Star of Arkansas
The Star of Arkansas Goes Into Full Production!
The Star of Arkansas was made available for general production use on Wednesday, May 21, 2008.Star is approximately eight times more capable than our first supercomputer, Red Diamond. Up to 93% of the available compute nodes are being utilized by applications from computational researchers from the University of Arkansas and jobs are waiting for resources. Jobs have been submitted by users from four different departments from two colleges on campus. The amount of computing accomplished on the Star of Arkansas during its first five days of production operation exceeds the amount of computing that can be accomplished on Red Diamond during an entire month!
On May 21, 2008, the Star of Arkansas ran the largest parallel user application ever to have run on a supercomputer in Arkansas. Sergey Lisenkov, Research Associate in the Department of Physics, ran a job across 1240 cores before we opened the queue to general submissions. His job used 15224.84 seconds of user time and 1.360000 seconds of system time. This job would have taken at least one year to complete as a serial application, but it completed overnight on the Star of Arkansas!
Another user, Stella Huff, Ph.D. candidate in the Department of Chemistry and Biochemistry, has run applications over a three day period that would have taken three weeks to complete on Red Diamond.
For more information about the Star of Arkansas and access to High Performance Computing resources at the University of Arkansas please see http://hpc.uark.edu/, or send email to the Senior Linux Cluster Administrator, Jeff Pummill jpummil@uark.edu
Acxiom Research
Acxiom Corporation informed the Computer Science and Computer Engineering Department today that they will provide $40K continuation funding in 2008-2009 for two projects started in June 2007. Drs. Wingning Li and Craig Thompson and graduate research assistants Reid Phillips, Wesley Deneke, and Joshua Eno staff the projects. Acxiom is a leader in customer data integration. They use grids of thousands of PCs to process huge data sets for their customers. Each task is specified as a workflow that can take hundreds of files as input and can process the inputs to organize, improve, and augment data quality. The two projects contribute as follows:
* Layout Inferencing Project - Presently, Acxiom receives thousands of files a month, each consisting of structured records but in varying formats with a wide range of contents. Humans have to discover and specify file formats. This research project samples files statistically augmented with a partial specification of common content types and suggests a file layout format, partially automating the task that humans perform manually now.
* Domain Specific Modeling Language - For each customer and collection of files, Acxiom associates currently manually define a complex workflow to improve the data quality, merge, and augment the contents. The DSML research project aims to define a high level intent language that associates can use to specify the goal of the workflow. A problem solver automatically generate a workflow that defines how to achieve the goal.
ASTA Grant Awarded
Craig Thompson (CSCE) and Ron Hardy (Director of Program Facilitation, EAST Initiative) have received a $26,600 summer grant from Arkansas Science and Technology Authority to introduce 3D virtual world technology to high school students in the EAST program. This summer, six EAST students will join Thompson's CSCE special topics course Healthcare Logistics in a Virtual World to work alongside undergraduates and graduates.
The project is aligned with the goals of the Center for Innovation in Healthcare Logistics (CIHL) and the RFID Research Center and also working with the Center for Advanced Spatial Technology (CAST). The course is modeling logistics flow in a hospital using the Second Life virtual world platform. Teleport to "University of Arkansas" island to visit the virtual hospital. So far we have developed smart pill bottles that only the owner can open and that know their pill count, smart shelves that know when to reorder, a restocking robot, wheelchairs that can follow way points, and virtual RFID. A description of the project and instructions for visiting the island are here: http://www.csce.uark.edu/~cwt/COURSES/2008-05--SL/2008-05--Healthcare-Logistics-in-Second-Life--Craig-Thompson.doc
