Recent Changes - Search:


edit SideBar


Parallel ROOT processing

ROOT implements a system called PROOF for providing parallel computation ability to the interactive shell. Batch-like parallel functionality is achieved by the ability to start a job interactively and then disconnect from the interactive session, with the ability to reconnect/check progress later.

xrootd and proofd (now deprecated) handle the communication between nodes.

The question arises of how to integrate this system with existing scheduling/load balancing. Condor has a module called COD (Computing on Demand), which allows certain users immediate access to certain computational resources, when they request it. Apparently PROOF supports this, although I'm not finding documentation online. Admittedly I haven't got the PROOF distribution yet and looked there, but my hard drive is toast with many bad sectors, making such efforts inconvenient. In this regard, here is a sketch of somebody else's setup: PROOF at UW CMS Tier-2. Example code utilizing the described setup is here.

There will be some restrictions from the condor side too. From the condor manual: "Regardless of how the COD application's ClassAd is defined, the application's executable and input data must be pre-staged at the node. This is a current limitation of Condor's support for COD that will eventually go away. For now, there is no mechanism to transfer files for a COD application, and all I/O must be performed locally or onto a network file system that is accessible by a node."

Edit - History - Print - Recent Changes - Search
Page last modified on March 18, 2009, at 11:47 AM