[gmx-users] Re: Benching against different compilers - Help needed
David van der Spoel
spoel at xray.bmc.uu.se
Wed May 3 15:22:56 CEST 2006
Jones de Andrade wrote:
> Hi again.
>
> Well, here we go:
>
> On 4/30/06, David van der Spoel <spoel at xray.bmc.uu.se> wrote:
>>
>> Jones de Andrade wrote:
>> > Up to now, I could finally compile it with Portland, in both single and
>> > double precision version. But, on the other hand, while the single
>> > precision works properly, the double precision fails in some tests,
>> like
>> > the "mixed" test on the "tutor" directory. It yelds me the following
>> > error message:
>> this is hard to debug, but what you could do is run the same tpr with
>> different executables and do gmxcheck -e -e2 to compare energies. that
>> might give a clue.
>> > step 0
>> > Step 1 Warning: pressure scaling more than 1%, mu: 1.01018 1.02193
>> 1.08117
>> > Step 2 Warning: pressure scaling more than 1%, mu: 1.01124 1.00787
>> 1.08362
>> > Step 143 Warning: pressure scaling more than 1%, mu: 1.00304 1.00017
>> 1.01009
>> > step 240, remaining runtime: 40 s
>> > Program mdrun_d, VERSION 3.3
>> > Source code file: stat.c, line: 283
>> > Fatal error:
>> > XTC error
>
> Ok, I did gmxcheck for both double precision versions of gromacs I got: one
> compiled with GCC, and the other with PGI. Here is what I got. Everything
> looks to be ugly and different, btw; no idea where to look first:
> *****************************************************************************************************************
>
> comparing energy file ener.edr and ener2.edr
>
> Opened ener.edr as double precision energy file
> Opened ener2.edr as double precision energy file
> There are 44 terms in the energy files
>
> There are 44 terms to compare in the energy files
>
> Reading frame 1 time 0.000 Pressure (bar) step 0:
> 2.26575e+06, step 0: -1212.92
> pV step 0: 3.53269e+06, step 0: -1891.14
> Vir-XX step 0: -475104, step 0: 2135.61
> Vir-XY step 0: 491116, step 0: -544.385
> Vir-XZ step 0: 763516, step 0: 180.112
> Vir-YX step 0: 491225, step 0: -544.385
> Vir-YY step 0: -1.02488e+06, step 0: 2008.1
> Vir-YZ step 0: -1.53177e+06, step 0: 558.876
> Vir-ZX step 0: 763218, step 0: 180.112
> Vir-ZY step 0: -1.53059e+06, step 0: 558.876
> Vir-ZZ step 0: -3.79554e+06, step 0: 2209.48
you have different pressure and virial, that means different forces.
maybe you can try to compile with portland but without optimization.
> Pres-XX (bar) step 0: 610883, step 0: -1286.57
> Pres-XY (bar) step 0: -630004, step 0: 664.233
> Pres-XZ (bar) step 0: -979427, step 0: -271.252
> Pres-YX (bar) step 0: -630144, step 0: 664.233
> Pres-YY (bar) step 0: 1.3161e+06, step 0: -1124.86
> Pres-YZ (bar) step 0: 1.96481e+06, step 0: -753.997
> Pres-ZX (bar) step 0: -979044, step 0: -271.252
> Pres-ZY (bar) step 0: 1.9633e+06, step 0: -753.997
> Pres-ZZ (bar) step 0: 4.87027e+06, step 0: -1227.32
> #Surf*SurfTen step 0: 9.18092e+06, step 0: -50.7689
> Reading frame 3 time 0.100 Angle step
> 50: 1570.43, step 50: 306.588
> LJ (SR) step 50: -449.51, step 50: -188.033
> Disper. corr. step 50: -48.4547, step 50: -193.09
> Coulomb (SR) step 50: -1821.5, step 50: -14751.3
> Potential step 50: -749.035, step 50: -14825.8
> Kinetic En. step 50: 4695.12, step 50: 3702.53
> Total Energy step 50: 3946.09, step 50: -11123.3
> Temperature step 50: 402.631, step 50: 317.511
> Pressure (bar) step 50: 35297.3, step 50: -836.341
> Box-X step 50: 6.85856, step 50: 4.68882
> Box-Y step 50: 3.81751, step 50: 2.34589
> Box-Z step 50: 3.93067, step 50: 2.34792
> Volume step 50: 102.915, step 50: 25.8259
> Density (SI) step 50: 174.458, step 50: 695.211
> pV step 50: 218762, step 50: -1300.74
> Vir-XX step 50: -34277.1, step 50: 1588.65
> Vir-XY step 50: -19728.5, step 50: 108.132
> Vir-XZ step 50: -14866.4, step 50: 14.285
> Vir-YX step 50: -20282.2, step 50: 108.132
> Vir-YY step 50: -185709, step 50: 2157.54
> Vir-YZ step 50: -102950, step 50: -196.053
> Vir-ZX step 50: -15598.4, step 50: 14.285
> Vir-ZY step 50: -102762, step 50: -196.053
> Vir-ZZ step 50: -103462, step 50: 1907.44
> Pres-XX (bar) step 50: 11541.9, step 50: -445.503
> Pres-XY (bar) step 50: 6367.62, step 50: -183.086
> Pres-XZ (bar) step 50: 4787.13, step 50: -18.5143
> Pres-YX (bar) step 50: 6546.3, step 50: -183.086
> Pres-YY (bar) step 50: 60435.8, step 50: -1150.69
> Pres-YZ (bar) step 50: 33194.4, step 50: 289.951
> Pres-ZX (bar) step 50: 5023.35, step 50: -18.5143
> Pres-ZY (bar) step 50: 33133.8, step 50: 289.951
> Pres-ZZ (bar) step 50: 33914.2, step 50: -912.829
> #Surf*SurfTen step 50: -8154.99, step 50: -269.382
> Pcoupl-Mu-YY step 50: 1.00103, step 50: 0.999981
> Mu-X step 50: 3.81572, step 50: 6.43379
> Mu-Y step 50: -28.1838, step 50: -29.6693
> Mu-Z step 50: -17.7016, step 50: -12.1778
> T-MeOH step 50: 448.795, step 50: 306.87
> T-SOL step 50: 348.773, step 50: 329.927
> Lamb-MeOH step 50: 0.997024, step 50: 0.999745
> .
> .
> .
> Reading frame 10 time 0.500
> End of file on ener.edr but not on ener2.edr
> ***************************************************************************************************************
>
>
> I'm sorry for posting such a big output here, but I had no idea what to
> select, and it is really starting to look like a not-yet documented issue,
> at least for the gromacs+pgi combination.
>
> Anyone with more experience on gmxcheck can help? Or did someone any time
> found this or a problem like this?
>
> Thanks a lot for all help already provided. :D
>
> Jones
> _______________________________________________
> gmx-users mailing list gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the www
> interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
--
David.
________________________________________________________________________
David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596, 75124 Uppsala, Sweden
phone: 46 18 471 4205 fax: 46 18 511 755
spoel at xray.bmc.uu.se spoel at gromacs.org http://folding.bmc.uu.se
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
More information about the gromacs.org_gmx-users
mailing list