Problem with cassandra_gfortran_openMP
- mostafa.razavi
-
Topic Author
- Offline
- Junior Member
-
Less
More
- Posts: 22
- Thank you received: 1
7 years 5 months ago #256
by mostafa.razavi
Hello everyone,
I have problem using parallel Cassanda.
I have a 8 core machine, CentOS 6.5 and I use gfortran compiler. I've been able to run a serial simulation using Cassandra without any problem, but when I try to run a parallel simulation only one CPU core is used.
This is how I compiled the code:
1- In src diectory I run:
make -f Makefile.gfortran.openMP
2- I copy the created executable file to Examples/GEMC/Methane directory, and I run:
export OMP_NUM_THREADS=8
./cassandra_gfortran_openMP.exe methane.inp
I would appreciate if anybody can help me why I cannot use more than one core. Also, for a system of 1200 interaction sites, is it even worth it to use multiple cores? Does anybody have a benchmark on this?
Thank you,
-Mostafa
I have problem using parallel Cassanda.
I have a 8 core machine, CentOS 6.5 and I use gfortran compiler. I've been able to run a serial simulation using Cassandra without any problem, but when I try to run a parallel simulation only one CPU core is used.
This is how I compiled the code:
1- In src diectory I run:
make -f Makefile.gfortran.openMP
2- I copy the created executable file to Examples/GEMC/Methane directory, and I run:
export OMP_NUM_THREADS=8
./cassandra_gfortran_openMP.exe methane.inp
I would appreciate if anybody can help me why I cannot use more than one core. Also, for a system of 1200 interaction sites, is it even worth it to use multiple cores? Does anybody have a benchmark on this?
Thank you,
-Mostafa
Please Log in to join the conversation.
- ryangmullen
-
- Offline
- Administrator
-
Less
More
- Posts: 124
- Karma: 4
- Thank you received: 24
7 years 5 months ago #258
by ryangmullen
Hi Mostafa,
Eliseo has run some benchmarking results that we will be publishing soon. If your 1200 sites are LJ butane, I don't think you'll see much of a performance boost from running in parallel. You'll see a bigger boost if you are simulating longer molecules (e.g. octane, which benefits from the parallelized CBMC routines) or systems with electrostatics (e.g. water, which benefits from the parallelized Ewald routines).
Are you running with your jobs with the
# Pair_Energy
true
option? That will speed up your sims by storing the energy between pairs of molecules in active memory.
Ryan
Eliseo has run some benchmarking results that we will be publishing soon. If your 1200 sites are LJ butane, I don't think you'll see much of a performance boost from running in parallel. You'll see a bigger boost if you are simulating longer molecules (e.g. octane, which benefits from the parallelized CBMC routines) or systems with electrostatics (e.g. water, which benefits from the parallelized Ewald routines).
Are you running with your jobs with the
# Pair_Energy
true
option? That will speed up your sims by storing the energy between pairs of molecules in active memory.
Ryan
Please Log in to join the conversation.
Time to create page: 0.097 seconds