Date: Wed, 15 Jan 1997 01:43:28 GMT
Server: Apache/1.1.1
Content-type: text/html
Content-length: 6537
Last-modified: Tue, 10 Dec 1996 15:52:02 GMT
College of Computing - Systems Research Group Home Page
Systems Research Group
What's New!
Motivation
Parallel and distributed machines have become the premiere computational
engines for both business and research endeavors, and their use is
permeating into many other disciplines. In addition, new ways of interacting
with computers have evolved, enabling higher levels of interaction among end
users as well as enhanced interactivity between end users and their programs
than previously possible. As a result of these developments, large-scale
applications routinely use multiple distributed and prallel machines
connected by high performance communcation media.
The architectural environment for facilitating the construction and use of
interactive distributed and parallel applications is multigranular --
consisting of fine-grain SIMD machines, medium-grain MIMD machines, and
coarse-grain mulitcomputers comprised of a farm of high-peformance
workstations, all of which are interconnected via high speed networks. While
such hardware and some of the software components are readily available,
what is lacking is the technology for easily assembling them to form
complete applications that are efficient, flexible, and highly usable for
non-computer scientists. The development of such technology is a fundamental
challenge for systems researchers.
The systems group in the College of Computing addresses this challenge in an
application-conscious way. The research projects currently being investigated
cross the boundaries of architecture, operating systems, compilers and
programming languages, and usability:
- architectural mechanisms for scalable parallel systems, scalability
studies of parallel and distributed systems, system abstractions for
representing shared state in distributed systems, availability and
fault-tolerance issues in distributed systems, operating mechanisms for
high-performance parallel and real-time systems, peformance monitoring tools
for parallel and distributed systems, and system mechanisms for parallel and
distributed discrete event simulations.
Current Projects:
- HPPCEL - High Performance Parallel Computing and Experimentation Lab
- Cthreads - Portable
user-level threads package
- FALCON - Threads based
application monitoring and steering tools
- PBIO --
Portable Binary Input/Output. A binary data
meta-representation library.
- DataExchange --
A communications library built on top of PBIO that eases
construction of networks of cooperating enties.
- Distributed Labs - Virtual
Labs from geographically dislocated computers
- COBS - Configurable Objects for
High Performance Systems
- Catamaran - Replicated
Data Systems
- PROGRESS - Program Steering Toolkit
- Beehive
- Scalable Shared Memory Multiprocessor Project
- PADS
- Parallel and Distributed Discrete Event Simulation
- TASS
- A Top-Down Approach to Scalabilty Study
Faculty:
Visiting Researchers:
Students:
Alumni:
Contact information:
At College of Computing :
Vernard Martin
College of Computing
801 Atlantic Drive
Georgia Institue of Technology
Atlanta, Ga 30332
E-mail : vernard@cc.gatech.edu