Optimized Use of Parallel Programming Interfaces in Multithreaded Embedded Architectures

Lorenzon, Arthur F.; Sartor, Anderson L.; Márcia, C.; Beck, Antonio Carlos Schneider

doi:10.1109/isvlsi.2015.85

Cited by 16 publications

(6 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The pseudo‐applications were used to analyze performance and energy consumption on embedded systems and general purpose processors. In other works, the authors classified the pseudo‐applications in CPU‐Bound, Weakly Memory‐Bound, and Memory‐Bound, according to the following criteria. Reads/writes to memory ‐ represents the number of accesses to the shared and private memory addresses of the processor, considering read and write operations for each pseudo‐application. Data dependence ‐ means that at least one thread/process can only start its execution when the computation result of one or more threads/ processes is over.…”

Section: Pampar Pseudo‐applicationsmentioning

confidence: 99%

“…To fill this gap, this work proposes PAMPAR, a new benchmark using a set of 11 pseudo‐applications developed with the purpose of evaluating the performance and energy consumption in multicore architectures. These pseudo‐applications were developed and classified according to different criteria in previous studies . These studies have shown that these pseudo‐applications have characteristics that are distinct enough to be representatives in different scenarios.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

PAMPAR: A new parallel benchmark for performance and energy consumption evaluation

Garcia

Schepke

Girardi

2019

Concurrency and Computation

View full text Add to dashboard Cite

Summary This paper presents PAMPAR, a new benchmark to evaluate the performance and energy consumption of different Parallel Programming Interfaces (PPIs). The benchmark is composed of 11 algorithms implemented in PThreads, OpenMP, MPI‐1, and MPI‐2 (spawn) PPIs. Previous studies have used some of these pseudo‐applications to perform this type of evaluation in different architectures since there is no benchmark that offers this variety of PPIs and communication models. In this work, we measure the energy and performance of each pseudo‐application in a single architecture, varying the number of threads/processes. We also organize the pseudo‐applications according to their memory accesses, floating‐point operations, and branches. The goal is to show that this set of pseudo‐applications has enough features to build a parallel benchmark. The results show that there is no single best case that provides both better performance and low energy consumption in the presented scenarios. Moreover, the pseudo‐applications usage of the system resources are different enough to represent different scenarios and be efficient as a benchmark.

show abstract

Section: Pampar Pseudo‐applicationsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

PAMPAR: A new parallel benchmark for performance and energy consumption evaluation

Garcia

Schepke

Girardi

2019

Concurrency and Computation

View full text Add to dashboard Cite

show abstract

“…Neste sentido, diferentes técnicas têm sido propostas com o objetivo de melhorar o custo benefício entre o desempenho e consumo de energia, representado pela métrica EDP (energy-delay product) [Lorenzon and Beck 2019]. Por um lado, considerando que muitas aplicações paralelas não escalam conforme o número de threads executando a aplicação aumenta, estratégias que ajustam dinamicamente o número de threads conseguem agir no nível de software para otimizar o EDP de tais aplicações sem a necessidade de modificação nas características do hardware [Lorenzon et al 2015. Por outro lado, dado que componentes de hardware permitem que sua frequência e tensão de operação possam ser alteradas em tempo de execução, Sistemas Operacionais implementam técnicas para a gerência de tais componentes [Sartor et al 2017].…”

Section: Introductionunclassified

PampaFreq: Otimizando o EDP de Aplicações Paralelas em Processadores AMD

Costa

Marques

Medeiros

et al. 2020

Anais Do XXI Simpósio Em Sistemas Computacionais De Alto Desempenho (SSCAD 2020)

View full text Add to dashboard Cite

O DVFS (Dynamic Voltage and Frequency Scaling) tem sido amplamente utilizado para melhorar o uso dos recursos computacionais quando aplicações paralelas estão sendo executadas. No entanto, as aplicações paralelas têm comportamentos distintos e se relacionam de diferentes maneiras com as políticas de modiﬁcação de frequência do DVFS. Neste sentido, é necessário utilizar métodos otimizados de DVFS para melhorar o custo-benefício entre desempenho e consumo de energia, representado pelo EDP (energy-delay product). Dito isso, através de uma extensa exploração de espaço e projeto de diferentes políticas de DVFS, níveis de frequência de operação da CPU e modo de operação de técnicas de boosting com a execução de dezesseis aplicações paralelas em três arquiteturas multicore, nós propomos PampaFreq, uma metodologia que otimiza o EDP em processadores AMD considerando as características da aplicação em tempo de execução. No caso mais signiﬁcativo, PampaFreq otimiza o EDP em até 38% quando comparado com governor ondemand.

show abstract

“…In parallel programming interfaces that use shared memory to communicate during execution (e.g. OpenMP), the communication represents an important part of the total energy consumed by the system [4]. Furthermore, the performance will increase as a result of the system parallelization, promoting energy reductions in the system [4].…”

Section: Introductionmentioning

confidence: 99%

“…OpenMP), the communication represents an important part of the total energy consumed by the system [4]. Furthermore, the performance will increase as a result of the system parallelization, promoting energy reductions in the system [4]. However, the increase in performance is not linear and may not scale with the number of threads, due to external bandwidth and the data-synchronization [5].…”

Section: Introductionmentioning

confidence: 99%

Energy Consumption Estimation in Parallel Applications: an Analysis in Real and Theoretical Models

Solveira¹,

Moro²,

Cruz³

et al. 2016

Anais Do XVII Simpósio Em Sistemas Computacionais De Alto Desempenho (SSCAD 2016)

View full text Add to dashboard Cite

This paper presents a detailed energy consumption analysis, considering the energy consumption related to CPU, cache memory and main memory of parallel applications on HPC systems. Furthermore, this paper also presents the correlation between energy consumption, Speedup, and execution time. Experiments are conducted with the NAS parallel benchmarks using three different measurement tools: 1) Intel PCM, 2) Perf Linux, and 3) HP CACTI. The results show a comparison between two approaches to obtain energy consumption results. One using PCM and other using Perf and CACTI. The DRAM results show an average variation between these approaches of 47% for sequential applications, and 19% for parallel applications. The system results show that the lowest energy consumption occurs only when all physical cores are used, showing that the hyper-threading system did not bring beneﬁts in energy consumption to the system. Moreover, the cache memories results show that the cache miss rate (regardless of the level) increases with the number of threads. However, a parallel application has lower cache memory energy consumption when compared to its sequential version.

show abstract

Optimized Use of Parallel Programming Interfaces in Multithreaded Embedded Architectures

Cited by 16 publications

References 16 publications

PAMPAR: A new parallel benchmark for performance and energy consumption evaluation

PAMPAR: A new parallel benchmark for performance and energy consumption evaluation

PampaFreq: Otimizando o EDP de Aplicações Paralelas em Processadores AMD

Energy Consumption Estimation in Parallel Applications: an Analysis in Real and Theoretical Models

Contact Info

Product

Resources

About