<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 10pt;
font-family:Verdana
}
--></style>
</head>
<body class='hmmessage'>
NO!<br>The problem does not vanish!<br>The temperature will seem ok, because NH will compensate all heat<br>introduced by integration errors. So you temperature looks ok,<br>but your sampling might be completely wrong.<br>In an NVT ensemble you can see this as an enormous drift<br>of the conserved energy quantity.<br><br>Berk<br><br>> Date: Thu, 11 Mar 2010 16:42:45 +0100<br>> From: schlesi@uni-mainz.de<br>> To: gmx-users@gromacs.org<br>> Subject: [gmx-users] Re: problem with icc compiler<br>> <br>> @ Erik:<br>> Have you also tried the nose-hoover thermostat? Because with it the <br>> problem vanishes for me (even if i use pme).<br>> <br>> > <br>> > Hi,<br>> > I have reported this problem to Bugzilla some months ago:<br>> > <br>> > http://bugzilla.gromacs.org/show_bug.cgi?id=378<br>> > <br>> > I don't exactly know what the problem is, but it is definitely related<br>> > to the MKL FFTs, which can be seen by switching from PME to cutoff, and<br>> > the problem should go away.<br>> > <br>> > I have reproduced this behavior on several computers, as I have written<br>> > in the bugzilla report.<br>> > <br>> > / Erik Brandt<br>> > <br>> > tor 2010-03-11 klockan 01:34 +0100 skrev Mark Abraham:<br>> > <br>> >> On 11/03/2010 3:04 AM, Thomas Schlesier wrote:<br>> >>> Gromacs was compiled on a xeon (woodcrest). for the simulations i used<br>> >>> an old xeon (don't know what chip, but also 64bit system) and a i7.<br>> >> Well, don't do that. IMO, dynamic linking should work in that kind of<br>> >> scenario, but something is clearly broken.<br>> >><br>> >>> About static / dynamic libraries:<br>> >>> I used there default settings. At the end of the configure command it<br>> >>> tells me the following:<br>> >>> * On most platforms you can save 10X space with dynamic libraries,<br>> >>> although the binaries might be less portable. Why not try --enable-shared?<br>> >>> So i think the libraries are static.<br>> >> Your internal GROMACS linking is static, which you could change with<br>> >> --enable-shared. You will need to kick the linker harder than that to<br>> >> force it to link statically to system libraries also.<br>> >><br>> >>> I tried Mark's suggestion and compiled a new version, where i change<br>> >>> '--with-fft=mkl' with '--enable-fft=fftpack' (the restr of the configure<br>> >>> command was the same then before). With that, the error didn't appear.<br>> >>> Does that mean that the linking to mkl did not work for mdrun, but<br>> >>> worked for mdrun_mpi (because there the temperature was right)?<br>> >> Yep, that is strong evidence for my hypothesis. Either compile GROMACS<br>> >> on the target execution system (even submit a cluster job for the<br>> >> compilation!), or (if the former doesn't work) get the MKL documentation<br>> >> and read it about how to enforce static linking. There may be some<br>> >> cunning linker option for that, or you may need to explicitly cite the<br>> >> static versions of the libraries. Or, get FFTW installed and link to<br>> >> that. I have found near-negligible performance differences between MKL<br>> >> and FFTW on my machine.<br>> >><br>> >>> One thing for the temperature jump:<br>> >>> Temperature starts at around 300 K (from 'gen_temp') then goes in 1-2ps<br>> >>> up to around 425 K and then stays there. the simulation was 100ps long,<br>> >>> in the end i had an average value of about 425 K (from log file).<br>> >>><br>> >>> Here are the first 20ps from g_energy<br>> >>> 0.000000 305.240509<br>> >>> 1.000000 381.614166<br>> >>> 2.000000 410.572906<br>> >>> 3.000000 434.954956<br>> >>> 4.000000 414.660400<br>> >>> 5.000000 394.799591<br>> >>> 6.000000 389.087128<br>> >>> 7.000000 414.893982<br>> >>> 8.000000 449.444183<br>> >>> 9.000000 417.877472<br>> >>> 10.000000 442.470306<br>> >>> 11.000001 446.170258<br>> >>> 12.000001 448.844666<br>> >>> 13.000001 412.847473<br>> >>> 14.000001 454.549744<br>> >>> 15.000001 447.908478<br>> >>> 16.000000 404.607422<br>> >>> 17.000000 404.629944<br>> >>> 18.000000 441.559448<br>> >>> 19.000000 396.328400<br>> >>> 20.000000 421.386017<br>> >> That's broken all right!<br>> >><br>> >> Mark<br>> >><br>> >>> and:<br>> >>> Energy Average RMSD Fluct. Drift Tot-Drift<br>> >>> ----------------------------------------------------------------<br>> >>> Temperature 424.625 21.7645 21.6696 0.0703201 7.03215<br>> >>><br>> >>><br>> >>><br>> >>>> On 10/03/2010 12:21 AM, Thomas Schlesier wrote:<br>> >>>>> Could anybody reproduce that error or has an idea what is happening?<br>> >>>>> Or i am alone with that problem?<br>> >>>> Nothing looks obviously wrong, but it's hard to be sure in the absence<br>> >>>> of information about your hardware. The most likely issue is some kind<br>> >>>> of dynamically-linked library mismatch. This can happen if your<br>> >>>> execution environment differs from your linking environment. Try forcing<br>> >>>> linking to static versions of the libraries, which will prevent this.<br>> >>>><br>> >>>> Also try disabling things until you get sensible behaviour in all cases,<br>> >>>> like --enable-fft=fftpack. That would reveal that the problem was with<br>> >>>> linking to mkl.<br>> >>>><br>> >>>> Also 1-2ps is a bit too short to expect convergence of temperature -<br>> >>>> check the plot of T against t with g_energy.<br>> >>>><br>> >>>> Mark<br>> >>>><br>> >>>>>> Date: Fri, 5 Mar 2010 23:11:45 +0100<br>> >>>>>> From: Thomas Schlesier <schlesi@uni-mainz.de><br>> >>>>>> Subject: [gmx-users] problem with icc compiler<br>> >>>>>> To: "gmx-users@gromacs.org" <gmx-users@gromacs.org><br>> >>>>>> Message-ID: <4B9181A1.7060106@uni-mainz.de><br>> >>>>>> Content-Type: text/plain; charset="ISO-8859-1"; format=flowed<br>> >>>>>><br>> >>>>>> Hi,<br>> >>>>>> i observed the following problem. if i simulate water (spc or tip4p)<br>> >>>>>> with gromacs 4.0.5 i get with v-rescale or berendsen thermostat the<br>> >>>>>> wrong temperature (ref_t = 300K -> average around 425K, in about<br>> >>>>>> 1-2ps),<br>> >>>>>> but only in serial, not in parallel runs.<br>> >>>>>> non-water molecules or nose-hoover thermostat make no problems.<br>> >>>>>> see also<br>> >>>>>> http://lists.gromacs.org/pipermail/gmx-users/2010-March/049248.html<br>> >>>>>> for mdp and log file.<br>> >>>>>><br>> >>>>>> gromacs was compiled with the following comands:<br>> >>>>>> and in the file 'configure' all '-lmkl' were deleted (don't ask me why,<br>> >>>>>> i don't really understand that stuff, the command were from our<br>> >>>>>> previous<br>> >>>>>> phd student).<br>> >>>>>><br>> >>>>>> ./configure CC="icc"<br>> >>>>>> CPPFLAGS="-I/share/apps/intel/mkl/10.0.011/include"<br>> >>>>>> LDFLAGS="-L/share/apps/intel/mkl/10.0.011/lib/em64t<br>> >>>>>> -lmkl_solver_lp64_sequential -Wl,--start-group -lmkl_intel_lp64<br>> >>>>>> -lmkl_sequential -lmkl_core -Wl,--end-group -lpthread" --with-fft=mkl<br>> >>>>>> --prefix="/share/apps/gromacs/4.0.5"<br>> >>>>>> make<br>> >>>>>> make install<br>> >>>>>> make clean<br>> >>>>>> ./configure CC="icc"<br>> >>>>>> CPPFLAGS="-I/share/apps/intel/mkl/10.0.011/include"<br>> >>>>>> LDFLAGS="-L/share/apps/intel/mkl/10.0.011/lib/em64t<br>> >>>>>> -lmkl_solver_lp64_sequential -Wl,--start-group -lmkl_intel_lp64<br>> >>>>>> -lmkl_sequential -lmkl_core -Wl,--end-group -lpthread" --with-fft=mkl<br>> >>>>>> --prefix="/share/apps/gromacs/4.0.5" --enable-mpi --disable-nice<br>> >>>>>> --program-suffix=_mpi<br>> >>>>>> make mdrun<br>> >>>>>> make install-mdrun<br>> >>>>>><br>> >>>>>> for gromacs 4.0.5 i used the icc 9.1.046 compiler.<br>> >>>>>><br>> >>>>>> i also tried gromacs 4.0.7 with icc 9.1.046 and icc 10.1.008 with spc<br>> >>>>>> water, v-rescale thermostat.<br>> >>>>>> -> serial: too high temperature 425K iso 300K<br>> >>>>>> -> parallel: no problems<br>> >>>>>> with non-water (mesitylene) i have no problem in serial.<br>> >>>>>><br>> >>>>>> the problem does not come from grompp because i can use same tpr-file<br>> >>>>>> for serial and parallel runs with the above results.<br>> >>>>>><br>> >>>>>> if someone needs more informations about this please tell me.<br>> >>>>>><br>> >>>>>> greetings<br>> >>>>>> thomas<br>> >>>>>><br>> -- <br>> gmx-users mailing list gmx-users@gromacs.org<br>> http://lists.gromacs.org/mailman/listinfo/gmx-users<br>> Please search the archive at http://www.gromacs.org/search before posting!<br>> Please don't post (un)subscribe requests to the list. Use the <br>> www interface or send it to gmx-users-request@gromacs.org.<br>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php<br> <br /><hr />Express yourself instantly with MSN Messenger! <a href='http://clk.atdmt.com/AVE/go/onm00200471ave/direct/01/' target='_new'>MSN Messenger</a></body>
</html>