What is it about?
Many digital systems use hardware with feedback paths that limit how quickly individual tasks can be completed. This paper explores how such hardware can still deliver high overall performance by sharing it across multiple users or tasks. By allowing different computations to take turns using the same logic, the system stays busy and processes more work in total. We develop a model based on queueing theory to understand and predict performance in this shared setup. We also show how different design choices—such as how many tasks are allowed to wait in line—can affect system efficiency. Several example applications demonstrate how this approach can be tuned to get the most out of the hardware.
Featured Image
Photo by Ciprian Pardău on Unsplash
Read the Original
This page is a summary of: Utilizing Virtualized Hardware Logic Computations to Benefit Multi-User Performance, Electronics, March 2021, MDPI AG,
DOI: 10.3390/electronics10060665.
You can read the full text:
Contributors
The following have contributed to this page
