Effacer les filtres
Effacer les filtres

Matlab 2017a very slow in linux

1 vue (au cours des 30 derniers jours)
Alan
Alan le 1 Déc 2017
Commenté : Walter Roberson le 1 Déc 2017
I have a Linux machine (CentOS 7) running Matlab 2017a that gets very sluggish after running a Simulink model. The machine has 8 cores, 32 GB of memory. Running the model uses up about 7 GB of memory. While examining the results of the model, I noticed that any command that I run in the command window either a) takes a few minutes to execute or b) quickly spits out the answer but then Matlab becomes unresponsive for a few minutes. I notice by watching the System Monitor while Matlab is out to lunch that it seems to be slowly allocating about 100 MB of memory and then quickly releasing it - over and over and over again. This behavior occurs even when I ask Matlab to do something simple, like
>> 2*2
I have been scouring these answers for something that resolves this issue. There is no network drive in my search path. I have no figures open. I have allocated 4 GB of java heap space. I am not sure if this is a java, Simulink, or Matlab problem. I thought that it was a display issue (running GEFORCE GTX TITAN X) but by eliminating plots and only asking Matlab to solve a simple arithmetic problem, I eliminated that possibility.
Thanks for your help.
  3 commentaires
Alan
Alan le 1 Déc 2017
I have the simulink model window still up, but that is about it besides the Matlab GUI and the Editor. No graphs are being plotted through Matlab or Simulink. When I give it a simple arithmetic problem (2*2), what fplot() is happening? It is something that running the simulink model with this much data is doing to the memory manager, because after starting Matlab it does not behave this way until running simulink. I can bring up the java runtime and run the gc and no change in behaviour.
Walter Roberson
Walter Roberson le 1 Déc 2017
You probably have no fplot() running -- I was demonstrating that in at least some cases, graphics keep processing in the background for complicated graphics, even after returning to the command line.

Connectez-vous pour commenter.

Réponses (0)

Catégories

En savoir plus sur Simulink Coder dans Help Center et File Exchange

Produits

Community Treasure Hunt

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

Start Hunting!

Translated by