OpenSHMEM Extensions and a Vision for Its Future Direction

Poole, Stephen W; Shamis, Pavel; Welch, Aaron; Pophale, Swaroop; Venkata, Manjunath Gorentla; Hernández, Óscar; Koenig, Gregory A.; Curtis, Tony; Hsu, Chung-Hsing

doi:10.1007/978-3-319-05215-1_11

Cited by 6 publications

(2 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We introduced the teams extension in [16] and now expand the concept by providing teams within teams, explicit indexing of members of a team and providing a team split functionality. Traditional active sets in OpenSHMEM were implicitly defined as part of collective calls and did not conceptually live outside of individual calls.…”

Section: Teamsmentioning

confidence: 98%

Extending the OpenSHMEM Memory Model to Support User-Defined Spaces

Welch

Pophale²,

Shamis

et al. 2014

Proceedings of the 8th International Conference on Partitioned Global Address Space Programming Models

Self Cite

View full text Add to dashboard Cite

OpenSHMEM is an open standard for SHMEM libraries.With the standardisation process complete, the community is looking towards extending the API for increasing programmer flexibility and extreme scalability. According to the current OpenSHMEM specification (revision 1.1), allocation of symmetric memory is collective across all PEs executing the application. For better work sharing and memory utilisation, we are proposing the concepts of teams and spaces for OpenSHMEM that together allow allocation of memory only across user-specified teams. Through our implementation we show that by using teams we can confine memory allocation and usage to only the PEs that actually communicate via symmetric memory. We provide our preliminary results that demonstrate creating spaces for teams allows for less consumption of memory resources than the current alternative. We also examine the impact of our extensions on Scalable Synthetic Compact Applications #3 (SSCA3), which is a sensor processing and knowledge formation kernel involving file I/O, and show that up to 30% of symmetric memory allocation can be eliminated without affecting the correctness of the benchmark.

show abstract

Section: Teamsmentioning

confidence: 98%

Extending the OpenSHMEM Memory Model to Support User-Defined Spaces

Welch

Pophale²,

Shamis

et al. 2014

Proceedings of the 8th International Conference on Partitioned Global Address Space Programming Models

Self Cite

View full text Add to dashboard Cite

show abstract

“…As part of the OpenSHMEM community's advance towards the next generation of the OpenSHMEM specification [5], we would like to explore how the productivity of the Open-SHMEM interface can be improved without sacrificing its functionality and performance. In this work we explore how the semantics of OpenSHMEM can be integrated in the context of high-level programming languages.…”

Section: Introductionmentioning

confidence: 99%

PySHMEM: A High Productivity OpenSHMEM Interface for Python

Welch

Shamis

Peng

et al. 2015

2015 9th International Conference on Partitioned Global Address Space Programming Models

Self Cite

View full text Add to dashboard Cite

OpenSHMEM is a well known high performance communication library implementing the Partitioned Global Address Space (PGAS) programming model. It exposes a broad range of one-sided communication semantics which maps well to modern network technologies and can achieve a level of performance that is close to that of network hardware. In this paper we explore how OpenSHMEM semantics can be integrated within the context of Python, a high level programming language that is widely used by the scientific community. In this work we introduce the PySHMEM library that provides simple and convenient interfaces for accessing OpenSHMEM functionality. The goal of this paper is twofold: (1) investigate how high-level programming semantics can be used to expose a high productivity interface for accessing low-level communication semantics, and (2) evaluate performance overheads imposed by Python with respect to native (C) OpenSHMEM performance.

show abstract