Hardware Multithreaded Transactions

Abstract

Speculation with transactional memory systems helps programmers and compilers produce profitable thread-level parallel programs. Prior work shows that supporting transactions that can span multiple threads, rather than requiring transactions be contained within a single thread, enables new types of speculative parallelization techniques for both programmers and parallelizing compilers. Unfortunately, software support for multi-threaded transactions (MTXs) comes with significant additional inter-thread communication overhead for speculation validation. This overhead can make otherwise good parallelization unprofitable for programs with sizeable read and write sets. Some programs using these prior software MTXs overcame this problem through significant efforts by expert programmers to minimize these sets and optimize communication, capabilities which compiler technology has been unable to equivalently achieve. Instead, this paper makes speculative parallelization less laborious and more feasible through low-overhead speculation validation, presenting the first complete design, implementation, and evaluation of hardware MTXs. Even with maximal speculation validation of every load and store inside transactions of tens to hundreds of millions of instructions, profitable parallelization of complex programs can be achieved. Across 8 benchmarks, this system achieves a geomean speedup of 99% over sequential execution on a multicore machine with 4 cores. CCS Concepts • Computer systems organization ! Multicore architectures; • Software and its engineering ! Multithreading

Harmonizing Speculative and Non-Speculative Execution in Architectures for Ordered Parallelism

“…Unfortunately, escape actions and similar mechanisms, such as open-nested transactions [59,63,64], are incompatible with architectures for ordered parallelism and many recent HTMs. These architectures perform speculative forwarding [4,29,44,71,72,74,86,88], which lets tasks access data written by earlier, uncommitted tasks. 1 Speculative forwarding is crucial because ordered tasks may take a long time to commit.…”

“…Some hardware [16,36,71] and software [12,32] TMs let programmers control the commit order among transactions, bridging the gap between TM and TLS. Other TMs order transactions internally, either to avoid pathologies [11,62] or to implement conflict serializability [4,29,44,72,74]. However, this order is not controllable by programmers.…”

“…Using classic HTM, each database transaction must execute on a single thread as a long memory transaction. By contrast, ordered parallelism allows breaking each database transaction into many short, ordered tasks, exploiting abundant intra-transaction parallelism [29,89].…”