UMIACS researchers are addressing a number of fundamental issues in
high-performance computing that encompass algorithms, software, and
applications. The UMIACS Parallel
Computing Laboratory is one of the best equipped HPC laboratories
in the nation and includes
an IBM
SP2,
and a 10 node DEC alpha farm.
-
Experimental Parallel Algorithmics:
- A fundamental problem in parallel computing is to design high-level,
architecture independent, algorithms that execute efficiently on general
purpose parallel machines. The purpose of this project is to advance our
understanding of the main factors required for designing practical parallel
algorithms and to develop techniques and data sets for experimentally
validating the results. As a byproduct, we are developing portable parallel
programs and data sets for a number of specific important problems arising in
combinatorial computing and image processing.
- Land Cover Dynamics:
- The goal of this NSF Grand Challege
project is to develop new high performance computing tools, emphasizing
object oriented programming and parallel I/O of large scale images
and maps, as well as new parallel algorithms and systems for image
processing and spatial data handling, and apply them to important
scientific problems in land cover dynamics.
- High Performance
Systems Laboratory (HPSL):
- The development of sophisticated runtime support forms the foundation
of this research effort. Two major research areas are currently under
intensive study. The first is continued development of the runtime
library (CHAOS/PARTI) for parallelizing applications with irregular
data access patterns onto distributed memory parallel machines. The
second area is providing runtime and compiler support for problems
with irregular data access, where the data cannot fit into the main
memory of the parallel machine
- Massively Parallel Support for Knowledge Representation and
Case-based Reasoning:
- This project focuses on using the power of massive parallelism to
develop tools that can be used in Artificial
Intelligence research. In particular, a new frame-based
represented language has been developed, which can exploit the power
of parallelism for performing extremely fast inferencing.
- Fundamentals of Parallel Computing:
- The emphasis is on understanding and anticipating the principles
guiding parallel computing. Examples include modeling trade-offs between
communication throughput and time in scalable parallel systems;
understanding the role that parallel programming could play in machines
that demonstrate modest, but increasing amounts of parallelism;
developing automatic load balancing methods; and studying various modeling
issues that arise in parallel computing research.
Back to the Research Page