MATLAB: Is there a limit on how much data I can transfer using the Distributed Computing Toolbox 2.0.1 (R2006a)

50mbdctMATLAB Parallel ServermdceParallel Computing Toolbox

I am using the FileDependencies property of my distributed job to send a data file that is over 50 MB in size. However, I receive memory-related errors similar to the following when doing so:

 ??? Not enough Java memory to construct Java arguments

Best Answer

This bug has been fixed in Release 2007a (R2007a). For previous product releases, read below for any possible workarounds:

The "Troubleshooting" section of the Distributed Computing Toolbox documentation discusses some default limitations on how much data can be transferred. The default limit is approximately 50 MB, determined by the Java Virtual Machine (JVM) memory allocation limit. To work around this, follow the suggestions outlined in the documentation. For more information, type the following command at the MATLAB command prompt:

docsearch('JOB_MANAGER_MAXIMUM_MEMORY')

Best Answer

Related Solutions

MATLAB: Do I receive an empty data set when I try to transfer huge data ( in order of 1.5 GB) between clients and workers while using Distributed Computing Toolbox 3.2 (R2007b)

MATLAB: Do I get an error initializing MPI when validating the cluster

Related Question