parpool problems: fewer running processes than workers
1 Ansicht (letzte 30 Tage)
Ältere Kommentare anzeigen
I'm currently trying to debug a problem with my parallel computing toolbox setup. I'm running MATLAB 2016a in a CentOS virtual machine, and have assigned the guest machine 22 cores. My host machine has 88 cores.
I have started a parpool with 22 workers, and if I terminate my job and look at the pool properties, it does, in fact, indicate that I am running 22 workers:
p = gcp('nocreate')
p =
Pool with properties:
Connected: true
NumWorkers: 22
Cluster: local
AttachedFiles: {}
IdleTimeout: 30 minute(s) (30 minutes remaining)
SpmdEnabled: true
Here is the problem: If I look at 'htop' in the guest, only 12 worker processes will be running at a time, as shown in this screenshot:
I can confirm that my guest vm IS utilizing 22 different cores on the host machine, but will only fully utilize 12 of those cores at a time.
Any suggestions?
1 Kommentar
Thomas Ibbotson
am 27 Jul. 2016
I think we'd need to see what code you are running on the pool to have any chance of figuring this out. Can you provide that?
You clearly have plenty of workers, in fact I count 39 in your screenshot, so you should be able to use them all. The fact you're only getting 12 is a little suspicious because that's the default number of workers Parallel Computing Toolbox uses for pools.
Antworten (0)
Siehe auch
Kategorien
Mehr zu Parallel Computing Fundamentals finden Sie in Help Center und File Exchange
Produkte
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!