Academia.eduAcademia.edu

Outline

Title
Executive Summary
Evaluation Overview
Methodology
Materials Science
Performance Results
References
All Topics
Computer Science
Distributed Systems

Cray X1 Evaluation Status Report

Profile image of Anthony MezzacappaAnthony Mezzacappa

2004

https://doi.org/10.2172/885593
visibility

…

description

49 pages

link

1 file

Related papers

An overview of high performance computing for the physical sciences
Geoffrey Fox

1993

ABSTRACT We present an overview of the state of the art and future trends in high performance computing, and in particular data parallel computing. We describe recent progress in defining a standardized, portable, high level parallel language called High Performance Fortran, an extension of Fortran 90 designed for e cient implementation of data parallel applications on parallel, vector and sequential high performance computers.

View PDFchevron_right
Performance comparison of scientific applications on Cray architectures
Vincent Betro

Current HPC architectures are changing drastically and rapidly as compared to mature scientific applications, which usually evolve at a much slower rate. Newly introduced architectures promise to impact the performance of these heavily used scientific applications. Therefore, it is prudent to understand how the supposed performance benefits and improvements of new architectures translate to the applications. In this paper, we attempt to quantify the differences between the theoretical performance improvements (due to changes in architecture) and the "real-world" improvements in applications by gathering performance data of selected applications from the fields of chemistry, climate, weather, materials science, fusion, and astrophysics running on three different Cray architectures: XT5, XE6, and XC30. Of particular interest is the fact that these three Cray platforms spans three different generations of interconnects, namely: SeaStar2+, Gemini, and Aries, respectively and that all three platforms (or its variant) are either in active use or coming online at HPC centers. Therefore, a performance evaluation of these selected applications on these three different architectures gives a user some useful perspective into the benefits of these architectures. These evaluations are done by comparing the improvements of numerical (micro)-benchmarks to the improvements of the selected applications when run across these architectures.

View PDFchevron_right
Performance evaluation of the SX-6 vector architecture for scientific computations, Concurrency and Computation; Practice and Experience
Rob Van der Wijngaart

2005

The growing gap between sustained and peak performance for scientific applications is a well-known problem in high performance computing. The recent development of parallel vector systems offers the potential to reduce this gap for many computational science codes and deliver a substantial increase in computing capabilities. This paper examines the intranode performance of the NEC SX-6 vector processor, and compares it against the cache-based IBM Power3 and Power4 superscalar architectures, across a number of key scientific computing areas. First, we present the performance of a microbenchmark suite that examines many low-level machine characteristics. Next, we study the behavior of the NAS Parallel Benchmarks. Finally, we evaluate the performance of several scientific computing codes. Overall results demonstrate that the SX-6 achieves high performance on a large fraction of our application suite and often significantly outperforms the cache-based architectures. However, certain cla...

View PDFchevron_right
Science Driven Supercomputing Architectures: AnalyzingArchitectural Bottlenecks with Applications and Benchmark Probes
J. Shalf

2005

There is a growing gap between the peak speed of parallel computing systems and the actual delivered performance for scientific applications. In general this gap is caused by inadequate architectural support for the requirements of modern scientific applications, as commercial applications and the much larger market they represent, have driven the evolution of computer architectures. This gap has raised the importance of developing better benchmarking methodologies to characterize and to understand the performance requirements of scientific applications, to communicate them efficiently to influence the design of future computer architectures. This improved understanding of the performance behavior of scientific applications will allow improved performance predictions, development of adequate benchmarks for identification of hardware and application features that work well or poorly together, and a more systematic performance evaluation in procurement situations.

View PDFchevron_right
Performance evaluation of the SX-6 vector architecture for scientific computations
Rupak Biswas

Concurrency and Computation: Practice and Experience, 2005

The growing gap between sustained and peak performance for scientific applications has become a well-known problem in high performance computing. The recent development of parallel vector systems offers the potential to bridge this gap for a significant number of computational science codes and deliver a substantial increase in computing capabilities. This paper examines the intranode performance of the NEC SX6 vector processor and the cache-based IBM Power3/4 superscalar architectures across a number of key scientific computing areas. First, we present the performance of a microbenchmark suite that examines a full spectrum of low-level hardware characteristics. Next, we study the behavior of the NAS Parallel Benchmarks using a variety of optimization schemes. Finally, we evaluate the performance of several numerical codes from key scientific computing domains. Overall results demonstrate that the SX6 achieves high performance on a large fraction of our application suite and in many cases significantly outperforms the RISC-based architectures. However, certain classes of applications are not easily amenable to vectorization and would likely require extensive reengineering of both algorithm and implementation to utilize the SX6 effectively.

View PDFchevron_right
BSMBench: a flexible and scalable supercomputer benchmark from computational particle physics
Biagio Lucini

Benchmarking plays a central role in the evaluation of High Performance Computing architectures. Several benchmarks have been designed that allow users to stress various components of supercomputers. In order for the figures they provide to be useful, benchmarks need to be representative of the most common real-world scenarios. In this work, we introduce BSMBench, a benchmarking suite derived from Monte Carlo code used in computational particle physics. The advantage of this suite (which can be freely downloaded from http://www.bsmbench.org) over others is the capacity to vary the relative importance of computation and communication. This enables the tests to simulate various practical situations. To showcase BSMBench, we perform a wide range of tests on various architectures, from desktop computers to state-of-theart supercomputers, and discuss the corresponding results. Possible future directions of development of the benchmark are also outlined.

View PDFchevron_right
Scientific Application Performance on Leading Scalar and VectorSupercomputing Platforms
Stephane Ethier

2007

The last decade has witnessed a rapid proliferation of superscalar cache-based microprocessors to build high-end computing (HEC) platforms, primarily because of their generality, scalability, and cost effectiveness. However, the growing gap between sustained and peak performance for full-scale scientific applications on conventional supercomputers has become a major concern in high performance computing, requiring significantly larger systems and application scalability than implied by peak performance in order to achieve desired performance. The latest generation of custom-built parallel vector systems have the potential to address this issue for numerical algorithms with sufficient regularity in their computational structure. In this work we explore applications drawn from four areas: magnetic fusion (GTC), plasma physics (LBMHD3D), astrophysics (Cactus), and material science (PARATEC). We compare performance of the vector-based Cray X1, X1E, Earth Simulator, NEC SX-8, with performance of three leading commodity-based superscalar platforms utilizing the IBM Power3, Intel Itanium2, and AMD Opteron processors. Our work makes several significant contributions: a new data-decomposition scheme for GTC that (for the first time) enables a breakthrough of the Teraflop barrier; the introduction of a new three-dimensional Lattice Boltzmann magneto-hydrodynamic implementation used to study the onset evolution of plasma turbulence that achieves over 26Tflop/s on 4800 ES processors; the highest per processor performance (by far) achieved by the full-production version of the Cactus ADM-BSSN; and the largest PARATEC cell size atomistic simulation to date. Overall, results show that the vector architectures attain unprecedented aggregate performance across our application suite, demonstrating the tremendous potential of modern parallel vector systems.

View PDFchevron_right
Parallel computing comes of age: Supercomputer level parallel computations at Caltech
Geoffrey Fox

Concurrency: Practice and Experience, 1989

Parallel supercomputers are now in regular use at Caltech for several major scientific calculations. We use this experience to abstract a set of lessons for applications, decomposition, performance, hardware and software. We consider hypercubes, transputer arrays and the SIMD Connection Machine CM-2 and AMT DAP. These are contrasted, where possible, with CRAY and other high performance conventional computers.

View PDFchevron_right
Guest Editor's Introduction: Special Issue on High-Performance Computing with Accelerators
David A Bader

IEEE Transactions on Parallel and Distributed Systems, 2011

View PDFchevron_right
High-Performance Computing System for High Energy Physics
Laurențiu Dumitru

In this work we present technical details and recent developments for a computing cluster working in a GRID environment, configured for high energy physics experiments at the National Institute of Physics and Nuclear Engineering. Main ideas and concepts behind the GRID technology are described. Two Virtual Organizations (VO) LHCb and ILC using GRID resources for Monte Carlo production, data analysis and data storage are presented together with the recently initiated development of their specific tools.

View PDFchevron_right

References (69)

  1. IMPLEMENTATION AND EVALUATION PLAN FOR CRAY X1 ................................................
  2. 1 Motivation ...................................................................................................................................
  3. EVALUATION OVERVIEW .............................................................................................................. 2.1 Methodology................................................................................................................................
  4. CRAY X1 DESCRIPTION................................................................................................................... 3.1 Cray X1 Hardware and Software Description .............................................................................
  5. 2 Delivery, Installation, and Operations at CCS.............................................................................
  6. EVALUATION COMPONENTS......................................................................................................... 4.1 Micro-benchmarks.......................................................................................................................
  7. 2 Kernels and Performance Optimization.....................................................................................
  8. 3 Climate Science .........................................................................................................................
  9. 4 Fusion Science...........................................................................................................................
  10. 4.1 GYRO ...................................................................................................................................
  11. 4.2 Science accomplished on the X1 with GYRO ......................................................................
  12. 4.2.1 Transport physics in the vicinity of an s=0 (minimum q) surface ................................
  13. 4.2.2 Bohm to gyroBohm scaling transition..........................................................................
  14. 4.2.3 Examination of the effect of ion "corrugations" on electron temperature gradient (ETG) turbulence ...........................................................................................................................
  15. 4.3 NIMROD...............................................................................................................................
  16. 4.4 AORSA3D ............................................................................................................................
  17. 4.5 GTC.......................................................................................................................................
  18. 5 Materials Science.......................................................................................................................
  19. 5.1 DCA-QMC............................................................................................................................
  20. 5.2 LSMS ....................................................................................................................................
  21. 6 Chemistry ..................................................................................................................................
  22. 6.1 NWCHEM and ARMCI........................................................................................................
  23. 6.2 GAMESS...............................................................................................................................
  24. 7 Astrophysics ..............................................................................................................................
  25. 7.1 AGILE-BOLTZTRAN..........................................................................................................
  26. 8 I/O and System Software Evaluation.........................................................................................
  27. Stability .................................................................................................................................
  28. 8.2 Collaborative Activities with Cray........................................................................................
  29. ADIC SAN planning ....................................................................................................
  30. 8.2.2 Cray Network Server (CNS) optimization ...................................................................
  31. 8.3 Access control .......................................................................................................................
  32. 8.4 Developing the next OS for Black Widow............................................................................
  33. 9 Scalability Evaluation................................................................................................................
  34. 10 Biology: AMBER and NAMD ..................................................................................................
  35. 11 Atomic and Molecular Physics..................................................................................................
  36. 12 Dense Matrix Operations on Cray X1 .......................................................................................
  37. 13 Sparse Matrix Operations on Cray X1 ......................................................................................
  38. 14 sPPM..........................................................................................................................................
  39. 15 HPC Challenge Benchmarks .....................................................................................................
  40. 16 Cosmology.................................................................................................................................
  41. 16.1 X1 Porting.........................................................................................................................
  42. 16.2 Performance Results .........................................................................................................
  43. OUTREACH ACTIVITIES................................................................................................................ 5.1 Workshops.................................................................................................................................
  44. 1.1 Climate Modeling System Workshop ...................................................................................
  45. 1.2 ORNL Cray X1 Tutorial ....................................................................................................... iv
  46. Cray-Fusion UltraScale Computing Workshop ....................................................................
  47. Doe SciDAC Workshop: Porting CCSM to the Cray X1 .....................................................
  48. 1.5 Computational Materials Science: Early Evaluation of the Cray X1....................................
  49. 1.6 Prototype Application Workshops ........................................................................................
  50. 1.7 Biology Workshop ................................................................................................................
  51. 1.8 Cray Hands-On Workshop I..................................................................................................
  52. 9 Cray Hands-On Workshop II ................................................................................................
  53. 1.10 CCSM Vectorization Workshop (organized by NCAR)...................................................
  54. 2 Cray Users Group (CUG) Participation.....................................................................................
  55. 2.1 CUG 2003 .............................................................................................................................
  56. 2.2 CUG 2004 -"Peaks, Streams, and Cascades" .......................................................................
  57. RESULTING PUBLICATIONS......................................................................................................... 6.1 Papers ........................................................................................................................................
  58. 2 Presentations..............................................................................................................................
  59. REFERENCES ...................................................................................................................................
  60. ACKNOWLEDGEMENTS ................................................................................................................
  61. • M. R. Fahey, J. B. White III, "DOE Ultrascale Evaluation Plan of the Cray X1," 45th Cray User Group Conference, Hyatt on Capital Square, Columbus, Ohio, May 12-16, 2003.
  62. • J. B. White III, "An Optimization Experiment with the Community Land Model on the Cray X1," 45th Cray User Group Conference, Hyatt on Capital Square, Columbus, Ohio, May 12-16, 2003.
  63. • P. H. Worley, T. H. Dunigan, "Early Evaluation of the Cray X1 at Oak Ridge National Laboratory," 45th Cray User Group Conference, Hyatt on Capital Square, Columbus, Ohio, May 13, 2003.
  64. • J. B. White III, "Scalable Supercomputer Solving Superconductivity," SC2003, Cray Exhibit Booth, Phoenix Convention Center, Phoenix, Arizona, November 19, 2003.
  65. REFERENCES
  66. OpenMP Architecture Review Board, "OpenMP: A Proposed Standard API for Shared Memory Programming," October 1997. Available from http://www.openmp.org/openmp/mp- documents/paper/paper.ps.
  67. Z. Lin et at., Phys Rev. Lett. 88, 195004, 2002.
  68. P.A. Agarwal, Oak Ridge National Laboratory R.A. Alexander, Oak Ridge National Laboratory E. Apra, Pacific Northwest National Laboratory S. Balay, Argonne National Laboratory A.S. Bland, Oak Ridge National Laboratory J. Colgan, Auburn University E.F. D'Azevedo, Oak Ridge National Laboratory J. Dongarra, University of Tennessee J. B. Drake, Oak Ridge National Laboratory T.H. Dunigan, Oak Ridge National Laboratory T.H. Dunning, Jr., Oak Ridge National Laboratory M.R. Fahey, Oak Ridge National Laboratory R.A. Fahey, Oak Ridge National Laboratory A. Geist, Oak Ridge National Laboratory B. Gorda, Lawrence Berkeley National Laboratory M. Gordon, Pacific Northwest National Laboratory W. D. Gropp, Argonne National Laboratory R. J. Harrison, Oak Ridge National Laboratory R. Kendall, Ames Laboratory D. Keyes, Columbia University D. Kaushik, Argonne National Laboratory M. Krishnakumar, Pacific Northwest National Laboratory P. Luszczek, University of Tennessee A. Mezzacappa, Oak Ridge National Laboratory J.A. Nichols, Oak Ridge National Laboratory J. Nieplocha, Pacific Northwest National Laboratory L. Oliker, Lawrence Berkeley National Laboratory T. Packwood, Cray Inc.
  69. M.S. Pindzola, Auburn University T.C. Schulthess, Oak Ridge National Laboratory H. Simon, Lawrence Berkeley National Laboratory R. Stevens, Argonne National Laboratory J.S. Vetter, Oak Ridge National Laboratory J.B. White, III, Oak Ridge National Laboratory T.L. Windus, Pacific Northwest National Laboratory P.H. Worley, Oak Ridge National Laboratory T. Zacharia, Oak Ridge National Laboratory

Related papers

Scientific Application Performance On Leading Scalar and Vector Supercomputering Platforms
Leonid Oliker

International Journal of High Performance Computing Applications, 2008

The last decade has witnessed a rapid proliferation of superscalar cache-based microprocessors to build high-end computing (HEC) platforms, primarily because of their generality, scalability, and cost effectiveness. However, the growing gap between sustained and peak performance for full-scale scientific applications on conventional supercomputers has become a major concern in high performance computing, requiring significantly larger systems and application scalability than implied by peak performance in order to achieve desired performance. The latest generation of custom-built parallel vector systems have the potential to address this issue for numerical algorithms with sufficient regularity in their computational structure. In this work we explore applications drawn from four areas: magnetic fusion (GTC), plasma physics (LBMHD3D), astrophysics (Cactus), and material science (PARATEC). We compare performance of the vector-based Cray X1, X1E, Earth Simulator, NEC SX-8, with performance of three leading commodity-based superscalar platforms utilizing the IBM Power3, Intel Itanium2, and AMD Opteron processors. Our work makes several significant contributions: a new data-decomposition scheme for GTC that (for the first time) enables a breakthrough of the Teraflop barrier; the introduction of a new three-dimensional Lattice Boltzmann magneto-hydrodynamic implementation used to study the onset evolution of plasma turbulence that achieves over 26Tflop/s on 4800 ES processors; the highest per processor performance (by far) achieved by the full-production version of the Cactus ADM-BSSN; and the largest PARATEC cell size atomistic simulation to date. Overall, results show that the vector architectures attain unprecedented aggregate performance across our application suite, demonstrating the tremendous potential of modern parallel vector systems.

View PDFchevron_right
Evaluation of Cache-based Superscalar and Cacheless Vector Architectures for Scientific Computations
carter john

2003

The growing gap between sustained and peak performance for scientific applications is a well-known problem in high end computing. The recent development of parallel vector systems offers the potential to bridge this gap for many computational science codes and deliver a substantial increase in computing capabilities. This paper examines the intranode performance of the NEC SX-6 vector processor and the cache-based IBM Power3/4 superscalar architectures across a number of scientific computing areas. First, we present the performance of a microbenchmark suite that examines low-level machine characteristics. Next, we study the behavior of the NAS Parallel Benchmarks. Finally, we evaluate the performance of several scientific computing codes. Results demonstrate that the SX-6 achieves high performance on a large fraction of our applications and often significantly outperforms the cache-based architectures. However, certain applications are not easily amenable to vectorization and would require extensive algorithm and implementation reengineering to utilize the SX-6 effectively. * Employee of Computer Sciences Corporation. 1 Also referred to as the Cray SX-6 due to Cray's agreement to market NEC's SX line.

View PDFchevron_right
Performance of Ultra-scale applications on leading vector and scalar HPC platforms
Leonid Oliker

2005

The last decade has witnessed a rapid proliferation of superscalar cache-based microprocessors to build high-end capability and capacity computers primarily because of their generality, scalability, and cost effectiveness. However, the constant degradation of superscalar sustained performance, has become a well-known problem in the scientific computing community. This trend has been widely attributed to the use of superscalar-based commodity components who's architectural designs offer a balance between memory performance, network capability, and execution rate that is poorly matched to the requirements of large-scale numerical computations. The recent development of massively parallel vector systems offers the potential to increase the performance gap for many important classes of algorithms. In this study we examine four diverse scientific applications with the potential to run at ultrascale, from the areas of plasma physics, material science, astrophysics, and magnetic fusion. We compare performance between the vector-based Earth Simulator (ES) and Cray X1, with leading superscalar-based platforms: the IBM Power3/4 and the SGI Altix. Results demonstrate that the ES vector systems achieve excellent performance on our application suite -the highest of any architecture tested to date.

View PDFchevron_right
Impact of Quad-Core Cray XT4 System and Software Stack on Scientific Computation
Heike Jagode

Lecture Notes in Computer Science, 2009

An upgrade from dual-core to quad-core AMD processor on the Cray XT system at the Oak Ridge National Laboratory (ORNL) Leadership Computing Facility (LCF) has resulted in significant changes in the hardware and software stack, including a deeper memory hierarchy, SIMD instructions and a multi-core aware MPI library. In this paper, we evaluate impact of a subset of these key changes on large-scale scientific applications. We will provide insights into application tuning and optimization process and report on how different strategies yield varying rates of successes and failures across different application domains.

View PDFchevron_right
Evaluation of leading scalar and vector architectures for scientific computations
Leonid Oliker

2004

Applications scientists have observed a frustrating trend of stagnating application performance despite dramatic increases in claimed peak performance of high-performance computing (HPC) systems. This effect has been widely attributed to systems composed of commodity components, whose architectural designs are unbalanced and inefficient for large-scale scientific computations. The recent development of parallel vector systems offers the potential to bridge this performance gap for a significant number of scientific codes, and to increase computational power substantially. In order to quantify what a vector capability entails for scientific communities that rely on modeling and simulation, it is critical to evaluate it in the context demanding computational algorithms.

View PDFchevron_right

Related topics

  • Computer Science
  • Parallel Computing
  • Design
  • Architecture
  • Biology
  • Astrophysics
    • 580 California St., Suite 400
    San Francisco, CA, 94104
    © 2025 Academia. All rights reserved