Hierarchical Processors-and-Memory Architecture for High Performance Computing
Jose A.B. Fortes, Rudolph Eigenmann
Purdue University
Valerie Taylor
Northwestern University
The proposed work consists of refining and evaluating a design of a 100-Teraops multiprocessor system organized as a hierarchy of processors-and-memory subsystems (HPAM) by carefully considering its technological feasibility along with the usability of the programming environment in the context of important representative applications. A programming environment capable of supporting a combination of automatic translation of standard programs, user-specified machine specific programs or optimizations and library-based programming is to be investigated. Two classes of applications will be considered: time-constrained, fixed-size problems and large-scale, complex problems. The first consists of scientific simulations used with interactive immersive visualization systems. The second includes the recently introduced SPEChpc96 benchmarks and 2D/3D finite-element modeling applications.
comments to: Larry Picha
posted: 19 April 96PAWS'96