JVM Out of memeory error when using multiple cluster profiles
3 vues (au cours des 30 derniers jours)
Afficher commentaires plus anciens
Hello,
I am working on a project which involves loading a large number of images, extracting certain features and saving these features to the disk.
I am using 64 bit Matlab R2013a on a linux server with 64 cores.
I created 4 copies of the local(default) cluster profile, so that I have 5 cluster profiles which can create 60 workers.
I am running into JVM memory problems when using all the cluster profiles parallely. I am able to run 2 cluster profiles simulaneously, but when I try to start a 3rd cluster profile parallely, matlab crashes with the follwoing error:
---------------------------------------- Error -------------------------------------------------------
Stage: SPMD job test (createCommunicatingJob)
Status: Failed
Description:The job errored or did not reach state finished.
Command Line Output:(none)
Error Report:(none)
Debug Log:
LOG FILE OUTPUT:
[7]Opening log file: /home/nchavali/java.log.21286
[2]Opening log file: /home/nchavali/java.log.21291
[10]Opening log file: /home/nchavali/java.log.21283
[9]#
[9]# A fatal error has been detected by the Java Runtime Environment:
[9]#
[9]# java.lang.OutOfMemoryError: Cannot create GC thread. Out of system resources.
[9]#
[9]# Internal Error (gcTaskThread.cpp:38), pid=21284, tid=139962136192768
[9]# Error: Cannot create GC thread. Out of system resources.
[9]#
[9]# JRE version: 6.0_17-b04
[9]# Java VM: Java HotSpot™ 64-Bit Server VM (14.3-b01 mixed mode linux-amd64 )
------------------------------------------------------------------
My code does not involve transfer of a large amount of data from client to workers.. My workers load image from a location specified in a variable, and save data into disk. And each worker saves data in different locations.
Can someone please help me is solving this problem.
This is a snippet of code that calls matlab workers:
clear;
load('foldernames.mat');
%loads folderNames variable
matlabpool open LocalProfile2;
parfor i=1:12
cffinal(folderNames(i,:));
end
exit;
0 commentaires
Réponses (1)
Voir également
Catégories
En savoir plus sur Parallel for-Loops (parfor) dans Help Center et File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!