a href="mailto:pronk@cbr.su.se" target="_blank">pronk@cbr.su.se</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div style="word-wrap:break-word">Thanks! <div>I've implemented automatic flags for pgi and xlc (CMake doesn't appear to be able to detect pathscale); could you (or somebody else) test these, perhaps?</div>
<div><div><br></div><div>Sander<br><div><br></div><div><br><div><div><div></div><div><div>On Jun 30, 2010, at 23:39 , Roland Schulz wrote:</div><br></div></div><blockquote type="cite"><div><div></div><div>
<br><br><div class="gmail_quote">On Wed, Jun 30, 2010 at 7:56 AM, Sander Pronk <span dir="ltr"><<a href="mailto:pronk@cbr.su.se" target="_blank">pronk@cbr.su.se</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I've implemented auto-detection of the CFLAGS (and CXXFLAGS). Right now, it works for gcc and icc. There are a few compilers I'd like to add, that I don't have access to. Specifically:<br>
<br>
xlc on Linux and AIX<br>
the pgi compilers on x86<br></blockquote><div>-fastsse. </div><div>Not sure whether IPA helps. If one wants IEEE correct math one needs to add -Kieee. Does someone know whether this is required for GROMACS? One can also gain with GCC by adding -ffast-math. Is this safe?</div>
<div>Currently PGI doesn't compile. But Peter told me he has a patch almost ready.</div><div><br></div><div>Pathscale:</div><div>-O3 -OPT:Ofast -fno-math-errno -ffast-math (if fast-math and no-math-errno is OK, again not sure if ipa helps)</div>
<div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">and any compiler on Bluegene.<br></blockquote>-O3 -qarch=auto -qtune=auto<br><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Would anybody who has insight on what compiler options to set (and test for) on these platforms, mind looking at cmake/gmxCFlags.cmake or tell me what they think should be added?<br>
<br>
I'm specifically looking for how to enable syntactical elements (like inline functions) and how to set the best optimization and debug settings.</blockquote><div><br></div><div>I hope this helps. Let me know if you need something further or want me to test something.</div>
<div><br></div><div>Roland</div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><font color="#888888"><font><br>
</font></font><font color="#888888">
<br>
--<br>
gmx-developers mailing list<br>
<a href="mailto:gmx-developers@gromacs.org" target="_blank">gmx-developers@gromacs.org</a><br>
<a href="http://lists.gromacs.org/mailman/listinfo/gmx-developers" target="_blank">http://lists.gromacs.org/mailman/listinfo/gmx-developers</a><br>
Please don't post (un)subscribe requests to the list. Use the<br>
www interface or send it to <a href="mailto:gmx-developers-request@gromacs.org" target="_blank">gmx-developers-request@gromacs.org</a>.<br>
</font></blockquote></div><br><br clear="all"><br>-- <br>ORNL/UT Center for Molecular Biophysics <a href="http://cmb.ornl.gov/" target="_blank">cmb.ornl.gov</a><br>865-241-1537, ORNL PO BOX 2008 MS6309<br></div></div>
-- <br><div>gmx-developers mailing list<br><a href="mailto:gmx-developers@gromacs.org" target="_blank">gmx-developers@gromacs.org</a><br><a href="http://lists.gromacs.org/mailman/listinfo/gmx-developers" target="_blank">http://lists.gromacs.org/mailman/listinfo/gmx-developers</a><br>
Please don't post (un)subscribe requests to the list. Use the <br>www interface or send it to <a href="mailto:gmx-developers-request@gromacs.org" target="_blank">gmx-developers-request@gromacs.org</a>.</div></blockquote>
</div><br></div></div></div></div><br>--<br>
gmx-developers mailing list<br>
<a href="mailto:gmx-developers@gromacs.org" target="_blank">gmx-developers@gromacs.org</a><br>
<a href="http://lists.gromacs.org/mailman/listinfo/gmx-developers" target="_blank">http://lists.gromacs.org/mailman/listinfo/gmx-developers</a><br>
Please don't post (un)subscribe requests to the list. Use the<br>
www interface or send it to <a href="mailto:gmx-developers-request@gromacs.org" target="_blank">gmx-developers-request@gromacs.org</a>.<br></blockquote></div><br><br clear="all"><br>-- <br>ORNL/UT Center for Molecular Biophysics <a href="http://cmb.ornl.gov" target="_blank">cmb.ornl.gov</a><br>
865-241-1537, ORNL PO BOX 2008 MS6309<br>
</div>