Hardware-Centric Analysis of Network Performance for MPI Applications

Brown, K.; Domke, Jens; Matsuoka, Satoshi

doi:10.1109/icpads.2015.92

Cited by 7 publications

(2 citation statements)

References 23 publications

(21 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Brown et al (2015) proposed a visualizing system of message traffics in a communication network and they succeeded to identify hot spots. In their case study using the samplesort program running on TSUBAME 2.5, 5% performance gain was obtained by avoiding the hot spots which they discovered by using their tool.…”

Section: Related Workmentioning

confidence: 99%

Overhead of using spare nodes

Hori

Yoshinaga

Hérault

et al. 2020

The International Journal of High Performance Computing Applica

View full text Add to dashboard Cite

With the increasing fault rate on high-end supercomputers, the topic of fault tolerance has been gathering attention. To cope with this situation, various fault-tolerance techniques are under investigation; these include user-level, algorithm-based fault-tolerance techniques and parallel execution environments that enable jobs to continue following node failure. Even with these techniques, some programs with static load balancing, such as stencil computation, may underperform after a failure recovery. Even when spare nodes are present, they are not always substituted for failed nodes in an effective way. This article considers the questions of how spare nodes should be allocated, how to substitute them for faulty nodes, and how much the communication performance is affected by such a substitution. The third question stems from the modification of the rank mapping by node substitutions, which can incur additional message collisions. In a stencil computation, rank mapping is done in a straightforward way on a Cartesian network without incurring any message collisions. However, once a substitution has occurred, the optimal node-rank mapping may be destroyed. Therefore, these questions must be answered in a way that minimizes the degradation of communication performance. In this article, several spare node allocation and failed node substitution methods will be proposed, analyzed, and compared in terms of communication performance following the substitution. The proposed substitution methods are named sliding methods. The sliding methods are analyzed by using our developed simulation program and evaluated by using the K computer, Blue Gene/Q (BG/Q), and TSUBAME 2.5. It will be shown that when failures occur, the stencil communication performance on the K and BG/Q can be slowed around 10 times depending on the number of node failures. The barrier performance on the K can be cut in half. On BG/Q, barrier performance can be slowed by a factor of 10. Further, it will also be shown that almost no such communication performance degradation can be seen on TSUBAME 2.5. This is because TSUBAME 2.5 has an Infiniband network connected with a FatTree topology, while the K computer and BG/Q have dedicated Cartesian networks. Thus, the communication performance degradation depends on network characteristics.

show abstract

Section: Related Workmentioning

confidence: 99%

Overhead of using spare nodes

Hori

Yoshinaga

Hérault

et al. 2020

The International Journal of High Performance Computing Applica

View full text Add to dashboard Cite

show abstract

“…PERUSE was an international effort to design a callback interface to collect internal information from MPI implementations. PERUSE was implemented in Open MPI [16] and used by selected projects in the MPI community [3,4,17].…”

Section: Perusementioning

confidence: 99%

Enabling callback-driven runtime introspection via MPI_T

Hermanns

Hjlem

Knobloch

et al. 2018

Proceedings of the 25th European MPI Users' Group Meeting

View full text Add to dashboard Cite

Understanding the behavior of parallel applications that use the Message Passing Interface (MPI) is critical for optimizing communication performance. Performance tools for MPI currently rely on the PMPI Profiling Interface or the MPI Tools Information Interface, MPI_T, for portably collecting information for performance measurement and analysis. While tools using these interfaces have proven to be extremely valuable for performance tuning, these interfaces only provide synchronous information, i.e., when an MPI or an MPI_T function is called. There is currently no option for collecting information about asynchronous events from within the MPI library. In this work we propose a callback-driven interface for event notification from MPI implementations. Our approach is integrated in the existing MPI_T interface and provides a portable API for tools to discover and register for events of interest. We demonstrate the functionality and usability of the interface with a prototype implementation in Open MPI, a small logging tool (MEL) and the measurement infrastructure Score-P. CCS CONCEPTS• General and reference → Measurement; Performance; • Software and its engineering → Software libraries and repositories; KEYWORDS MPI, callback functions, runtime introspection ACM Reference Format:

show abstract

xCCL: A Survey of Industry-Led Collective Communication Libraries for Deep Learning

Weingram

et al. 2023

J. Comput. Sci. Technol.

View full text Add to dashboard Cite

Hardware-Centric Analysis of Network Performance for MPI Applications

Cited by 7 publications

References 23 publications

Overhead of using spare nodes

Overhead of using spare nodes

Enabling callback-driven runtime introspection via MPI_T

xCCL: A Survey of Industry-Led Collective Communication Libraries for Deep Learning

Contact Info

Product

Resources

About