Difference between revisions of "Enabling multicore jobs and jobs requesting large amounts of memory"

Revision as of 16:15, 21 June 2012

TO BE DONE

@@ Line 1: / Line 1: @@
-'''THIS ARTICLE IS WORK IN PROGRESS'''
+TO BE DONE
-This article describes the test implementation that enables grid jobs to use multiple cores on one (or more!) physical compute node(s) and which supports allocation of flexible amounts of memory to the jobs.
-The implementation is based on the UMD-1 version of the CreamCE, the batch system Torque and the Maui scheduler.
-'''''Disclaimer'''''
-Please note that the "production" grid systems at Nikhef currently do not support this implementation. Certain parts may work, others may not.