|
|
8th International Workshop on
Software and Compilers for Embedded Systems
SCOPES 2004
Workshop Program - Presentation Abstract
| A Fuzzy Adaptive Algorithm for Fine Grained Cache Paging |
Susmit Bagchi - NTNU
Mads Nygaard - NTNU
|
| The performance of page replacement algorithms used by the virtual
memory management system and file system of operating systems is
important because of the higher disk access latency. The situation
complicates further under the limitation of memory. There exists a set
of page replacement algorithms, however, a majority of them are based on
static policies. In this paper, a novel Fuzzy Adaptive Page Replacement
algorithm (FAPR) is proposed. The FAPR algorithm applies fuzzy inference
technique based on an adaptive rule-base and online priority control.
The algorithm is simulated in a highly memory-constrained environment
under various load distributions. The results demonstrate that the FAPR
algorithm enhances the performance in comparison to the commonly used
algorithms such as, LRU and LFU. In the worst-case execution model, FAPR
algorithm averages the performance of the LRU and LFU algorithms. This
paper describes the design and simulation of the FAPR algorithm along
with the results. |
|