About the project

The project Xgrid@Stanford is run by Charles Parnot, and is hosted by Brian Kobilka's lab, in the Molecular and Cellular Physiology department, Stanford University.

We are currently running large calculations that need a lot of time on a computer, the purpose being to modelize the conformational changes of the beta 2 adrenergic receptor, and have a better understanding of its pharmacology (read more in the Project section). To have these calculations run faster, we are now running them on a cluster of more than 200 computers, where each computer actually belongs to a different person, connected to the internet somewhere in the world. This allows us to run a calculation in 1 week instead of 1 year!!

The idea is to take advantage of all these computers sitting idle at night (and often during the day too), that could run these calculations for us. But it also has to be secure, and the calculations should stop when the normal user needs its computer at its full speed. So basically, you want it to be just like a screensaver, only popping up when you don't need your computer.

For this, Apple has written a very simple and elegant program called Xgrid. The communications are completety secure, it does not take any space on the hard drive, has no access to personal files, and has no access to the System, so no risk of crash, bug, corrupted file and no piracy/privacy problem.

So help science go fast forward and make your computer happy: keep it busy even on its free time! Follow the installation instructions.

Latest News

June 25, 2008. The license for GridStuffer has now been released under the modified BSD license, instead of the more restrictive GPL license. The underlying GridEZ framework remains LGPL.

December 3, 2007. The Xgrid@Stanford project has now entirely switched its computation needs to the OpenMacGrid cluster. The agents from this cluster will be shared with other scientists that have submitted projects to MacResearch.

August 21, 2007. A bunch of our Goodies have been updated, with Xgrid FUSE version 0.3.1, GridStuffer version 0.4.5, and GridEZ.framework version 0.4.5.

May 5, 2007. First public release of Xgrid FUSE, version 0.2.0. Xgrid FUSE transforms an Xgrid controller into a file system. Accessing your controller is now just as easy as plugging an external hard drive, all of this thanks to the fantastic MacFUSE project.

April 26, 2007. New version of GridStuffer, now at 0.4.4, that fixes a (non-crashing) bug in the parsing of the input file.

April 19, 2007. Available on MacResearch, the second part of the Xgrid tutorials series was just posted. In this new instalment, I cover GridStuffer basics.

March 12, 2007. New version of xgridstatus, a command-line utility that brings to the terminal many features otherwise only accessible in Xgrid Admin. New features include more report options, such as job lists and statistics, as well as automatic reconnection after controller crashes. There are also a number of small bug fixes.

Also available are the versions 0.4.2 of GridStuffer and GridEZ.framework. One crasher fixed, and improved file downloads, all within GridEZ, are the main changes.

February 26, 2007. New versions of GridStuffer and GridEZ, both compiled as universal binaries that will run happily on either PowerPC or x86 macs. These two open-source projects are now developed in parallel, with GridEZ providing the core Xgrid functionality. GridStuffer features a lot of under-the-hood improvements, mostly coming from the new much-improved GridEZ framework. For interested developers, the framework also includes a new metajob class that allows the submission and retrieval of complex multipart jobs, with only a few lines of code. Check out this 30-line 30-minute tutorial to write your first Xgrid-enabled application.

January 8, 2007. MacResearch has started a series of tutorials on Xgrid, that I will be authoring. The first installment is online, and goes through some of the basics of distributed computing and Xgrid. It should be a good starting point for scientists curious about Xgrid, and interested in using distributed computing for their research.

July 22, 2006. Thanks to an unexpected post on digg and an unexpectedly high interest from digg users, the number of visits went from an average of 100 per day to a stunning 10,000 on July 18th, and 5,000 on July 19th. With a flow of new agents, the cluster is registering speeds close to 800 GHz this week.