Date of Award

2023

Degree Type

Thesis

Degree Name

Master of Science in Electrical Engineering (MSEE)

Department

Electrical, Computer, and Biomedical Engineering

First Advisor

Resit Sendag

Abstract

In modern computer systems, memory access latencies can have a significant impact on performance, and caching and prefetching are used to reduce these latencies. Caches store recently accessed data to allow the central processing unit (CPU) to access it quickly, while prefetchers attempt to predict future data requests and load them into the cache in advance.

However, the simultaneous use of different prefetchers at different levels of the cache hierarchy can be challenging, as interactions between them can potentially lead to performance degradation. As prefetchers are often designed in isolation, without considering how they interact with prefetchers at other levels, these prefetcher interactions have not been studied extensively.

Therefore, this thesis investigates these inter-level prefetcher interactions. Various prefetcher and system evaluation metrics are presented and applied using adaptations made to the simulator used. Two different approaches are introduced for the further analysis of harmful prefetcher interactions. By applying these approaches for demonstration purposes, it was possible to identify, reduce or even eliminate some performance degrading interactions between prefetchers.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.