Date of Award

2025

Degree Type

Thesis

Degree Name

Master of Science in Computer Engineering

Department

Electrical, Computer, and Biomedical Engineering

First Advisor

Resit Sendag

Abstract

This work proposes replacing fully static ASIC functional units with programmable functional units that leverage tightly integrated programmable logic within CPU cores. The programmable units can be configured to match the execution behavior of a specific workload while still allowing the CPU to remain general-purpose across all workloads. This approach reduces the total area required for functional units without sacrificing performance. We evaluate two configurations: an Ideal implementation, where programmable and ASIC functional units have equal area, and an Area-Larger implementation, where programmable units occupy twice the area of their ASIC counterparts. To establish feasibility, we first conduct binary instrumentation and micro-op analysis of the x86 instructions executed across several benchmarks. The results indicate strong potential for this approach, as each benchmark exercised only a subset of available functional units. Simulations of a CPU core equipped with programmable functional units show that performance remains unchained with up to a 60% functional-unit area reduction in the Ideal configuration and up to 50% area reduction in the Area-Larger configuration. Furthermore, repurposing the saved area to increase L1D or L1I cache capacity yields an average performance improvement of 5.8%.

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.