[gmx-users] compiling on different architecture than the compute nodes architecture
mark.j.abraham at gmail.com
Fri Feb 8 15:01:36 CET 2013
On Thu, Feb 7, 2013 at 10:00 AM, Broadbent, Richard <
richard.broadbent09 at imperial.ac.uk> wrote:
> Dear Silard,
> cmake -DGMX_CPU_ACCELERATION=AVX_256 -DGMX_DEFAULT_SUFFIX=OFF
> -DGMX_BINARY_SUFFIX="_avx" -DGMX_OPENMP=ON -DGMX_MPI=ON -DGMX_DOUBLE=ON
> -DGMX_GPU=OFF -DGMX_PREFER_STATIC_LIBS=ON -DGMX_FFT_LIBRARY=mkl
> Compiled, linked and installed without warning using intel-suite/2013,
> mpi/intel-3.1, and cmake/2.8.9 I was compiling release-4-6 out of the git
> repository ( 06935659a3125d9b44b84c57e67cb6788fda42c1 ). The job using
> that executable is still in the queue, however, another one built using
> the sse2 kernels ran a minimisation on a single core without problems.
> How much faster fftw3 is than mkl for gromacs, is the difference likely to
> be on the scale of 1-2% or 10%?
The former, I would expect. Naturally it depends on your hardware, how you
compiled FFTW, and whether/how you use PME in GROMACS. If your PME-PP load
balance has too much work for the PP nodes, then FFT performance doesn't
matter. The .log file prints lots of analysis about this, and mdrun tries
by default to do what it can to balance.
Recent versions of icc and MKL do make compiling for MKL much easier, so
we'll probably add that for 5.0
> On 07/02/2013 02:17, "Szilárd Páll" <szilard.pall at cbr.su.se> wrote:
> >On Wed, Feb 6, 2013 at 6:03 PM, Richard Broadbent <
> >richard.broadbent09 at imperial.ac.uk> wrote:
> >> Dear All,
> >> I would like to compile gromacs 4.6 to run with the correct acceleration
> >> on the compute nodes on our local cluster. Some of the nodes have intel
> >> sandy-bridge whilst others only have sse4.1 and some (including the
> >> and single core job nodes) are still stuck on ssse3 (gmx would use sse2
> >> acceleration here).
> >> Installing several versions is not a problem however, I'm not sure how
> >> make cmake build a version of the code that is not using the
> >> for the system on which the code is being compiled. Restrictions on job
> >> sizes makes running the compilation on the sandy-bridge nodes almost
> >> impossible. Can anyone let me know which flags cmake needs to enable
> >> avx-256 acceleration?
> >> my standard cmake line is:
> >> $ CC=mpiicc CXX=mpiicpc ; cmake -DGMX_OPENMP=ON -DGMX_MPI=ON
> >> -DGMX_DOUBLE=ON -DGMX_GPU=OFF -DGMX_PREFER_STATIC_LIBS=ON
> >> -DGMX_FFT_LIBRARY=mkl -DMKL_INCLUDE_DIR=$MKLROOT/**include
> >> -DMKL_LIBRARIES="$MKLROOT/lib/**intel64/libmkl_core.so;$**
> >> -DCMAKE_INSTALL_PREFIX=$HOME/**libs/gromacs ../
> >Note that MKL (without the fftw wrappers) is known to not work out of the
> >box. Making it work requires a fairly simple workaround described here:
> >Additionally, note that FFTW is in most cases faster than MKL (but if you
> >find the contrary do let us know).
> >> Thanks,
> >> Richard
> >> --
> >> gmx-users mailing list gmx-users at gromacs.org
> >> * 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
> >gmx-users mailing list gmx-users at gromacs.org
> >* Please search the archive at
> >http://www.gromacs.org/Support/Mailing_Lists/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/Support/Mailing_Lists
> gmx-users mailing list gmx-users at gromacs.org
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/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/Support/Mailing_Lists
More information about the gromacs.org_gmx-users