Logo: to the web site of Uppsala University

uu.sePublications from Uppsala University
Change search
ExportLink to record
Permanent link

Direct link
BETA

Project

Project type/Form of grant
Project grant
Title [sv]
Application-specific Coherence for Concurrent Acceleration of Managed Language
Title [en]
Application-specific Coherence for Concurrent Acceleration of Managed Language
Abstract [en]
Managed languages dominate computing: enterprises run Java/C#, Python drives AI, and the web runs on JavaScript. This success comes from the productivity of automating common error-prone tasks (such as memory management) and by enabling rapid prototyping (via dynamic languages). Yet the price is performance: managed languages must constantly perform checks and manipulations that introduce high overhead.Our insight is that much of the overhead comes from the inability to offload runtime checks to the hardware due to the cost of sharing and synchronizing detailed program information. Yet this is changing: recent hardware advances make it practical to share program metadata and build small hardware accelerators.This project will extend these ideas into something much more powerful and flexible: application-specific coherence and synchronization supported in hardware. Such a layer will allow us to implement much of today’s application-specific synchronization (locking, barriers, etc.) with direct hardware support, thereby greatly accelerating them.¬†Managed languages provide the ideal testbed for this as they require frequent, algorithm-specific, fine-grained synchronization. To drive this work we will develop concurrent hardware accelerators for two key managed language tasks (garbage collection and data structure transformations) to learn how application-specific coherence support can address the fine-grained synchronization challenges between the hardware and software.
Principal InvestigatorBlack-Schaffer, David
Coordinating organisation
Uppsala University
Funder
Period
2020-01-01 - 2023-12-31
National Category
Software EngineeringComputer Engineering
Identifiers
DiVA, id: project:6688Project, id: 2019-04275_VR

Search in DiVA

Software EngineeringComputer Engineering

Search outside of DiVA

GoogleGoogle Scholar