Ahh, I appreciate this explanation. This is a good thing to keep in mind. I read up on how functions like `top` measure the processor load. Since the OS cannot distinguish between physical and virtual cores, `top` does not provide an accurate measurement, hence, I should not be using it as an absolute judge of the load on my system.
Thank you, Namu On Fri, Apr 10, 2015 at 3:40 PM, Damien <dam...@khubla.com> wrote: > Namu, > > Hyperthreads aren't real cores, they're really just another hardware > thread (two per physical core). You have two CPUs with 6 physical cores > each. If you start 2 simulations, each with 6 MPI processes running, your > 12 physical cores will each be running at 100%. Adding another simulation > with another 6 MPI processes will oversubscribe your physical cores (you're > asking for 150%), which is why you're still seeing 12 processors at 100%, > and everything else very low. Your physical cores are switching hardware > threads, but each core can't go any faster. Hyperthreads only help when > your software doesn't load a core to 100%. Then the other hyperthread on > that core can switch in and use leftover capacity. Hardware thread > switching is much faster than software thread switching, which is why it's > there. > > Most simulation software will load cores to 100% (even if it doesn't use > that 100% wisely, which is a whole other flame war) and hyperthreading > doesn't help you. > > Damien > > On 2015-04-10 2:22 PM, namu patel wrote: > > Hello All, > > I am using OpenMPI 1.8.4 on my workstation with 2 CPUs, each with 12 > processors (6 with hyper-threading). When I run simulations using mpiexec, > I'm noticing a strange performance issue. If I run two simulations, each > with 6 processors, then everything is fine and 12 processors are under 100% > load. When I start a 3rd simulation with 6 processors, I notice throttling > in all 3 simulations and only 12 processors are at 100%, the rest are below > 10%. My guess is that somehow the processes from the 3rd simulations are > doubling onto the already busy processors. How can I be certain that this > is the case? > > Thanks, > namu > > > _______________________________________________ > users mailing listus...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.open-mpi.org/community/lists/users/2015/04/26671.php > > > > _______________________________________________ > users mailing list > us...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.open-mpi.org/community/lists/users/2015/04/26673.php >
