|
The broad goal of our research is to
develop
novel compiler optimization techniques and programming environments
for scientific and
distributed applications
Our research group is presently working on two broad projects. Click on the project titles below for more details about each
one:
The goal of this work is to enable arbitrary object code optimization long
after static compile-time, including link-time, runtime, and when porting to
new systems. Click here for more details.
This work is supported by an NSF CAREER award and the
NSF Next Generation Software Program.
The goal of this work is to provide a programming framework (a programming
model, language, compiler, and runtime environment) that enables programmers
to design, develop, and optimize the performance of adaptive distributed
applications. Click here for more details.
This work is supported by an NSF CAREER award , the
NSF Next Generation Software Program, and the NSF Operating
Systems and Compilers Program.
Past
Projects
An
environment for performance modeling of highly scalable parallel applications
on large-scale parallel and distributed systems. This project involves 6
different institutions with complementary interests. Our work at Illinois
focuses on compiler support for a performance modeling techniques, and on
integrating compilers and performance models for runtime control of adaptive
parallel applications. The specific research issues
addressed in our work include:
An
application representation that
can be synthesized using parallelizing compiler technology.
Compiler
support for parallel discrete-event simulation:
We have achieved major improvements to the state of the art of
simulation of parallel (message-passing) programs. In particular, we have
been able to simulate programs that are 10-100 times larger than previous
state-of-the-art simulation tools based on PDES.
The POEMS project was funded by DARPA
ITO and by the NSF Next
Generation Software Program (EIA-9975024).
dHPF
Compiler
Advanced
compiler techniques for computation partitioning, communication optimization
and code-generation for data-parallel languages.
A more detailed overview of the goals and results of this project is
available here.
The
dHPF project is funded by DARPA ITO, NSF, the Texas Advanced Technology
Program, and NASA.
D
System Tools
Integrating
performance tools with compilers for source-level performance evaluation. A
more detailed overview of the goals and results of this project is available here.
    
Last
Revised: April 03, 2002 05:36 PM
.
|
