Jeff Gardner's Homepage

Scalable Computing Resources for UW Researchers

The Athena cluster is jointly owned and operated by the UW Physics and Astronomy departments, the Institute for Nuclear Theory (INT), and the Center for Experimental Nuclear Physics and Astrophysics (CENPA). It is comprised of 140 compute nodes linked together by a high speed InfiniBand network fabric. The theoretical peak performance of the system is over 10 TeraFlop/s (10 trillion floating point operations per second). All compute nodes are connected to 23.7 TB of shared storage.
The NSF TeraGrid is an open scientific discovery infrastructure funded by the NSF. It combines leadership-class resources at eleven partner sites to create an integrated, persistent, national computational resource. Grants of time is available to any U.S. researcher. Please contact Jeff if you would like to acquire free TeraGrid time.

The Cyberinfrastructure Tutor at NCSA contains an excellent set of introductory and intermediate web-based tutorials on a variety of topics in High Performance Computing. I highly recommend that you take some time to browse the offerings. You need to create an account with the system, but that is free. The account is also useful because it will keep track of your progress through each tutorial for you.
Supercomputing in Plain English is a web workshop series offered roughly once per year by the Oklahoma University Supercomputing Center for Education & Research. The last offering was in Spring 2009. Videos of the past year as well as powerpoint slides and exercised are available on their website. This is a very good introduction to basic issues in using and developing applications for HPC systems. It can move a little slowly at times, but the advantage of watching an archived video of the presentation is that you can fast-forward :).
HPC University is an effort sponsored primarily by the NSF to aggregate information on courses and fellowships that available for High Performance Computing. They are still compiling information, and I still think they need to work on it some more, but they do have a pretty complete listing of fellowship opportunities.
NSF TeraGrid Online Training also has links to general training in a number of technology areas as well as to TeraGrid specific learning resources.

The UW eScience Institute lists courses that are offered at UW that relate to scientific computing. The best general purpose offering is the AMATH 500A, which is planned to be offered yearly, probably in the spring. In Spring 2009, quite a number of Physics and Astronomy students took the class and found it worthwhile.

Click here for a list of lectures that Jeff has given while teaching classes on high performance computing.
Click here for a list of talks that Jeff has made on behalf of PSC and TeraGrid.
Click here for a list of list popular talks that Jeff has made on Astronomy and High Performance Computing to general audiences.

Ntropy is a framework that enables rapid development of highly scalable parallel implementations of any algorithm that uses a tree data structure. Initial target applications are for astrophysics and cosmology, where nearly all data fast reduction methods (for both simulated and observed data) use kD trees. Ntropy, however, is useful for any application that interacts with unordered data in an N-dimensional parameter space. (Ntropy homepage)
Our Astronomy in the Cloud project is evaluating cloud platforms for research computing on astrophysics data. Our current focus platform is Hadoop, although we will also eventually target Microsoft Dryad.