Abstract. In this paper we investigate the issue of automatically identifying the "natural" degree of parallelism of an application using software transactional memory (STM), i.e., the workload-specific multiprogramming level that maximizes application's performance. We discuss the importance of adapting the concurrency level to the workload in two different scenarios, a shared-memory and a distributed STM infrastructure. We propose and evaluate two alternative self-tuning methodologies, explicitly tailored for the considered scenarios. In shared-memory STM, we show that lightweight, black-box approaches relying solely on on-line exploration can be extremely effective. For distributed STMs, we introduce a novel hybrid approach that combines model-driven performance forecasting techniques and on-line exploration in order to take the best of the two techniques, namely enhancing robustness despite model's inaccuracies, and maximizing convergence speed towards optimum solutions.

We present the Input Acceptance of Transactional Memory (TM). Despite the large interest for performance of TMs, no existing research work has investigated the impact of solving a conflict that does not need to be solved. Traditional solutions for a TM to be correct is to delay or abort a transaction as soon as it presents a risk to violate consistency. Both alternatives are costly and should be avoided if consistency is actually preserved. To address this problem, we introduce the input acceptance of a TM as its ability to commit transactions, we upper-bound the input acceptance of existing TMs and propose a new TM with higher input acceptance.

No abstract

As opposed to database transactional systems, transactional memory (TM) systems are constrained by real-time while treating their input workload. Nevertheless, there is no clear formalization of how a TM should react regarding to a specific input. While TM performance is often measured in terms of throughput, i.e., commit-rate (by time unit), we consider the commit-abort ratio of a TM for a given input, as the number of transactions this TM commits over the total number of input transactions. Building onto this, this paper defines the input acceptance of TMs depending on their commit-abort ratio on input classes. To this end, we exhibit several classes of workloads and identify bounds on the input acceptance of existing TM designs. Additionally, we propose a serializable STM that presents high input acceptance at the cost of a more complex algorithm than existing STMs. Finally, experimental validation compares the presented TM designs in terms of input acceptance with realistic workloads.

No abstract