Temporal Data Placement
Loading...
Files
Date
Keywords
data placement optimization, compile-time data placement, polyhedral model, polyhedral optimization, implementation of Farkas Lemma, linear programming, integer programming, mixed integer, non-linear programming
Degree Level
masters
Advisor
Degree Name
M. Eng.
Volume
Issue
Publisher
Memorial University of Newfoundland
Abstract
How can we provide data where it is required and when it is required to the execution units of parallel hardware? Program transformations have been a focus to improve the performance of parallel computing, whereas data optimizations like data placement, data layout transformation, data migration and data replications are overlooked especially in compiler domain. We are proposing a methodology, Temporal Data Placement, that will schedule and place data in both time and space. The use of our methodology will enhance the performance of parallel systems significantly, and it can also be automated.