- Inside UNM
Dorian Arnold, assistant professor in UNM's Computer Science Department is a member of a research team that has been recognized with a 2011 R&D; 100 Award. The award is for STAT, the stack trace analysis tool, which is being developed with researchers from Lawrence Livermore National Laboratory, (LLNL), the University of Wisconsin and UNM. The R&D; 100 awards are widely recognized as the "Oscars of Innovation." They celebrate the top technology products of the year.
STAT was designed to enable scalable debugging techniques for extremely large applications- that means applications that run on supercomputers comprised of hundreds of thousands of processor or more. When a researcher is running a complex problem such as climate modeling code on a supercomputer, a small snag in a few of the processors can cause serious problems.
STAT helps researchers to isolate problems in their application code so that debugging or performance analysis software can be used to diagnose and fix the problem. Or if a problem eludes the debugging software, a researcher can put the problematic source code into a folder, and send it to the code originator for debugging. It's a relatively simple way to save researchers time in finding those problematic needles that cause their programs to misbehave or run slowly in the haystack of large, complex code.
Arnold began working on the problem when he spent a summer visiting LLNL as a student scholar. He worked with his collaborators to design "lightweight debugging mechanisms", developed the first STAT prototype and published a research article documenting this work. Since that point, the research team has generated several publications, based on this tool and its groundbreaking concepts. STAT has since been elevated from "research quality" to a production tool with a myriad of features that make it easy for software developers to use. This is an open source tool and is available on the STAT website. If you want to know more about the tool, here's a helpful video.
At UNM Arnold teaches Computer Sciences courses primarily in Operating and Distributed Systems. His research focuses on high-performance computing and the performance and reliability of large scale distributed systems. He is co-director of the Scalable Systems Laboratory in the CS department where his students' research topics like scalable computer infrastructures, autonomous computer systems and computer system reliability.
Students interested in working with Arnold can visit his website.
Media contact: Karen Wentworth (505) 277-5627; e-mail: email@example.com