An analysis of the impact of multi-threading on communication performance

Brunet, Élisabeth; Denis, Alexandre

doi:10.1109/ipdps.2009.5160893

Cited by 14 publications

(12 citation statements)

References 9 publications

(9 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It divides rendezvous handshakes into multiple tasks and offloads them to background threads running only on idle cores. This approach, however, also suffers from a non-negligible overhead derived from the necessary multithreaded safety [22].…”

Section: Communication Asynchronous Progressmentioning

confidence: 99%

Dynamic Adaptable Asynchronous Progress Model for MPI RMA Multiphase Applications

Peña

Hammond

et al. 2018

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Abstract-Casper is a process-based asynchronous progress model for MPI one-sided communication on multi-and many-core architectures. The one-sided communication is not yet truly one-sided in most MPI implementations: the target process still relies on software progress to complete incoming operations. Casper allows the user to specify an arbitrary number of cores dedicated to background ghost processes and transparently redirects the user RMA operations to ghost processes by utilizing the PMPI redirection and MPI-3 shared-memory technologies. Although Casper is effective and efficient for applications suffering from lack of asynchronous progress, permanently redirecting operations to a small number of ghost processes might not support complex multiphase applications effectively, which often involve dynamically changing communication density and computing workloads. In this paper, we present an adaptive mechanism in Casper to address the limitation of static asynchronous progress in multiphase applications. We exploit two adaptive strategies, a precise user-guided strategy and a fully transparent and automatic strategy based on self-profiling and prediction, to dynamically reconfigure the asynchronous progress in Casper according to real-time performance characteristics during multiphase execution. We evaluate the adaptive approaches in both microbenchmarks and a real quantum chemistry application suite, NWChem, on the Cray XC30 supercomputer and an Intel Omni-Path cluster.

show abstract

Section: Communication Asynchronous Progressmentioning

confidence: 99%

Dynamic Adaptable Asynchronous Progress Model for MPI RMA Multiphase Applications

Peña

Hammond

et al. 2018

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

show abstract

“…The table presents an overview of the currently available techniques to evaluate the performance of mobile networks, and indicates the degree to which they satisfy the above-mentioned requirements. State-ofthe-art in evaluating software performance can be classified into real world evaluation [20,21,24], machine simulation [17,2,23], emulation [9], analytical modeling [19,18], and performance prediction techniques [22,6]. Real world evaluation is often infeasible with large-scale networks, due to monetary cost and the difficulty of conducting controllable and repeatable experiments.…”

Section: H H H H H H H H H H Hmentioning

confidence: 99%

Modeling communication software execution for accurate simulation of distributed systems

Kristiansen

Plagemann

Goebel

2013

Proceedings of the 1st ACM SIGSIM Conference on Principles of Advanced Discrete Simulation

View full text Add to dashboard Cite

Network simulation is commonly used to evaluate the performance of distributed systems, but these approaches do not account for the performance impact that protocol execution on nodes has on performance, which may be significant. We propose a methodology to capture execution models from communication software running on real devices where the execution models can be integrated with discrete event network simulators to improve their accuracy. We provide a set of rules to instrument the software to obtain the events of importance, and present techniques to create executable models based on the obtained traces. To make the models scalable, processing stages are reduced to statistical distributions. When the resulting models are executed in a device model with a scheduler simulator, we are able to model the dynamics of multithreading and parallel execution. Our initial results from a proof-of-concept extension to Ns-3 show that our models are able to accurately model protocol execution on the Google Nexus One with low simulation overhead.

show abstract

“…The pressure on data structures highly depends on the granularity of critical sections (the portions of codes that have to be protected by a lock). Fine-grain locking techniques [1] or lock-free algorithms [4] have to be applied to communication libraries in order to achieve good concurrency…”

Section: A Concurrent Communicationmentioning

confidence: 99%

“…The placement of tasks influences the performance of network on NUMA machines [7]. Even on SMP architectures, the network performance depends on cache effects [1]. It is thus important to take locality into account when designing a modern communication library.…”

Section: A Concurrent Communicationmentioning

confidence: 99%

“…Mixing threads and networking requires some precautions in communication libraries so as to avoid race conditions when threads access concurrently the library [1]. Moreover, we have shown that the communication library itself may benefit from multi-threading [2], for example to make non-blocking communication actually progress in background and overlap with computation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation