DMCA
Data-centric Multi-level Blocking (1997)
Cached
Download Links
- [www.cs.cornell.edu]
- [iss.cs.cornell.edu]
- [iss.ices.utexas.edu]
- DBLP
Other Repositories/Bibliography
| Citations: | 155 - 10 self |
BibTeX
@INPROCEEDINGS{Kodukula97data-centricmulti-level,
author = {Induprakas Kodukula and Nawaaz Ahmed and d Keshav Pingali},
title = {Data-centric Multi-level Blocking},
booktitle = {},
year = {1997},
pages = {346--357}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
We present a simple and novel framework for generating blocked codes for high-performance machines with a memory hierarchy. Unlike traditional compiler techniques like tiling, which are based on reasoning about the control flow of programs, our techniques are based on reasoning directly about the flow of data through the memory hierarchy. Our data-centric transformations permit a more direct solution to the problem of enhancing data locality than current control-centric techniques do, and generalize easily to multiple levels of memory hierarchy. We buttress these claims with performance numbers for standard benchmarks from the problem domain of dense numerical linear algebra. The simplicity and intuitive appeal of our approach should make it attractive to compiler writers as well as to library writers. 1 Introduction Data reuse is imperative for good performance on modern high-performance computers because the memory architecture of these machines is a hierarchy in which the cost of ...
Keyphrases
memory hierarchy data-centric multi-level blocking data-centric transformation library writer control flow good performance standard benchmark traditional compiler technique modern high-performance computer memory architecture high-performance machine problem domain current control-centric technique performance number multiple level novel framework data locality direct solution intuitive appeal introduction data reuse blocked code dense numerical linear algebra
