Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
End-To-End Sequential Consistency

Abhayendra Singh, Satish Narayanasamy, Daniel Marino, Todd Millstein, and Madanlal Musuvathi

Abstract

Sequential consistency (SC) is arguably the most intuitive behavior for a shared-memory multithreaded program. It is widely accepted that language-level SC could significantly improve programmability of a multiprocessor system. However, efficiently supporting end-to-end SC remains a challenge as it requires that both compiler and hardware optimizations preserve SC semantics. While a recent study has shown that a compiler can preserve SC semantics for a small performance cost, an efficient and complexity-effective SC hardware remains elusive. Past hardware solutions relied on aggressive speculation techniques, which has not yet been realized in a practical implementation.

This paper exploits the observation that hardware need not enforce any memory model constraints on accesses to thread-local and shared read-only locations. A processor can easily determine a large fraction of these safe accesses with assistance from static compiler analysis and the hardware memory management unit. We discuss a low-complexity hardware design that exploits this information to reduce the overhead in ensuring SC. Our design employs an additional unordered store buffer for fast-tracking thread-local stores and allowing later memory accesses to proceed without a memory ordering related stall.

Our experimental study shows that the cost of guaranteeing end-to-end SC is only 6.2% on average when compared to a system with TSO hardware executing a stock compiler’s output.

Details

Publication typeInproceedings
Published inInternational Symposium on Computer Architecture
PublisherACM
> Publications > End-To-End Sequential Consistency