Architecture of the Component Collective Messaging Interface

Kumar, Sameer; Dózsa, Gábor; Berg, Jeremy M; Cernohous, Bob; Miller, Douglas; Ratterman, Joe; Smith, Brian E.; Heidelberger, Philip

doi:10.1007/978-3-540-87475-1_10

Cited by 12 publications

(9 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Bruck [19] also studies multiporting in the context of allto-all collectives. In recent work, Kumar et al [20,21] incorporate a number of these optimizations in collective interfaces that exist at different layers of the messaging stack. They show that performance saturates at different number of links, depending on the message protocol used.…”

Section: Image Compositing and Collective Communication Algorithmsmentioning

confidence: 99%

A configurable algorithm for parallel image-compositing applications

Peterka

Goodell

Ross

et al. 2009

Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis

View full text Add to dashboard Cite

Collective communication operations can dominate the cost of large-scale parallel algorithms. Image compositing in parallel scientific visualization is a reduction operation where this is the case. We present a new algorithm called Radix-k that in many cases performs better than existing compositing algorithms. It does so through a set of configurable parameters, the radices, that determine the number of communication partners in each message round. The algorithm embodies and unifies binary swap and direct-send, two of the best-known compositing methods, and enables numerous other configurations through appropriate choices of radices. While the algorithm is not tied to a particular computing architecture or network topology, the selection of radices allows Radix-k to take advantage of new supercomputer interconnect features such as multiporting. We show scalability across image size and system size, including both powers of two and nonpowers-of-two process counts.

show abstract

Section: Image Compositing and Collective Communication Algorithmsmentioning

confidence: 99%

A configurable algorithm for parallel image-compositing applications

Peterka

Goodell

Ross

et al. 2009

Proceedings of the Conference on High Performance Computing Networking, Storage and Analysis

View full text Add to dashboard Cite

show abstract

“…Such operations discern start and completion of a collective operation, and therefore allow to execute a predefined group communication operation simultaneously to other computations on the main CPU. Asynchronous progression makes an abstraction of the group operation necessary and typical implementations [6], [7] use communication schedules to represent the communication. However, as far as our experience goes there is currently no efficient way to create and handle the necessary schedules.…”

Section: Background and Related Workmentioning

confidence: 99%

“…Schemes implemented in LibNBC [6] and IBM's Collective Component Messaging Interface [7] already rely on such a strict ordering. In order to improve the efficiency of the scheduler, we show how this ordering can be relaxed to provide more degrees of freedom at each given scheduling decision.…”

Section: A Universal Definition For Group Operationsmentioning

confidence: 99%

Group Operation Assembly Language - A Flexible Way to Express Collective Communication

Hoefler

Siebert²,

Lumsdaine

2009

2009 International Conference on Parallel Processing

View full text Add to dashboard Cite

Abstract-The implementation and optimization of collective communication operations is an important field of active research. Such operations directly influence application performance and need to map the communication requirements in an optimal way to steadily changing network architectures. In this work, we define an abstract domain-specific language to express arbitrary group communication operations. We show the universality of this language and how all existing collective operations can be implemented with it. By design, it readily lends itself to blocking and nonblocking execution, as well as to off-loaded execution of complex group communication operations. We also define several offline and online optimizations (compiler transformations and scheduling decisions, respectively) to improve the overall performance of the operation. Performance results show that the overhead to express current collective operations is negligible in comparison to the potential gains in a highly optimized implementation.

show abstract

“…An optimized version of the library was implemented for InfiniBand [7] and shows significant better performance for the communication time as well as asynchronous progression. IBM's Component Collective Messaging Interface (CCMI [8]) also implements optimized nonblocking collective operations for different architectures.…”

Section: Related Workmentioning

confidence: 99%