Computer Science Home | People | Degrees | Papers | Research | Events | Other FCMS Subjects
 
Theses In Progress

Neil Bradley
Analysing branch prediction algorithms using OpenPAT

Chris Griffiths
Web based simulation of medical devices for training and testing purposes

Brian Hardyment
O-Glycosylation site prediction

Emily Khong
Digitally prototyping a controller for multiple medical infusions pumps

Zane Tootell
Informing users about data leakage in the Cloud

Theses
Theses In Progress >> Dissertations
 
Neil Bradley - Analysing branch prediction algorithms using OpenPAT

Date Completed: in progress

Pipelining is a method for speeding up CPU cycles. But it comes at a cost - if subsequent instructions depend on the result of an instruction currently being processed, it can either be delayed (nop), or the processor can attempt to predict what the result will be. Whilst a prediction may result in wasted cycles, a nop always will, which reduces both speed and efficiency. In the same token, there is a trade-off between predictor complexity and chip size that also affects efficiency, speed and cost. Therefore it is necessary to get the best value for money, in terms of accuracy vs complexity, from a CPU's branch predictor. Given this, the aim of this project is to re-evaluate the classic McFarling branch predictor analysis, using OpenPAT and modern benchmarks, before pursuing further research in branch prediction algorithms.

  2007 FCMS. The University of Waikato - Te Whare Wananga o Waikato