North Carolina State University (NCSU) researchers have developed Dense Footprint Cache, a new method to boost the speed of computer applications by more than 9%.
The researchers say the speed improvement results from techniques that enable computer processors to retrieve data more efficiently, and it lets a processor's dynamic random-access memory (DRAM) cache learn over time which data the processor needs from each macroblock. The techniques enable the cache to compress the macroblock, retrieving only the relevant data, which in turn enables the cache to send data to the processor more efficiently. In addition, the new method frees up space in the cache that can be used to store other data.
The researchers tested the Dense Footprint Cache in a processor and memory simulator. They ran 3 billion instructions for each application tested via the simulator, and found the new method accelerated applications by 9.5% compared to state-of-the-art competing methods for managing die-stacked DRAM.
The researchers note Dense Footprint Cache also used 4.3% less energy, and led to a significant improvement in "last-level cache miss ratios."
From NCSU News
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA
No entries found