Why Transactional Memory Should Not Be Obstruction-Free

TL;DR

This paper argues that obstruction-free transactional memory is inherently inefficient, providing theoretical lower bounds and a lock-based implementation that outperforms obstruction-free approaches, highlighting a fundamental complexity gap.

Contribution

It establishes lower bounds on obstruction-free TMs and introduces a lock-based TM that surpasses these bounds, demonstrating the advantages of blocking techniques.

Findings

Obstruction-free TMs have significant theoretical lower bounds.

Blocking TMs can outperform obstruction-free TMs in complexity.

Lock-based TM implementation exceeds the efficiency of obstruction-free methods.

Abstract

Transactional memory (TM) is an inherently optimistic abstraction: it allows concurrent processes to execute sequences of shared-data accesses (transactions) speculatively, with an option of aborting them in the future. Early TM designs avoided using locks and relied on non-blocking synchronization to ensure obstruction-freedom: a transaction that encounters no step contention is not allowed to abort. However, it was later observed that obstruction-free TMs perform poorly and, as a result, state-of-the-art TM implementations are nowadays blocking, allowing aborts because of data conflicts rather than step contention. In this paper, we explain this shift in the TM practice theoretically, via complexity bounds. We prove a few important lower bounds on obstruction-free TMs. Then we present a lock-based TM implementation that beats all of these lower bounds. In sum, our results exhibit a considerable complexity gap between non-blocking and blocking TM implementations.