[gmx-users] Performance (GMX4.6.1): MPI vs Threads

[Thomas Schlesier]

Dear all,
if one performs a parallel calculation on a single node / computer with 
more than 1 core, is there a speed difference between MPI and Threads?

Problem is for gromacs 4.6.1 i'm facing problems to link it to OpenMpi. 
A serial version (without threads) worked, so i think i should be able 
to compile a version which supports threads.
Since we only run calculations on a single node on our cluster (no 
infini-band), i'm wondering if the programm with threads would be 
sufficient in our case.

Greetings
Thomas