Protocols and strategies for optimizing performance of remote memory operations on clusters

Nieplocha, Jaroslaw; Tipparaju, Vinod; Saify, A.; Panda, D.K.

doi:10.1109/ipdps.2002.1016563

Cited by 26 publications

(25 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our work on GASNet [20] and that of Nieplocha et al on ARMCI [35] show that in fact one-sided communication can often outperform two-sided message-passing communication. Moreover, the results of this paper in the context of Titanium and others in the context of CAF [37] and UPC [5] show that these performance advantages carry over to application-level performance in the NAS benchmarks.…”

Section: Zplmentioning

confidence: 99%

Titanium Performance and Potential: An NPB Experimental Study

Datta

Bonachea

Yelick

2006

Languages and Compilers for Parallel Computing

View full text Add to dashboard Cite

Titanium is an explicitly parallel dialect of Java TM designed for high-performance scientific programming. It offers objectorientation, strong typing, and safe memory management in the context of a language that supports high performance and scalable parallelism. We present an overview of the language features and demonstrate their use in the context of the NAS Parallel Benchmarks, a standard benchmark suite of kernels that are common across many scientific applications. We argue that parallel languages like Titanium provide greater expressive power than conventional approaches, enabling much more concise and expressive code and minimizing time to solution without sacrificing parallel performance. Empirical results demonstrate our Titanium implementations of three of the NAS Parallel Benchmarks can match or even exceed the performance of the standard MPI/Fortran implementations at realistic problem sizes and processor scales, while still using far cleaner, shorter and more maintainable code.

show abstract

Section: Zplmentioning

confidence: 99%

Titanium Performance and Potential: An NPB Experimental Study

Datta

Bonachea

Yelick

2006

Languages and Compilers for Parallel Computing

View full text Add to dashboard Cite

show abstract

“…Our original XLUPC Myrinet port implements multiple transfer protocols depending on message size [20]. Short messages are copied to avoid memory registration costs.…”

Section: Considerations For the Myrinet/gm Implementationmentioning

confidence: 99%

“…MPI implementations like OpenMPI [22] and MVAPICH [27] as well as one-sided messaging systems like ARMCI [20] follow a differential approach based on message size, switching between preallocated registered memory buffers (Bounce Buffers) for short messages and dynamic memory registration and de-registration as part of each data transfer (Rendezvous) for large ones. The crossover point between the protocols is dependent on the underlying network hardware and software, requiring tuning for each machine.…”

Section: Related Workmentioning

confidence: 99%

Scalable RDMA performance in PGAS languages

Farreras

Almási

Caşcaval

et al. 2009

2009 IEEE International Symposium on Parallel &Amp; Distributed Processing

View full text Add to dashboard Cite

Partitioned Global Address Space (PGAS) languages provide a unique programming model that can span shared-memory multiprocessor (SMP) architectures, distributed memory machines, or cluster of SMPs. Users can program large scale machines with easy-to-use, shared memory paradigms.In order to exploit large scale machines efficiently, PGAS language implementations and their runtime system must be designed for scalability and performance. The IBM XLUPC compiler and runtime system provide a scalable design through the use of the Shared Variable Directory (SVD). The SVD stores meta-information needed to access shared data. It is dereferenced, in the worst case, for every shared memory access, thus exposing a potential performance problem.In this paper we present a cache of remote addresses as an optimization that will reduce the SVD access overhead and allow the exploitation of native (remote) direct memory accesses. It results in a significant performance improvement while maintaining the run-time portability and scalability.

show abstract

“…ARMCI provides an excellent implementation substrate for global address space languages making use of coarse-grain communication because it achieves high performance on a variety of networks (including Myrinet, Quadrics, and IBM's switch fabric for its SP systems) as well as shared memory platforms (Cray X1, SGI Altix3000, SGI Origin2000) while insulating its clients from platform-specific implementation issues such as shared memory, threads, and DMA engines. A notable feature of ARMCI is its support for non-contiguous data transfers [12].…”

Section: Communication Librariesmentioning

confidence: 99%

Experiences with Sweep3D implementations in Co-array Fortran

2006

View full text Add to dashboard Cite

As part of the recent focus on increasing the productivity of parallel application developers, Co-array Fortran (CAF) has emerged as an appealing alternative to the Message Passing Interface (MPI). CAF belongs to the family of global address space parallel programming languages; such languages provide the abstraction of globally addressable memory accessed using one-sided communication. At Rice University we are developing cafc, an open source, multiplatform CAF compiler. Our earlier studies show that cafc-compiled CAF programs achieve similar performance to that of corresponding MPI codes for the NAS Parallel Benchmarks. In this paper, we present a study of several CAF implementations of Sweep3D on four modern architectures. We analyze the impact of using one-sided communication in Sweep3D, identify potential sources of inefficiencies and suggest ways to address them. Our results show that we achieve comparable performance to that of the MPI version on three cluster-based architectures and outperform it by up to 10% on the SGI Altix 3000.

show abstract

Protocols and strategies for optimizing performance of remote memory operations on clusters

Cited by 26 publications

References 9 publications

Titanium Performance and Potential: An NPB Experimental Study

Titanium Performance and Potential: An NPB Experimental Study

Scalable RDMA performance in PGAS languages

Experiences with Sweep3D implementations in Co-array Fortran

Contact Info

Product

Resources

About