Re: compiler flags.

From: Brian Bennion (brian_at_youkai.llnl.gov)
Date: Wed Jul 14 2004 - 23:41:22 CDT

Hi Gengbin and Jim and others

I just finished a compile of tcl, fftw, charm++ and namd all using the -xN
-O3 icc8.0 compiler flags.

Compared to the previous executable I get about a 13% gain in speed for
the same exact namd.config. I did notice that memory usage more than
doubled and this is troubling.
Some log snippets are below. Every second line is the same run....sorry
for the confusion. The first line is the newest binary.
any ideas what I might have done (wrong)?
Regards
Brian
Info: Entering startup phase 0 with 15024 kB of memory in use | 00: Info:
Entering startup phase 0 with 7139 kB of memory in
Info: Entering startup phase 1 with 15024 kB of memory in use | 00: Info:
Entering startup phase 1 with 7139 kB of memory in
Info: Entering startup phase 2 with 17456 kB of memory in use | 00: Info:
Entering startup phase 2 with 9572 kB of memory in
Info: Entering startup phase 3 with 17699 kB of memory in use | 00: Info:
Entering startup phase 3 with 9816 kB of memory in
Info: PATCH GRID IS 5 (PERIODIC) BY 5 (PERIODIC) BY 4 (PERIOD | 00: Info:
PATCH GRID IS 5 (PERIODIC) BY 5 (PERIODIC) BY 4 (PE
Info: REMOVING COM VELOCITY -0.000221351 -0.00606466 0.000429 | 00: Info:
REMOVING COM VELOCITY -0.000221351 -0.00606466 0.00
Info: LARGEST PATCH (33) HAS 402 ATOMS | 00: Info:
LARGEST PATCH (33) HAS 402 ATOMS
Info: Entering startup phase 4 with 22092 kB of memory in use | 00: Info:
Entering startup phase 4 with 14208 kB of memory in
Info: PME using 40 and 40 processors for FFT and reciprocal s | 00: Info:
PME using 40 and 40 processors for FFT and reciproc
Creating Strategy 4 | 00:
Creating Strategy 4
Creating Strategy 4 | 00:
Creating Strategy 4
Info: PME GRID LOCATIONS: 1 3 5 7 9 11 13 14 15 17 ... | 00: Info:
PME GRID LOCATIONS: 1 3 5 7 9 11 13 14 15 17 ...
Info: PME TRANS LOCATIONS: 1 3 5 7 9 11 13 14 15 17 ... | 00: Info:
PME TRANS LOCATIONS: 1 3 5 7 9 11 13 14 15 17 ...
Info: Entering startup phase 5 with 22098 kB of memory in use | 00: Info:
Entering startup phase 5 with 14215 kB of memory in
Info: Entering startup phase 6 with 18266 kB of memory in use | 00: Info:
Entering startup phase 6 with 10382 kB of memory in
Info: Entering startup phase 7 with 18270 kB of memory in use | 00: Info:
Entering startup phase 7 with 10387 kB of memory in
Info: COULOMB TABLE R-SQUARED SPACING: 0.0625 | 00: Info:
COULOMB TABLE R-SQUARED SPACING: 0.0625
Info: COULOMB TABLE SIZE: 705 POINTS | 00: Info:
COULOMB TABLE SIZE: 705 POINTS

On Tue, 13 Jul 2004, Gengbin Zheng wrote:

> I guess you just need these switches for NAMD. If you also want them for
> charm, you can add them to build script like:
> ./build charm++ net-linux -O3 ...
>
> Gengbin
>
> Brian Bennion wrote:
>
> >okay. now that I am going to do some testing...where do I put these
> >switches if i use the build script in charm?
> >
> >
> >On Tue, 13 Jul 2004, Gengbin Zheng wrote:
> >
> >
> >
> >>yes, i was using the default. No luck with it. -O1 or -O0 may work. I am
> >>not aware of -xWNP flags.
> >>
> >>Gengbin
> >>
> >>Brian Bennion wrote:
> >>
> >>
> >>
> >>>So you where using the default -O2 optimization and not and -xWNP flags?
> >>>
> >>>
> >>>
> >>>On Tue, 13 Jul 2004, Gengbin Zheng wrote:
> >>>
> >>>
> >>>
> >>>
> >>>
> >>>>unforturantely so far I have no luck with Intel 8 compiler. Get rid of
> >>>>all optimization flags let it work as I remember, however, very very bad
> >>>>performance. I have not tried to selectively get rid of some of the
> >>>>optimization flags and see the impact of the performance though. For
> >>>>example, remove some of the -f flags, or lower the optimization level.
> >>>>
> >>>>Gengbin
> >>>>
> >>>>Brian Bennion wrote:
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>>Hi Gengbin,
> >>>>>
> >>>>>I know I sent an email through the namd-l list, but I really could use a
> >>>>>quick answer either way. I need to get NAMD and charm running at their
> >>>>>best and soon. If I need to experiment with different options then I can,
> >>>>>but I really don't want to reinvent the wheel here.
> >>>>>
> >>>>>Tell me what you think.
> >>>>>brian
> >>>>>
> >>>>>
> >>>>>*****************************************************************
> >>>>>**Brian Bennion, Ph.D. **
> >>>>>**Computational and Systems Biology Division **
> >>>>>**Biology and Biotechnology Research Program **
> >>>>>**Lawrence Livermore National Laboratory **
> >>>>>**P.O. Box 808, L-448 bennion1_at_llnl.gov **
> >>>>>**7000 East Avenue phone: (925) 422-5722 **
> >>>>>**Livermore, CA 94550 fax: (925) 424-6605 **
> >>>>>*****************************************************************
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>
> >>>>
> >>>*****************************************************************
> >>>**Brian Bennion, Ph.D. **
> >>>**Computational and Systems Biology Division **
> >>>**Biology and Biotechnology Research Program **
> >>>**Lawrence Livermore National Laboratory **
> >>>**P.O. Box 808, L-448 bennion1_at_llnl.gov **
> >>>**7000 East Avenue phone: (925) 422-5722 **
> >>>**Livermore, CA 94550 fax: (925) 424-6605 **
> >>>*****************************************************************
> >>>
> >>>
> >>>
> >>>
> >>
> >>
> >
> >*****************************************************************
> >**Brian Bennion, Ph.D. **
> >**Computational and Systems Biology Division **
> >**Biology and Biotechnology Research Program **
> >**Lawrence Livermore National Laboratory **
> >**P.O. Box 808, L-448 bennion1_at_llnl.gov **
> >**7000 East Avenue phone: (925) 422-5722 **
> >**Livermore, CA 94550 fax: (925) 424-6605 **
> >*****************************************************************
> >
> >
>
>

*****************************************************************
**Brian Bennion, Ph.D. **
**Computational and Systems Biology Division **
**Biology and Biotechnology Research Program **
**Lawrence Livermore National Laboratory **
**P.O. Box 808, L-448 bennion1_at_llnl.gov **
**7000 East Avenue phone: (925) 422-5722 **
**Livermore, CA 94550 fax: (925) 424-6605 **
*****************************************************************

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:38:47 CST