jayteaftw Goto Github PK

followers: 1.0 following: 1.0 repos: 39.0 gists: 0.0

Name: JT

Type: User

Bio: AI Engineer, Cloud Engineer, Software Engineer, Everything Engineer 🙃

JT's Projects

avl-tree

Python implimentation of an AVL tree which is a self-balancing binary search tree where the heights of the 2 child subtrees of any node differe by at most one.

bear-predictor

The Bear Predictor is a PyTorch model created using fastAi that is able to predict 7 of the 8 kinds of bears: Grizzly, Black, Polar, Panda, Andean, Sloth, and Sun Bear.

bone-appetit-automatic-dog-feeder

An automatic dog feeder that uses the Dabble API to interface a smart phone/tablet with an Arduino using Bluetooth.

Branch prediction is a technique used in computer architecture schemes to reduce the overhead cost of branch operations. The present paper examines the application of the following static and dynamic branch prediction methods presented in Santa Clara University’s Advanced Computer Architecture (COEN 313) course: Always Taken, Always Not Taken, N-Bit, Correlating, and Gshare. The Prophet-Critic hybrid predictor is also implemented as a potential method of custom hybrid branch prediction by the authors. Analysis of the aforementioned methods was compared based on the resulting misprediction rates observed upon application to large datasets. Empirical results concluded a substantial performance improvement using dynamic and custom branch prediction techniques over static methods, yielding hit rates of up to 90%. The substantial improvements in performance can be attributed to the active analysis of the branch pattern behaviors. Together, these findings recommend that dynamic techniques of branch prediction are used to improve instruction pipelining and increase CPU efficiency.