Abstract

The majority of today’s digital circuits use synchronous pipelines. As the technology nodes get smaller, these pipelines are facing problems with area, power, and timing. One of the major sources of power consumption is the global clock and stall signals. These signals have to be routed across large sections of the chip, and with regards to stalling the pipeline, often face significant timing issues. One solution, developed by Hans M. Jacobson et al., is “Synchronous Interlocked Pipelines”. This pipeline design combines synchronous pipelines with the handshaking of asynchronous pipelines. Asynchronous pipelines are less power intensive because they send acknowledge and request signals to neighboring stages that allow stages to turn off when not being used. Jacobson et al. use this handshaking technique to create local valid and stall signals instead of using global ones. To test the benefits of this design, an asynchronous pipeline, synchronous pipeline, and interlocked synchronous pipeline were built using a generic 45 nm library. Comparisons showed that while the asynchronous and interlocked synchronous pipelines took up 4 times more area than the synchronous pipeline, the asynchronous pipeline had the highest throughput of the three pipeline designs, followed by the interlocked synchronous pipeline. The synchronous pipeline had the worst throughput.

Publication Date

5-2019

Document Type

Master's Project

Student Type

Graduate

Degree Name

Electrical Engineering (MS)

Department, Program, or Center

Electrical Engineering (KGCOE)

Advisor

Mark A. Indovina

Advisor/Committee Member

Sohail A. Dianat

Campus

RIT – Main Campus

Download

Share

COinS