Predictive Cache Modeling and Analysis
Abstract
This work applied particle swarm heuristic optimization techniques to the problem of finding a near-optimal order in which to schedule tasks in a real-time embedded system in order to minimize cache miss rates experienced by the software. Reducing the number of cache misses is an important component of runtime execution efficiency. We demonstrated runtime reductions of 3-5% in execution time, significant for embedded systems attempting to add new capability without upgrading hardware. The expectation is that these gains can be improved further by the use of hardware with pseudo-LRU cache behavior.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2011
- Accession Number
- ADA552968
Entities
People
- Anirudda Gokhale
- Brian Dougherty
- Jonathan Preston
- Jules White
- Russell Kegley
Organizations
- Lockheed Martin Aeronautics