1 / 7

Cache Replacement Scheme based on Back Propagation Neural Networks

Cache Replacement Scheme based on Back Propagation Neural Networks. -Rakesh Ramananda. Aim of the Project. To build a cache replacement Algorithm for a set associative memory using BPNN. Compare the test results with existing algorithms. Basic info. What is Cache?

yael
Télécharger la présentation

Cache Replacement Scheme based on Back Propagation Neural Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Cache Replacement Scheme based on Back Propagation Neural Networks -Rakesh Ramananda

  2. Aim of the Project • To build a cache replacement Algorithm for a set associative memory using BPNN. • Compare the test results with existing algorithms.

  3. Basic info • What is Cache? • Why its performance place an important role? • Existing Algorithm - LRU - MRU - FIFO Optimal replacement. Shadow Directory

  4. Implementation details of BPNN • Feature vector Memory Address being accessed for a 16 set 4 way set associative cache with line size=16 Input vector = TAG FIELD INDEX • Number of neurons in input layer = size of tag +size of index • Number of neurons in output layer = size of index.

  5. Work till now Training of BPNN for 16 bit addresses for a 4 way 16 sets cache. number of hidden layers is 2 (sigmoidal function) size of hidden layers is varied from 2,4,8,10. for line size 8 and 16. Hit ratio currently is 57%

  6. Future Work Train the network for various length of input data i.e. varying associativity, address width and no.of sets in the cache. Train network with different activation function. If time permits design a cache controller using fuzzy control and compare the functionality between the two.

  7. Question?? 

More Related