Filter löschen
Filter löschen

Increase number of parallel pools coworker by adding hardware?

3 Ansichten (letzte 30 Tage)
Hi, on my PC with Windows10, i9-10900K, 32GB RAM, i can launch max. 10 coworkers, which is the number I can reach on my machine, no matter what number i setup. But during processing data, my CPU get max. ca. 50% load, which I don't understand. So I am wondering whether it helps to add more RAM (32->64 GB) to increase the number of coworkers?
Does anyone know what is the relationship between CPU/RAM setups to number of coworkers?
Is there any suggestion for which CPU with how much RAM to get max. number of coworkers?
Thanks
Song

Akzeptierte Antwort

Walter Roberson
Walter Roberson am 15 Okt. 2021
Check your Parallel Preferences
And check the number of workers permitted in your "local" cluster; https://www.mathworks.com/help/parallel-computing/discover-clusters-and-use-cluster-profiles.html
The default permitted is the number of true cores you have -- excluding any effect for hyperthreading. (Hyperthreads that are not sometimes waiting for resources, such as waiting for disk or I/O, can end up making the overall execution slower)
Adding more RAM will not affect the number of workers permitted.
On the other hand, you should expect that workers will take a mininum of 2 Gb each, for the MATLAB image, plus the size of your data.
You should double check: is that 50% that you are seeing 50% of the nominal rate if you were using hyperthreads? Because 50% load measured against hypothetical hyperthreads means that your physical cores are at full load.

Weitere Antworten (0)

Kategorien

Mehr zu Parallel Computing Fundamentals finden Sie in Help Center und File Exchange

Produkte


Version

R2019b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by