CASS-MT News Archive

06/14/11 - Researchers at the Center for Adaptive Supercomputing Software (CASS), located at the Department of Energy’s Pacific Northwest National Laboratory in Richland, Wash., have released a report on the Workshop on Semantic Graph Database Search Patterns held in Seattle in April. The workshop brought together technical leaders from government, academia, and industry in semantic databases and high performance computing. CASS teamed with Cray, Inc. to host the workshop, which was the first in a series of meetings to be held over the next year. The intent of the meetings is to define a benchmark set of complex search patterns to derive development and evaluation of next-generation semantic database technologies. For more information about upcoming workshops, contact John Feo, 509-375-3768.

12/04/09 - Below are the presentations given during the XMT Birds of a Feather at Supercomputing 2009 in Portland, OR.

• SC09 BOF Agenda
• XMT Status and Roadmap, by Shoaib Mufti
• Characterizing and Analyzing Massiv Spatio-Temporal Graphs, by David Bader
• The MultiThreaded Graph Library, by Greg Mackey
• Recent Research on the Cray XMT, by Shahid Bokhari

07/15/09 - Daniel Chavarria, a senior computational researcher at Pacific Northwest National Laboratory, gave a presentation at the 2009 Society of Industrial & Applied Mathematics (SIAM) Annual Meeting as part of the symposium on High-Performance Computing on Massive Real-World Graphs. Chavarria’s invited talk, titled “Scalable & efficient parallelization of graph methods for Boolean satisfiability and power grid analysis on the Cray XMT,” focused on the challenges and successes in the implementation of graph-based algorithms on the Cray XMT system at PNNL.

“Graph-based algorithms have broad applicability to many different areas of science and technology. Unfortunately, executing them efficiently for large-scale graphs requires a significant time and resource investment on traditional HPC systems,” says Chavarria. “The Cray XMT multithreaded architecture provides specialized hardware designed for the efficient parallel execution of highly irregular codes, such as those derived from graph algorithms.”

The SIAM’09 meeting was held from July 6 through July 10.

04/24/09 - Research at Pacific Northwest National Laboratory to gain a greater understanding of electric grid failures is producing promising results. Using PNNL's Cray XMT machine, researchers are conducting advanced power grid contingency analysis, a valuable tool to help determine the impact of potential component failures. Keep reading...

04/23/09 - Congratulations to George Chin, Andres Marquez, and Sutanay Choudhury of PNNL’s Center for Adaptive Supercomputing Software Multithreaded Architectures (CASS-MT) for the acceptance and publication of their writings for the 2009 Workshop on Multithreaded Architectures and Applications (MTAAP).

The team’s paper entitled “Implementing and Evaluating Multithreaded Triad Census Algorithms on the Cray XMT” presents its experiences in developing and optimizing two implementations of a social network analysis method known as triadic analysis on the Cray XMT – a high performance computer operated by PNNL’s CASS-MT. The paper describes and explores the unique capabilities and features of the Cray XMT and its usefulness and applicability to an important class of graph algorithms.

MTAAP is being held in conjunction with the International Parallel and Distributed Processing Symposium in Rome, Italy in May 2009.

Abstract: Commonly represented as directed graphs, social networks depict relationships and behaviors among social entities such as people, groups, and organizations. Social network analysis denotes a class of mathematical and statistical methods designed to study and measure social networks. Beyond sociology, social network analysis methods are being applied to other types of data in other domains such as bioinformatics, computer networks, national security, and economics. For particular problems, the size of a social network can grow to millions of nodes and tens of millions of edges or more. In such cases, researchers could benefit from the application of social network analysis algorithms on high-performance architectures and systems.

The Cray XMT is a third generation multithreaded system based on the Cray XT-3/4 platform. Like most other multithreaded architectures, the Cray XMT is designed to tolerate memory access latencies by switching context between threads. The processors maintain multiple threads of execution and utilize hardware-based context switching to overlap the memory latency incurred by any thread with the computations from other threads. Due to its memory latency tolerance, the Cray XMT has the potential of significantly improving the execution speed of irregular data-intensive applications such as those found in social network analysis.

In this paper, we describe our experiences in developing and optimizing two implementations of a social network analysis method known as triadic analysis to execute on the Cray XMT. The two implementations possess different execution complexities, qualities, and characteristics. We evaluate how the various attributes of the codes affect their performance on the Cray XMT. We also explore the effects of different compiler options and execution strategies on the different triadic analysis implementations and identify general XMT programming issues and lessons learned.

George Chin, Andres Marquez, Sutanay Choudhury

04/23/09 - Congratulations to Chad Scherrer of PNNL’s Center for Adaptive Supercomputing Software Multithreaded Architectures (CASS-MT) for the acceptance and publication of his writings at MTAAP 2009.

Chad’s paper titled “Accelerating Numerical Calculation on the Cray XMT” describes an approach to improve the numerical performance of the Cray XMT – a high performance computer operated by PNNL’s CASS-MT.

The MTAAP conference is hosting workshops on multithreaded architectures and applications of high performance computing at Rome in May.

Abstract: The Cray XMT provides hardware support for parallel algorithms that would be communication- or memory-bound on other machines. Unfortunately, even if an algorithm meets these criteria, performance suffers if the algorithm is too numerically intensive. We present a lookup-based approach that achieves a significant performance advantage over explicit calculation. We describe an approach to balancing memory bandwidth against on-chip floating point capabilities, leading to further speedup. Finally, we provide table lookup algorithms for a number of common functions.

Scherrer C, TR Shippert, and A Marquez

04/23/09 - Congratulations to Oreste Villa of PNNL’s Center for Adaptive Supercomputing Software Multithreaded Architectures (CASS-MT) for the acceptance and publication of his writings at IPDPS 2009.

Oreste’s paper titled “Input-independent, Scalable and Fast String Matching on the Cray XMT” offers answers for improving the speed and real-time, string searching solutions made possible on the Cray XMT – a high performance computer operated by PNNL’s CASS-MT. Oreste and his colleagues recognize that string searching is at the core of many security and network applications like search engines, intrusion detection systems, virus scanners and spam filters.

The IPDPS is hosting an IEEE International Parallel and Distributed Processing Symposium at Rome in May.

Abstract: String searching is at the core of many security and network applications like search engines, intrusion detection systems, virus scanners and spam filters. The growing size of on-line content and the increasing wire speeds push the need for fast, and often real-time, string searching solutions. For these conditions, many software implementations (if not all) targeting conventional cache-based microprocessors do not perform well. They either exhibit overall low performance or exhibit highly variable performance depending on the types of inputs. For this reason, real-time state of the art solutions rely on the use of either custom hardware or Field-Programmable Gate Arrays (FPGAs) at the expense of overall system flexibility and programmability.

This paper presents a software based implementation of the Aho-Corasick string searching algorithm on the Cray XMT multithreaded shared memory machine. Our solution relies on the particular features of the XMT architecture and on several algorithmic strategies: it is fast, scalable and its performance is virtually content independent. On a 128-processor Cray XMT, it reaches a scanning speed of _ 28 Gbps with a performance variability below 10%. In the 10 Gbps performance range, variability is below 2.5%. By comparison, an Intel dual-socket, 8-core system running at 2.66 GHz achieves a peak performance which varies from 500 Mbps to 10 Gbps depending on the type of input and dictionary size.

Oreste Villa, Daniel Chavarria-Miranda, and Kristyn Maschhoff

04/15/09 - RICHLAND, Wash. – Computational scientist John Feo has joined the Department of Energy's Pacific Northwest National Laboratory to lead software development that takes advantage of a different kind of computer hardware. Supercomputers with multi-threaded architecture can analyze vast amounts of data found in complex networks more efficiently than conventional ones. Feo will serve as the director of a group of scientists developing software for the multi-threaded Cray XMT and similar computers.

The XMT's memory is built in such a way that it can handle seemingly random data faster than typical hardware. Feo will direct the fledgling Center for Adaptive Supercomputing Software on the PNNL campus. At the Center, researchers are developing software that can analyze and visualize data coming from many types of complex networks, such as the nation's electrical grid or biological communities in the subsurface, for example.

Feo comes to PNNL from Microsoft Corporation, where he led a group developing new applications for many core processors. Prior to working for Microsoft, Feo spent time at Cray, Inc., where he worked on the early generations of the Cray XMT, and Lawrence Livermore National Laboratory. Feo has a doctorate in computer science from the University of Texas, Austin, and a masters in astronomy.

The Center for Adaptive Computing Software is home to the only Cray XMT system available to external scientific users as an Open Science system. The Cray XMT consists of 16 multithreaded processors, 128 gigabyte RAM and a 2.3-terabyte Lustre parallel filesystem.

01/14/09 - Click for more information and registration links.

12/08/08 - This article was published in the Fall 2008 issue and highlights the capabilities and application potential of the Cray XMT. Read the article. (The SciDAC Review is published by DOE ASCR.)

12/02/08 - Materials from BOF meeting: Cray XMT Resources Management, XMT Overview SC08 BOF, SC08 CASS-MT Presentation
The BOF meeting took place on Tuesday, November 18th at the Courtyard Marriot hotel located adjacent to the main SC’08 conference center.

12/02/08 - PNNL’s Daniel Chavarria explores the center’s recent research advancements and provides information on how you can partner with CASS-MT to advance science. The original demonstration took place on Nov. 19th, 1:00 p.m., Austin, TX.

View a PDF of the presentation.

10/20/08 - View the article.