João Lourenço

Multi-threaded programs allow one to achieve better performance by doing a lot of work in parallel using multiple threads. Such parallel programs often contain code blocks that a thread must execute atomically, i.e., with no interference from the other threads of the program. Failing to execute these code blocks atomically leads to errors known as atomicity violations. However, frequently it not obvious to tell when a piece of code should be executed atomically, especially when that piece of code contains calls to some third-party library functions, about which the programmer has little or no knowledge at all. One solution to this problem is to associate a contract with such a library, telling the programmer how the library functions should be used, and then check whether the contract is indeed respected. For contract validation, static approaches have been proposed, with known limitations on precision and scalability. In this paper, we propose a dynamic method for contract validation, which is more precise and scalable than static approaches.

Neste artigo apresentamos um protocolo de controlo de concorrência que garante que a execução concorrente de transações é equivalente à sua execução sequencial por uma ordem predefinida. Isto permite executar programas que usam transações de forma determinista. O protocolo (1) permite, pela primeira vez, a execução determinista de programas que usam memória transacional por hardware; e (2) garante a execução determinista de programas que usam memória transacional por software com um desempenho claramente superior ao estado da arte.

Software Transactional Memory algorithms associate metadata with the memory locations accessed during a transactions lifetime. This metadata may be stored in an external table and accessed by way of a function that maps the address of each memory location with the table entry that keeps its metadata (this is the out-place or external scheme); or alternatively may be stored adjacent to the associated memory cell by wrapping them together (the in-place scheme). In transactional memory multi-version algorithms, several versions of the same memory location may exist. The efficient implementation of these algorithms requires a one-to-one correspondence between each memory location and its list of past versions, which is stored as metadata. In this chapter we address the matter of the efficient implementation of multi-version algorithms in Java by proposing and evaluating a novel in-place metadata scheme for the Deuce framework. This new scheme is based in Java Bytecode transformation techniques and its use requires no changes to the application code. Experimentation indicates that multi-versioning STM algorithms implemented using our new in-place scheme are in average 6 × faster than when implemented with the out-place scheme.

Distributed transactional memory (DTM) presents itself as a highly expressive and programmer friendly model for concurrency control in distributed programming. Current DTM systems make use of both data distribution and replication as a way of providing scalability and fault tolerance, but both techniques have advantages and drawbacks. As such, each one is suitable for different target applications, and deployment environments. In this paper we address the support of different data replication models in DTM. To that end we propose ReDstm, a modular and non-intrusive framework for DTM, that supports multiple data replication models in a general purpose programming language (Java). We show its application in the implementation of distributed software transactional memories with different replication models, and evaluate the framework via a set of well-known benchmarks, analysing the impact of the different replication models on memory usage and transaction throughput.

n/a

Transactional memory (TM) is an increasingly popular technique for synchronising threads in multi-threaded programs. To address both correctness and performance-related issues of TM programs, one needs to monitor and analyse their execution. However, monitoring concurrent programs (including TM programs) may have a non-negligible impact on their behaviour, which may hamper the objectives of the intended analysis. In this paper, we propose several approaches for monitoring TM programs and study their impact on the behaviour of the monitored programs. The considered approaches range from specialised lightweight monitoring to generic heavyweight monitoring. The implemented monitoring tools are publicly available to the scientific community, and the implementation techniques used for lightweight monitoring of TM programs may be used as an inspiration for developing other specialised lightweight monitors.

Software transactional memory (STM) algorithms associate metadata with the memory locations accessed during a transaction's lifetime. This metadata may be stored in an external table by resorting to a mapping function that associates the address of a memory cell with the table entry containing the corresponding metadata (out-place or external strategy). Alternatively, the metadata may be stored adjacent to the associated memory cell by wrapping the cell and metadata together (in-place strategy). The implementation techniques to support these two approaches are very different and each STM framework is usually biased towards one of them, only allowing the efficient implementation of STM algorithms which suit one of the approaches and inhibiting a fair comparison with STM algorithms suiting the other. In this paper, we introduce a technique to implement in-place metadata that does not wrap memory cells, thus overcoming the bias and allowing STM algorithms to directly access the transactional metadata. The proposed technique is available as an extension to Deuce and enables the efficient implementation of a wide range of STM algorithms and their fair (unbiased) comparison in a common STM framework. We illustrate the benefits of our approach by analyzing its impact in two popular transactional memory algorithms with several transactional workloads, TL2 and multiversioning, each befitting out-place and in-place, respectively.

n/a

Os sistemas de memória transacional distribuída atuais recorrem essencialmente à distribuição ou à replicação total para distribuir os seus dados pelos múltiplos nós do sistema. No entanto, estas estratégias de replicação de dados apresentam limitações. A distribuição não oferece tolerância a falhas e a replicação total limita a capacidade de armazenamento do sistema. Nesse contexto, a replicação parcial de dados surge como uma solução intermédia, que combina o melhor das duas anteriores com o intuito de mitigar as suas desvantagens. Esta estratégia tem sido explorada no contexto das bases de dados distribuídas, mas tem sido pouco abordada no contexto da memória transacional e, tanto quanto sabemos, nunca antes tinha sido incorporada num sistema de memória transacional distribuída para uma linguagem de propósito geral. Assim, neste artigo propomos e avaliamos uma infraestrutura para replicação parcial de dados para programas Java bytecode, que foi desenvolvida com base num sistema já existente de memória transacional distribuída. A modularidade da infraestrutura que apresentamos permite a implementação de múltiplos algoritmos e, por conseguinte, avaliar em que contextos de utilização (workloads, número de nós, etc.) a replicação parcial se apresenta como uma alternativa viável a outras estratégias de replicação de dados.

n/a

As técnicas e algoritmos desenvolvidos sobre diferentes infraestruturas específicas dificilmente podem ser comparados entre si. Este princípio também se aplica às infraestruturas para execução de Memória Transacional Distribuída (MTD), pois não só são muito escassas aquelas que permitem o desenvolvimento, teste e comparação de vários algoritmos e técnicas de implementação, como fornecem uma interface intrusiva para o programador. Sem uma comparação justa, não é possível aferir quais as técnicas e algoritmos mais apropriados em cada contexto de utilização (workload). Neste artigo propomos uma infraestrutura generalista, muito flexível, que possibilita a experimentação de várias estratégias de MTD, permitindo o desenvolvimento de uma grande variedade de algoritmos e de técnicas de implementação eficientes e otimizadas. Através da sua utilização, é agora possível a comparação de técnicas e algoritmos em diferentes contextos de utilização (workloads), recorrendo a uma única infraestrutura e com implicações mínimas no código da aplicação.

Software Transactional Memory (STM) algorithms correctness rely on metadata associated with the memory locations accessed during the transaction life-time. STM implementations may store this metadata either in-place, by wrapping the memory cells in a container that includes the memory cell itself and the corresponding metadata, or out-place, by resorting to a mapping function that associates the memory cell address to an external table with the corresponding metadata. The implementation techniques for these two approaches are very different and each STM framework is usually biased towards one of them, only allowing the efficient implementation of algorithms that fall into the appropriate category, and inhibiting the fair comparison with STM algorithms falling into the other. In this paper we introduce a technique that supports the use of in-place metadata without requiring to wrap memory cells, thus providing STM algorithms with direct access to the transactional metadata and overcoming the bias. The proposed technique is available as an extension to the DeuceSTM framework and allows the efficient implementation of a wide range of STM algorithms, thus enabling their fair (unbiased) comparison in a common STM infrastructure. We illustrate the benefits of our approach by analyzing its impact in two popular TM algorithms with two different transactional workloads, TL2 and multi-versioning, which bias to out-place and in-place respectively.

This document describes a work-in-progress development of NGen-VM, a distributed infrastructure that manages execution environments with run-time and programming language support targeting applications developed in the Java programming language, deployed over clusters of many-core computers. For each running application or suite of related applications, a dedicated single-system image will be provided, regardless of the concurrent threads running on a single machine (on several cores) or scattered on different computers. Such system images rely on a single model for concurrency management (Transactional Shared Memory Model), in order fill the gap between the hardware infrastructure of clusters of many-core nodes and the application runtime that is independent from that hardware infrastructure. Interactions between threads in the same tasks will be supported by a Transactional Memory framework that provides the programming language with Atomic and Isolated code regions. Interactions between thread on different machines will also use the Transactional Memory model, but now resorting to a Distributed Shared Memory abstraction.

We discuss the problem of building domain oriented environments by a composition of heterogeneous application components and tools. We describe several individual tools that support such environments, namely a distributed monitoring and control tool (DAMS), a process-based distributed debugger (PDBG) and a heterogeneous interconnection model (PHIS). We discuss our experience with the development of a Problem Oriented Environment in the domain of genetic algorithms, obtained by a composition of heterogeneous tools and application components.

We discuss debugging prototypes that can easily support new functionalities, depending on the requirements of high-level computational models, and allowing a coherent integration with other tools in a software engineering environment. Concerning the first aspect, we propose a framework that identifies two distinct levels of functionalities that should be supported by a parallel and distributed debugger using: a process and thread-level, and a coordination level concerning sets of processes or threads. An incremental approach is used to effectively develop prototypes that support both functionalities. Concerning the second aspect, we discuss how the interfacing with other tools has influenced the design of a process-level debugging interface (PDBG) and a distributed monitoring and control layer called (DAMS).