<div class="gmail_quote">On Fri, Jul 27, 2012 at 4:34 AM, Roland Schulz <span dir="ltr"><<a href="mailto:roland@utk.edu" target="_blank">roland@utk.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">On Thu, Jul 26, 2012 at 10:26 PM, Szilárd Páll <<a href="mailto:szilard.pall@cbr.su.se">szilard.pall@cbr.su.se</a>> wrote:<br>
> On Fri, Jul 27, 2012 at 4:06 AM, Szilárd Páll <<a href="mailto:szilard.pall@cbr.su.se">szilard.pall@cbr.su.se</a>><br>
> wrote:<br>
>><br>
</div><div><div class="h5">>> On Thu, Jul 26, 2012 at 2:09 PM, Jochen Hub <<a href="mailto:jhub@gwdg.de">jhub@gwdg.de</a>> wrote:<br>
>>><br>
>>> Hi,<br>
>>><br>
>>> I am trying to compile and run the git master on my Macbook air (OS X<br>
>>> Lion). Without success. If I compile with a newer gcc (4.5 or newer,<br>
>>> installed from Macports), I get errors like (does this have to do with AVX?)<br>
>>><br>
>>> [ 1%] Building C object<br>
>>> src/gromacs/CMakeFiles/libgromacs.dir/gmxpreprocess/add_par.c.o<br>
>>> /var/folders/ys/rh9lzqpj7854h34d2__mznph0000gn/T//ccPxJmjg.s:66:no such<br>
>>> instruction: `vmovups 0(%r13), %ymm0'<br>
>>> /var/folders/ys/rh9lzqpj7854h34d2__mznph0000gn/T//ccPxJmjg.s:69:no such<br>
>>> instruction: `vmovups %ymm0, 24(%rdi)'<br>
>>> /var/folders/ys/rh9lzqpj7854h34d2__mznph0000gn/T//ccPxJmjg.s:79:no such<br>
>>> instruction: `vmovss 0(%r13), %xmm1'<br>
>>> /var/folders/ys/rh9lzqpj7854h34d2__mznph0000gn/T//ccPxJmjg.s:83:no such<br>
>>> instruction: `vmovss %xmm1, 24(%rdi,%r9,4)'<br>
>>> /var/folders/ys/rh9lzqpj7854h34d2__mznph0000gn/T//ccPxJmjg.s:99:no such<br>
>>> instruction: `vmovss 0(%r13), %xmm2'<br>
>>> /var/folders/ys/rh9lzqpj7854h34d2__mznph0000gn/T//ccPxJmjg.s:102:no such<br>
>>> instruction: `vmovss %xmm2, 24(%rdi,%r9,4)'<br>
>><br>
>><br>
>> I can confirm this, works on Linux, fails on Mac OS Lion with gcc 4.5 -<br>
>> 4.7 with GMX_ACCELERATION=AVX_256. It looks like AVX instructions are<br>
>> somehow not recognized, although the aforementioned compilers are supposed<br>
>> to have support for it.<br>
>><br>
>>><br>
>>> On a gcc 4.4 or earlier, compiling works fine, but mdruns stops with a<br>
>>> segfault. A backtrace in gdb<br>
>><br>
>><br>
>> That's strange as gcc 4.4 is supposed to have AVX support<br>
>> (<a href="http://goo.gl/Rt3BM" target="_blank">http://goo.gl/Rt3BM</a>).<br>
><br>
><br>
</div></div><div class="im">> Just checked and, as expected, the same compiler errors show up with gcc<br>
> 4.4.7 as well. I'm not sure what did you change in the configuration to get<br>
> the code built, but you must have used GMX_ACCELERATION != AVX_256.<br>
><br>
>><br>
>><br>
>>><br>
</div><div><div class="h5">>>> gives the following. Seems like something goes wrong in FFTW (which was<br>
>>> compiled with the same gcc and with --enable-threads --enable-sse<br>
>>> --enable-sse2).<br>
>>><br>
>>> Program received signal EXC_BAD_ACCESS, Could not access memory.<br>
>>> Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000048<br>
>>> [Switching to process 44300 thread 0x1b03]<br>
>>> 0x0000000100050ebd in gomp_resolve_num_threads ()<br>
>>> (gdb) bt<br>
>>> #0 0x0000000100050ebd in gomp_resolve_num_threads ()<br>
>>> #1 0x0000000100050fc3 in GOMP_parallel_start ()<br>
>>> #2 0x00000001004c0bc2 in fft5d_plan_3d ()<br>
>>> #3 0x0000000100434a52 in gmx_parallel_3dfft_init ()<br>
>>> #4 0x000000010046b6fc in gmx_pme_init ()<br>
>>> #5 0x0000000100026ab3 in mdrunner (nthreads_requested=4, fplog=0x0,<br>
>>> cr=0x102100b40, nfile=36, fnm=0x103808200, oenv=0x101000c10, bVerbose=0,<br>
>>> bCompact=1, nstglobalcomm=-1, ddxyz=0x1013c0e04, dd_node_order=1, rdd=0,<br>
>>> rconstr=0, dddlb_opt=0x10002e26a "auto", dlb_scale=0.800000012, ddcsx=0x0,<br>
>>> ddcsy=0x0, ddcsz=0x0, nstepout=100, resetstep=-1, nmultisim=0,<br>
>>> repl_ex_nst=0, repl_ex_nex=0, repl_ex_seed=-1, pforce=-1, cpt_period=15,<br>
>>> max_hours=-1, deviceOptions=0x10002e276 "", Flags=7168) at<br>
>>> /Users/jhub/src/gmx/gromacs/src/programs/mdrun/runner.c:844<br>
>>> #6 0x0000000100024f2d in mdrunner_start_fn (arg=0x101005d60) at<br>
>>> /Users/jhub/src/gmx/gromacs/src/programs/mdrun/runner.c:173<br>
>>> #7 0x0000000100242bfb in tMPI_Thread_starter ()<br>
>>> #8 0x00007fff9785f8bf in _pthread_start ()<br>
>>> #9 0x00007fff97862b75 in thread_start ()<br>
>><br>
>><br>
>> Can't confirm this one right now as there's no gcc 4.4 on the Mac OS build<br>
>> server. Will check it later. However, it sounds like it's a similar issue to<br>
>> the ones we've had with an ordered clause in a #pragma omp statment.<br>
><br>
><br>
</div></div><div class="im">> Unfortunately, I can confirm this crash. It seems to occur with<br>
> OMP_NUM_THREADS=m and -nt n, where m>1 and n >1.<br>
<br>
</div>Could you check whether this the error goes away with a version prior<br>
to d75b7537ed4df95fd3796d9432e9c1e64adec26a<br>
Maybe the new locking added in that commit is not OK.<br></blockquote><div><br></div><div>No, it does not:</div><div><div><br></div><div>$ mdrun -version 2>&1 | tail -n 5</div><div>Version: VERSION 4.6-dev-20120704-103d98c</div>
<div>GIT SHA1 hash: 103d98cf31f064cce7c6ec14fcccd081aa2bbfac</div><div>Precision: single</div><div>Parallellization: thread_mpi</div><div>FFT Library: fftw3</div></div><div><br></div><div>and in fact it looks like it's a steange tMPI+OpenMP issue because the crash happens even with one omp thread:</div>
<div><br></div><div>$ OMP_NUM_THREADS=1 $mdrun -nt 2 -quiet -stepout 1 2>/dev/null</div><div><div>[...]</div><div><br></div><div>Segmentation fault: 11</div></div><div><br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<span class="HOEnZb"><br><font color="#888888">
Roland</font><br>
</span><div class="HOEnZb"><div class="h5"><br>
><br>
>><br>
>><br>
>>><br>
>>><br>
>>> Can anyone give me a hint how to fix this? Or is the master so<br>
>>> experimental that it is not interned to be used at all right now?<br>
>><br>
>><br>
>> Unfortunately, this affects the release-4-6 branch as well. Could you<br>
>> please file a bug on Redmine on the first issue and have a look at the<br>
>> related issue on the second.<br>
>><br>
>> Cheers,<br>
>> --<br>
>> Sz.<br>
>><br>
>>><br>
>>> Many thanks,<br>
>>> Jochen<br>
>>><br>
>>><br>
>>> --<br>
>>> ---------------------------------------------------<br>
>>> Dr. Jochen Hub<br>
>>> Computational Molecular Biophysics Group<br>
>>> Institute for Microbiology and Genetics<br>
>>> Georg-August-University of Göttingen<br>
>>> Justus-von-Liebig-Weg 11, 37077 Göttingen, Germany.<br>
>>> Phone: <a href="tel:%2B49-551-39-14189" value="+495513914189">+49-551-39-14189</a><br>
>>> <a href="http://cmb.bio.uni-goettingen.de/" target="_blank">http://cmb.bio.uni-goettingen.de/</a><br>
>>> ---------------------------------------------------<br>
>>> --<br>
>>> gmx-developers mailing list<br>
>>> <a href="mailto:gmx-developers@gromacs.org">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 www<br>
>>> interface or send it to <a href="mailto:gmx-developers-request@gromacs.org">gmx-developers-request@gromacs.org</a>.<br>
>><br>
>><br>
><br>
<br>
<br>
<br>
</div></div><div class="im HOEnZb">--<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 class="HOEnZb"><div class="h5">--<br>
gmx-developers mailing list<br>
<a href="mailto:gmx-developers@gromacs.org">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">gmx-developers-request@gromacs.org</a>.<br>
</div></div></blockquote></div><br>