skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Scientific Application Performance on Candidate PetaScalePlatforms

Abstract

After a decade where HEC (high-end computing) capability was dominated by the rapid pace of improvements to CPU clock frequency, the performance of next-generation supercomputers is increasingly differentiated by varying interconnect designs and levels of integration. Understanding the tradeoffs of these system designs, in the context of high-end numerical simulations, is a key step towards making effective petascale computing a reality. This work represents one of the most comprehensive performance evaluation studies to date on modern HEC systems, including the IBM Power5, AMD Opteron, IBM BG/L, and Cray X1E. A novel aspect of our study is the emphasis on full applications, with real input data at the scale desired by computational scientists in their unique domain. We examine six candidate ultra-scale applications, representing a broad range of algorithms and computational structures. Our work includes the highest concurrency experiments to date on five of our six applications, including 32K processor scalability for two of our codes and describe several successful optimizations strategies on BG/L, as well as improved X1E vectorization. Overall results indicate that our evaluated codes have the potential to effectively utilize petascale resources; however, several applications will require reengineering to incorporate the additional levels of parallelism necessary to achievemore » the vast concurrency of upcoming ultra-scale systems.« less

Authors:
; ; ; ; ; ; ; ; ; ;
Publication Date:
Research Org.:
Ernest Orlando Lawrence Berkeley NationalLaboratory, Berkeley, CA (US)
Sponsoring Org.:
USDOE Director. Office of Science. Advanced ScientificComputing Research
OSTI Identifier:
932484
Report Number(s):
LBNL-62952
R&D Project: K11121; BnR: KJ0101030; TRN: US200813%%77
DOE Contract Number:
DE-AC02-05CH11231
Resource Type:
Conference
Resource Relation:
Conference: International Parallel and Distributed ProcessingSymposium (IPDPS), Long Beach, CA, March 24-30,2007
Country of Publication:
United States
Language:
English
Subject:
42; ALGORITHMS; EVALUATION; PERFORMANCE; PROCESSING; SUPERCOMPUTERS

Citation Formats

Oliker, Leonid, Canning, Andrew, Carter, Jonathan, Iancu, Costin, Lijewski, Michael, Kamil, Shoaib, Shalf, John, Shan, Hongzang, Strohmaier, Erich, Ethier, Stephane, and Goodale, Tom. Scientific Application Performance on Candidate PetaScalePlatforms. United States: N. p., 2007. Web.
Oliker, Leonid, Canning, Andrew, Carter, Jonathan, Iancu, Costin, Lijewski, Michael, Kamil, Shoaib, Shalf, John, Shan, Hongzang, Strohmaier, Erich, Ethier, Stephane, & Goodale, Tom. Scientific Application Performance on Candidate PetaScalePlatforms. United States.
Oliker, Leonid, Canning, Andrew, Carter, Jonathan, Iancu, Costin, Lijewski, Michael, Kamil, Shoaib, Shalf, John, Shan, Hongzang, Strohmaier, Erich, Ethier, Stephane, and Goodale, Tom. Mon . "Scientific Application Performance on Candidate PetaScalePlatforms". United States. doi:. https://www.osti.gov/servlets/purl/932484.
@article{osti_932484,
title = {Scientific Application Performance on Candidate PetaScalePlatforms},
author = {Oliker, Leonid and Canning, Andrew and Carter, Jonathan and Iancu, Costin and Lijewski, Michael and Kamil, Shoaib and Shalf, John and Shan, Hongzang and Strohmaier, Erich and Ethier, Stephane and Goodale, Tom},
abstractNote = {After a decade where HEC (high-end computing) capability was dominated by the rapid pace of improvements to CPU clock frequency, the performance of next-generation supercomputers is increasingly differentiated by varying interconnect designs and levels of integration. Understanding the tradeoffs of these system designs, in the context of high-end numerical simulations, is a key step towards making effective petascale computing a reality. This work represents one of the most comprehensive performance evaluation studies to date on modern HEC systems, including the IBM Power5, AMD Opteron, IBM BG/L, and Cray X1E. A novel aspect of our study is the emphasis on full applications, with real input data at the scale desired by computational scientists in their unique domain. We examine six candidate ultra-scale applications, representing a broad range of algorithms and computational structures. Our work includes the highest concurrency experiments to date on five of our six applications, including 32K processor scalability for two of our codes and describe several successful optimizations strategies on BG/L, as well as improved X1E vectorization. Overall results indicate that our evaluated codes have the potential to effectively utilize petascale resources; however, several applications will require reengineering to incorporate the additional levels of parallelism necessary to achieve the vast concurrency of upcoming ultra-scale systems.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Mon Jan 01 00:00:00 EST 2007},
month = {Mon Jan 01 00:00:00 EST 2007}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • This work presents the ScalaBLAST Web Application (SWA), a web based application implemented using the PHP script language, MySQL DBMS, and Apache web server under a GNU/Linux platform. SWA is an application built as part of the Data Intensive Computer for Complex Biological Systems (DICCBS) project at the Pacific Northwest National Laboratory (PNNL). SWA delivers accelerated throughput of bioinformatics analysis via high-performance computing through a convenient, easy-to-use web interface. This approach greatly enhances emerging fields of study in biology such as ontology-based homology, and multiple whole genome comparisons which, in the absence of a tool like SWA, require a heroicmore » effort to overcome the computational bottleneck associated with genome analysis. The current version of SWA includes a user account management system, a web based user interface, and a backend process that generates the files necessary for the Internet scientific community to submit a ScalaBLAST parallel processing job on a dedicated cluster.« less
  • This paper compares specific procedures of two data base management system available at Brookhaven National Laboratory, System-2000 (S2K) and Scientific Information Retrieval (SIR). The computer on which these reside is a CDC-6600 and the operating system is NOS-BE level 488. For this comparison study a new S2K data base and a SIR data base were created using a subset of the actual raw data from the original Acid Rain data base. Identical data was loaded in both data bases and the procedures followed were duplicated as closely as possible on both. All jobs were run in batch mode to getmore » accurate statistics on computer time and resources. This paper does not try to compare the two data base management systems exhaustively, but rather discusses the results of certain specific procedures in terms of computer resources and the ease of implementing these procedures from the user's standpoint.« less
  • This paper describes the Sets and Fields (SAF) scientific data modeling system. It is a revolutionary approach to interoperation of high performance, scientific computing applications based upon rigorous, math-oriented data modeling principles. Previous technologies have required all applications to use the same data structures and/or meshes to represent scientific data or lead to an ever expanding set of incrementally different data structures and/or meshes. SAF addresses this problem by providing a small set of mathematical building blocks--sets, relations and fields--out of which a wide variety of scientific data can be characterized. Applications literally model their data by assembling these buildingmore » blocks. A short historical perspective, a conceptual model and an overview of SAF along with preliminary results from its use in a few ASCI codes are discussed.« less
  • Graphite moderated gas cooled reactors led the way into the nuclear age with the Chicago Pile-1 reactor, which provided the first sustained critical reaction in December, 1942. The first commercial nuclear plant, Calder Hall in the UK, went critical in 1956 with an outlet gas temperature of {approx}345 C. As depicted in Fig. 1, in five decades since Calder Hall, outlet temperature increased rapidly, reaching a plateau of {approx}950 C. This apparent ceiling is in large part due to limitations in the structural materials utilized within the core (e.g. control systems) and primary loop (hot duct, heat-exchangers etc.) Simply, themore » operating temperatures of Generation III (HTGR's) are very near performance limits of the structural alloys used, both in terms of elevated temperature and as-irradiated properties. This limitation remains today and is the reason the outlet temperature of the Generation IV Very High Temperature Reactor (VHTR) continues to be revised downward from the original optimistic goal of {approx}1200 C, to it's current target outlet temperature of {approx}950 C, a temperature consistent with the previous generation of HTGR's. An example of the challenges facing Generation IV VHTR is found by considering the control rods. For the Fort St. Vrain Reactor the control system consisted of thirty tubes each containing B4C control material. Alloy 800, originally developed by Inco in the 1950's under the trade-name Incoloy 800 and 800H had found widespread application is steam generators, turbines and was selected for this control rod application. These materials are included for Class 1 Nuclear Component by ASME section III. In addition to Ft. St Vrain, alloy 800H has found control rod application in the German HTR and Japanese HTTR reactors and is the primary choice Pebble Bed Modular Reactor (a HTGR) reactivity control system. Figure 2 gives the ASME allowed stress for Alloy 800H. Due to the loss in creep rupture strength, the allowed design stress for decreases substantially above {approx}500 C. For the operating temperature (and current maximum code temperature) of the Fort St. Vrain reactor ({approx}760 C) the control rod designed from Alloy 800H would be allowed approximately 20 MPa primary stress. For the anticipated temperature of the VHTR that stress would likely be under 10 MPa in the non-irradiated condition. Selection of VHTR in-core structural material will depend on the ability for the material to withstand extreme temperatures and irradiation. The use of Alloy 800H for control rod applications is preferred, but will require designing with a relatively low strength material which becomes brittle upon modest irradiaton. Presently, more fidelity in both design and the available Alloy 800H database are required to judge the lifetime this alloy will have in service. The only currently envisioned alternative to Alloy 800H are the CFC and SiC/SiC composites, which are currently challenged by code qualification issues. As with Alloy 800H, the composite irradiation data-base is inadequate, though it appears that the CFC service lifetime will be limited due to its anisotropic dimensional change under irradiation to doses somewhat higher than Alloy 800H (1-5 x 10{sup 25} n/m{sup 2}), while SiC/SiC composites appear to retain their mechanical properties to doses in excess of the VHTR lifetime dose. As the path to qualification of both CFC and SiC/SiC are similar, it appears that SiC/SiC is more attractive for high-dose Gen IV application unless high thermal conduction is required.« less