[gmx-users] gromacs on itanium2

[David van der Spoel]

Marius Retegan wrote:
> Hi,
> 
> Thank you for the suggestions. I've managed to get the version 4.0.4
> compiled by using --enable-fortran and --disable-ia64-asm.
> Nevertheless some of the kernel tests still fail.

Did you use the new test set?

> 
> FAILED. Check files in kernel020
> FAILED. Check files in kernel120
> FAILED. Check files in kernel121
> FAILED. Check files in kernel122
> FAILED. Check files in kernel123
> FAILED. Check files in kernel124
> FAILED. Check files in kernel220
> FAILED. Check files in kernel221
> FAILED. Check files in kernel222
> FAILED. Check files in kernel223
> FAILED. Check files in kernel224
> FAILED. Check files in kernel320
> FAILED. Check files in kernel321
> FAILED. Check files in kernel322
> FAILED. Check files in kernel323
> FAILED. Check files in kernel324
> 16 out of 63 kernel tests FAILED
> 
> Is this due to the fact that I've disable the assembly loops for ia64?
> I also have some problems with the parallel version.
> 
> Testing acetonitrilRF . . . FAILED. Check files in acetonitrilRF
> Testing aminoacids . . . FAILED. Check files in aminoacids
> Testing argon . . . FAILED. Check files in argon
> Testing butane . . . PASSED but check mdp file differences
> Testing dec+water . . . PASSED but check mdp file differences
> Testing ethyleenglycol . . . PASSED but check mdp file differences
> Testing fe_test . . . PASSED but check mdp file differences
> Testing field . . . PASSED but check mdp file differences
> Testing nacl . . . PASSED but check mdp file differences
> Testing sw . . . FAILED. Check files in sw
> Testing tip4p . . . FAILED. Check files in tip4p
> Testing tip4pflex . . . PASSED but check mdp file differences
> Testing urea . . . FAILED. Check files in urea
> Testing water . . . FAILED. Check files in water
> 7 out of 14 complex tests FAILED
> 
> For the failed test I get the following error
> 
> -------------------------------------------------------
> Program mdrun_mpi_d, VERSION 4.0.4
> Source code file: domdec.c, line: 5444
> 
> Fatal error:
> ns type Simple is not supported with domain decomposition,
> use particle decomposition: mdrun -pd
> -------------------------------------------------------
> 
> Also in the gmxtest.pl script the option -np for the grompp should be removed.
> Finally, I was wondering if the use of mkl in place of fftw3 would
> improve performance.
> 
> Thanks,
> Marius
> 
> 
> 
> 
> On Tue, Mar 3, 2009 at 10:02 PM, Erik Lindahl <lindahl at cbr.su.se> wrote:
>> Hi,
>>
>> There have been some reports on newer ia64 processors being quite fast with
>> the Fortran kernels instead (even faster than asm!), so I would try that.
>>
>> This has to do with the brain-dead architecture on ia64. The asm kernels
>> were written for original itanium2 timings, but with the register rotation
>> loops and lack of dynamic instruction scheduling you would get lots of
>> pipeline stalls/bubbles if Intel ever changed their instruction latencies,
>> and that might have happend now.
>>
>> Considering the limited adoption of ia64 I don't think we're very likely to
>> keep updating those kernels - the first thing on the agenda right now will
>> rather be improved performance of x86 SSE kernels for Gromacs 4.1.
>>
>> Cheers,
>>
>> Erik
>>
>>
>> On Mar 3, 2009, at 8:14 PM, David van der Spoel wrote:
>>
>>> Marius Retegan wrote:
>>>> Hello
>>>> Since I was unable to get a working version for Gromacs 4.0.4 on a
>>>> Itanium 2 machine with the ia64 nonbonded kernel, I was wondering what
>>>> would be the lost in speed if I would disable the assembly loops?
>>> About a factor of two. What's the problem with compiling?
>>>
>>>> Thanks,
>>>> Marius
>>>> _______________________________________________
>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>>>> Please search the archive at http://www.gromacs.org/search before
>>>> posting!
>>>> 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 van der Spoel, Ph.D., Professor of Biology
>>> Molec. Biophys. group, Dept. of Cell & Molec. Biol., Uppsala University.
>>> Box 596, 75124 Uppsala, Sweden. Phone:  +46184714205. Fax: +4618511755.
>>> spoel at xray.bmc.uu.se    spoel at gromacs.org   http://folding.bmc.uu.se
>>> _______________________________________________
>>> gmx-users mailing list    gmx-users at gromacs.org
>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>>> Please search the archive at http://www.gromacs.org/search before posting!
>>> 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
>> _______________________________________________
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://www.gromacs.org/mailman/listinfo/gmx-users
>> Please search the archive at http://www.gromacs.org/search before posting!
>> Please don't post (un)subscribe requests to the list. Use thewww interface
>> or send it to gmx-users-request at gromacs.org.
>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> 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 van der Spoel, Ph.D., Professor of Biology
Molec. Biophys. group, Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205. Fax: +4618511755.
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://folding.bmc.uu.se