LIVEwire News Briefs:

Intel Awards Seven Universities Research Grants At SC2004

Dedication to developing innovative ideas for emerging technologies has earned seven leading U.S. research universities grants from Intel Corp. The announcement was made at SC2004, the world's leading conference on high- performance computing, networking and storage.

The grants were made through the Intel Advanced Computing Center, which is dedicated to accelerating innovation in mainstream, volume computer technologies. The grants, totaling $600,000, include those awarded to international universities. These grants will be announced at a later date.

American universities receiving grants were the University of Illinois at Urbana-Champaign (UIUC), Purdue University, the University of Arizona, the University of Wisconsin, Princeton University, The Ohio State University and the University of Florida.

"The Intel Advanced Computing Center is focused on accelerating innovation in mainstream, or volume, computer technologies by working with leading commercial and academic partners to develop technologies once limited to proprietary high-performance computing systems," said center director Wilfred Pinfold.

UIUC was a double recipient. For a project on speculative multithreading for software development, Intel is working with a group led by Josep Torrellas, UIUC professor and Willett Faculty Scholar. Their work will lead to enhancing software productivity and enable multithreaded codes with coarsely tuned synchronization to help deliver competitive performance.

A second grant awarded to the University of Illinois was for the exploration of parallel file systems and software on Itanium processor-based clusters, a project led by Prof. Wen-Mei Hwu.

Purdue University's grant is in the area of sparse solvers. Under the guidance of Ahmed Sameh, professor of computer science on the main campus in Indianna, Intel and Purdue were able to create the SPIKE Algorithm which is able to solve banded linear systems, dense or sparse, in the range of applications with, depending on the needs, direct or iterative solvers. The algorithm also is effective for matrices for nanostructures and strong for SMP and MPI performance. The goal is for the algorithm to help fight back in areas infamously bad for cluster architecture and to help support advanced application areas.

Intel also awarded a grant to the University of Arizona for work on "Interconnection Network for Scalable Systems." The project, through the cooperation of Prof. Ahmed Louri of the Electrical and Computer Engineering Department, explores the application of optical interconnects for designing scalable interconnection networks for building large parallel systems. Goals of the project include achieving scalable bandwidth, significant reduction in remote latency, considerably reduced network costs and large processor counts.

The University of Wisconsin grant is for work on "Scalable Performance Instrumentation," tools that are used to help unify the daemons for multithreaded applications, improve MPI support and help in the handling of fork and exec calls. Led by Prof. Barton Miller of the Computer Sciences Department, the project's goal is to develop instruments that use automatic instrumentation and will search for performance problems.

Other grants announced:

• Princeton University, "Toward Supercomputer Performance for Advanced Bit Manipulations on Commercial Microprocessors," Ruby Lee, Forrest G. Hamrick professor.
• The Ohio State University, "Advanced Message Passing Algorithms for RDMA- enabled Interconnects," Dhabaleswar K. Panda, professor.
• University of Florida in the area of "Dynamic Provisioning," Sanjay Ranka and Sartaj Sahni, professors.

Further information about the Intel ACC is available at http://www.intel.com/technology/systems/acc.

Intel, the world's largest chip maker, is also a leading manufacturer of computer, networking and communications products. Additional information about Intel is available at http://www.intel.com/pressroom.