By Topic

The Cache-Core Architecture to Enhance the Memory Performance on Multi-Core Processors

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

2 Author(s)
Mori, Y. ; Grad. Sch. of Inf. Sci. & Eng., Tokyo Inst. of Technol., Tokyo, Japan ; Kise, K.

Multi-core architectures are focused on improving the processor performance. However, multi-core processors cannot achieve their peak performance when application program has a little thread level parallelism. On the other hand, it is difficult to extract thread level parallelism from application programs. Because of these factors, in this paper, we propose the cache-core. The cache-core is a mechanism where the excess core behaves like an L2 data cache by executing software managed cache program. As the cache-core supplies other core with data as an L2 cache, the number of global memory accesses can decrease. So the cache-core enhances the performance of multi-core processor. We evaluate the cache-core on Cell/B. E. processor in detail. Our evaluation results show that the cache-core can improve computational performance. Furthermore, this paper describes consideration of the cache-core for a many-core processor. We evaluate the cache-core when a lot of cores massively access the global memory. The evaluation results show effectiveness of the cache-core on a many-core processor.

Published in:

Parallel and Distributed Computing, Applications and Technologies, 2009 International Conference on

Date of Conference:

8-11 Dec. 2009