Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Automatic Data and Computation Partitioning on Scalable Shared Memory Multiprocessors

Summary: Automatic Data and Computation Partitioning on
Scalable Shared Memory Multiprocessors
Sudarsan Tandri and Tarek S. Abdelrahman
Department of Electrical and Computer Engineering
The University of Toronto
Toronto, Ontario, Canada, M5S 3G4
e-mail: ftandri,tsag@eecg.toronto.edu
1 Introduction
Scalable Shared Memory Multiprocessors(SSMMs) are becoming increasingly popular
as platforms for parallel scientific computing. Recent commercial systems such as the
Convex Exemplar and the Cray T3E offer not only scalability previously exclusive
to distributed memory multiprocessors, but also the convenience of a single coherent
view of memory. The presense of shared memory initially suggests that parallelizing
compilers for SSMMs need not be concerned with the data management issues that
compilers for distributed memory must contend with. However, the non-uniformity
of memory accesses and limited operating system data management policies suggest
that compilers should play a more active role in data management on SSMMs. A data
partitioning based approach to data management can improve application performance
on SSMMs [1].
In thispaper, we address the problemof automaticallyderivingdata and computation


Source: Abdelrahman, Tarek S. - Department of Electrical and Computer Engineering, University of Toronto


Collections: Computer Technologies and Information Sciences