Parallel Performance Project

Professor Edward S. Davidson

The objective of our work is to develop and implement automated means of assessing the potential performance of applications on targeted supercomputers, high performance workstations and parallel computer systems, identifying the causes of degradations in delivered performance, and restructuring the application codes and their data structures so that they can achieve their full performance potential. For a more detailed description, please click here .

Project Members

The Parallel Performance Project is part of the ongoing research being performed at the Advanced Computer Architecture Laboratory in the Department of Electrical Engineering and Computer Science of the University of Michigan. Current project members include:

• Professor Edward S. Davidson
• Murali Annavaram - Computer Architecture and Compiler Interactions.
• Shih-Hao Hung
• Jude Rivers
• Mikhail Smelyanskiy
• Viji Srinivasan - Multiscalar Architecture Performance Analysis.
• Edward Tam
• Stevan Vlaovic

Research Publications

The research being performed by the Parallel Performance Project is subdivided in the following catagories:

• Performance Evaluation
• Domain Decomposition & Synchronization
• Cache Optimization & Data Layout
• Instruction Scheduling & Register Allocation
• Microprocessor Architecture & Pipeline Design

Presentations

• Origin 2000 Design Enhancements for Communication Intensive Applications. (PACT, Oct 1998).
• Reducing Communication Cost in Scalable Shared Memory Systems (Dissertation Defense, 1998).
• Effects of Architectural and Technological Advances on the HP/Convex Exemplar's Memory and Communication Performance. (ISCA, July 1998).
• A Comparative Study of Cache-Coherent Nonuniform Memory Access Systems (HPCS, May 1998).
• Configuration Independent Analysis for Characterizing Shared-Memory Applications (IPPS, April 1998).
• Modeling the Communication Performance of the IBM SP2 (IPPS, April 1996).

Recent Graduates and Former Project Members

• Professor Santosh G. Abraham
• Gheith Abandah - Parallel Systems Characterization and Design.
• Waqar Azeem
• Eric L. Boyd
• Chuan-Hua Chang
• Alexandre E. Eichenberger - Software pipelining, resource modeling
• Eric Hao
• Dave Hudak
• Bassam Kahhaleh
• Hsien-Hsin Lee
• Waleed Meleis - Instruction scheduling, register allocation, combinatorial optimizations
• Bill Mangione-Smith
• John Nguyen
• Tien-Pao Shih
• Rabin Sugamar
• Karen Tomko
• Perry Wang
• John-David Wellman
• Daniel Windheiser

Contributed Software and Tools

• The Parallel Performance Project is making some of its research tools available to other researchers. These tools are intended to help in developing better shared memory applications and systems. They enable trace collection, application analysis, existing-system calibration, and system simulation. Please click here to access these tools.
  
  

Back to Advanced Computer Architecture Laboratory or Department of Electrical Engineering and Computer Science .