[gmx-users] Gromacs using MKL with Intel 11.1 compilers
gmx3 at hotmail.com
Fri Sep 18 08:59:16 CEST 2009
I don't know if this is worth all the troube.
Gromacs performance is not very dependent on FFT speed (unless you
choose your parameters badly). FFTW is faster than anything else
on most platforms (often including hardware vendor specific FFT's).
Simply sticking to FFTW will probably never result in a significant
From: darth.vasya at gmail.com
Date: Fri, 18 Sep 2009 09:08:48 +0400
Subject: Re: [gmx-users] Gromacs using MKL with Intel 11.1 compilers
To: cousins at umit.maine.edu; gmx-users at gromacs.org
Anybody have any real-world comparisons of using MKL vs. FFTW3?
Axel Kohlmeyer suggests the following (April 2008):
"(I've) found that Intel is now turning on multithreading without any
warning and that is for example why their FFT seems faster than
FFTW. For serial and OpenMP based runs this makes no difference (in
fact the multi-threaded FFT helps), but if you run MPI locally, you
actually lose performance. Also if you use the 'numactl' tool on linux
to bind a job to a specific cpu core, MKL will still try to use all
available cores (and slow down badly). The cleanest way of avoiding
this mess is to either link with
-lmkl_intel_lp64 -lmkl_sequential -lmkl_core (on 64-bit: x86_64, ia64)
-lmkl_intel -lmkl_sequential -lmkl_core (on 32-bit, i.e. ia32 )
or edit the libmkl_'platform'.a file (I'm using now a file libmkl10.a with:
GROUP (libmkl_intel_lp64.a libmkl_sequential.a libmkl_core.a)
It works like a charm".
So, this might contribute to your problem. Please tell us if Axel's suggestion works for you!
Express yourself instantly with MSN Messenger! Download today it's FREE!
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the gromacs.org_gmx-users