 |
|
 |

|
|
성공사례 - 휴스톤 대학(University of Houston)
휴스톤 대학(University of Houston)
프로파일 휴스톤 대학(University of Houston)
"With the strengths found in the AMD64 platform, we greatly enhance our ability to contribute new insights in leading-edge national research initiatives and to serve university educational needs. We chose AMD Opteron processors because they met our performance requirements, and our experience with AMD’s reliability has been tremendous. AMD processor-based computers have been serving as the workhorses in classrooms within TLC2, so based on past experience, we know they stand up to the demands of scientific and academic use."
-Dr. Lennart Johnsson, Director of ACRL and TLC2, University of Houston
Company Profile
The University of Houston’s Texas Learning and Computation Center (www.tlc2.uh.edu) is a promising research center within the university, supporting the computational sciences and engineering, and computer science as cornerstones of the university’s most innovative research. Named TLC2, the center offers students and faculty researchers – from across the many different colleges and departments making up the University of Houston – invaluable high-performance computer and networking resources, state-of-the-art visualization and large-scale storage facilities for exploring and solving some of the most complex and computationally demanding questions facing science today.
The Advanced Computing Research Laboratory carries out research in algorithms and software methodologies, and develops software for efficient use of high-performance platforms for large-scale computations in science and engineering. Through this research and development it facilitates the early adoption of new technologies and platforms.
Challenge
Large-scale scientific and engineering computations used in the frontiers of science and engineering design have depended on 64-bit platforms for a long time. The emergence of industry-standard 64-bit platforms and Open Software have allowed such research and development to push the frontiers of science and engineering much further. Additionally, there has been a shift of emerging disciplines within the life sciences (i.e., bioinformatics), from being largely experimental to becoming computationally intensive. This shift creates strong demands for high-performance, industry-standard 64-bit computing and large high-performance memory and I/O systems.
The challenge for ACRL is to direct its research and development efforts to systems based on new architectures that will dominate the scientific and engineering marketplace before they begin to dominate the marketplace i.e., the time the platforms based on new architectures become available. If the judgment about the future success of these systems is correct then ACRL’s R&D will be a good investment and a valuable contribution to the adoption and future success of the new architectures.
The challenge for TLC2 is to provide resources that enable nationally and internationally competitive research and education on a very limited budget compared to national centers, and to migrate users whose needs grow beyond the capabilities of 32-bit platforms to 64-bit platforms. TLC2 scientists include some of the largest users at national facilities.
Solution
The ACRL concluded in 2002 that the AMD Opteron™ processor should have a great chance to become a significant architecture for large-scale scientific and engineering simulations. With the support of the Los Alamos Computer Science Institute (LACSI), the ACRL decided to acquire a modest size cluster for its R&D efforts. During the spring of 2003, TLC2 scientists in the bioinformatics area came to the conclusion that for their needs 4-way AMD Opteron processor-based platforms with a large amount of memory would be the most cost-effective solution. “The AMD Opteron processor’s ability to run both 32- and 64-bit code based applications without the need to recompile was a critical part of the bioinformatics group’s evaluation,” said Dr. Johnson.
With the shared interest and needs of ACRL and one of TLC2’s strongest user groups, ACRL and TLC2 acquired a 20 node cluster. The cluster has 16 nodes – each 2P AMD Opteron processor Models 246, with 4 GB of memory and four nodes with 4P AMD Opteron processor Models 842, with 16 GB each. All nodes are interconnected with a 4x Infiniband network. The ACRL team was particularly interested in the integrated memory controllers in the AMD Opteron processor and saw it as an interesting solution to a nagging performance bottleneck with conventional designs.
Impact
Dr. Johnsson anticipates the overall effects of the new AMD Opteron processor-based cluster will be far reaching. “The AMD Opteron processor provides a very competitive solution for the ever growing performance requirements of leading-edge research and development in the sciences and engineering design,” he said. “The impact of ACRL’s research should be significant, and the research by TLC2’s students and faculty should be greatly enhanced by the ability to test many more theories and perform more case studies for validation of the theories – aided by shortened computational turn-around times, and they can deal with increasingly larger and more detailed data sets.”
ACRL’s and TLC2‘s new AMD Opteron processor-based cluster will also directly support one of the important initiatives being led by Dr. Johnsson – to provide better data sharing and collaboration across research communities. “In the past, many communities only shared data and findings through journal and conference publications, a form of sharing that requires many months or even years. With improved network capabilities and software environments for secure sharing of data, many communities are now beginning to engage in data sharing at a time scale measured in hours to weeks regardless of the location of the members of those communities. Such improved and secure sharing will benefit many of TLC2’s research and educational initiatives. The AMD Opteron processor-based cluster will play an important role to help us move in this direction, in particular for our bioinformatics researchers.”
About AMD
AMD (NYSE:AMD) designs and produces innovative microprocessors, Flash memory devices and low-power processor solutions for the computer, communications and consumer electronics industries. AMD is dedicated to helping its customers deliver standards-based, customer-focused solutions for technology users, ranging from enterprises and governments to individual consumers. For more information, visit www.amd.com.
About ACRL
The Advanced Computing Research Laboratory (ACRL) carries out research on new and innovative ways to harness computational resources for scientific and engineering applications and participates in several national and international research efforts in high-performance computing, storage and networking. More information about ACRL is available at http://www.cs.uh.edu/~johnsson.
About TLC2
The Texas Learning and Computation Center (TLC2) fosters and supports interdisciplinary research, education and training in computational sciences and engineering through centers, laboratories and individual faculty initiatives. TLC2 has state-of-the-art computation, visualization and educational facilities for environmental studies, biological, biomedical and energy research, undergraduate and graduate education and teacher training. More information about TLC2 is available at http://www.tlc2.uh.edu.
|
|
|
 |
|