1 / 25

Linux Clusters as an MSC.Nastran High-Performance Computing Platform

Linux Clusters as an MSC.Nastran High-Performance Computing Platform. Chris Smith WW CAE Portfolio Manager Hewlett-Packard c_smith@hp.com +1 (303) 649-5263. Why Linux and Clusters?. Price-Performance Stability Highly scalable from laptops to high-performance cluster supercomputing

tillie
Download Presentation

Linux Clusters as an MSC.Nastran High-Performance Computing Platform

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Linux Clusters as an MSC.Nastran High-Performance Computing Platform Chris SmithWW CAE Portfolio ManagerHewlett-Packardc_smith@hp.com+1 (303) 649-5263

  2. Why Linux and Clusters? • Price-Performance • Stability • Highly scalable from laptops to high-performance cluster supercomputing • Easy and inexpensive to expand • Older, slower, assets can be redistributed to “office” tasks • Supported by all major hardware vendors • Leverage existing processes & UNIX expertise • Enthusiastic supporters, Open Source and GPL

  3. Issues with Linux Clusters? • Lack of knowledge • Lack of applications • Lack of proven track record for production work • Lack of application vendor and hardware vendor support • Integration work required

  4. The HP/MSC Linux Cluster Initiative • Working with MSC, deliver a turn-key, production quality, high-performance computational server environment, based on Linux workstations, as a platform for running MSC and other CAE applications

  5. The HP/MSC Linux Cluster InitiativePress Release • MSC.Software Announces High Performance MSC.Linux Appliance May 12, 2000 9:00 AM EDTLOS ANGELES and COSTA MESA, Calif.--(BUSINESS WIRE)--May 12, 2000 MSC.Software Will Deliver Turnkey MSC.Linux Cluster Solutions, Using Hardware From HP, to Satisfy the Market Demand for Extreme Performance Computing MSC.Software Corporation (MSC) (NYSE:MNS) today announced plans to develop a range of MSC.Linux cluster appliances, based on Hewlett-Packard Company (NYSE: HWP) platforms, to meet the high performance requirements of the world wide scientific markets. This HP hardware based MSC.Linux clusters will be marketed by MSC.Software. The initial MSC.Linux appliance is a turnkey cluster for MSC.Software's highly parallel simulation applications. "Our customers, using the MSC.Linux cluster solution, can now conveniently embrace tremendous price-performance benefits when dealing with their most demanding computational situations," said Frank Perna, Jr., chairman and CEO of MSC.Software. "The MSC.Linux appliance, directed at the scientific community which includes all the product development activities of our customer base who represent the top 5000 manufacturers in the world, immensely expands our market opportunities." "MSC.Linux appliances offer scientific and engineering organizations a powerful and flexible platform for solving complex problems," said Knute Christensen, mechanical design marketing manager for HP's Technical Computing Division. "At HP we are committed to providing solutions that meet customer requirements, and we're excited to collaborate with MSC.Software on this innovative initiative."

  6. High Performance Clusters • Cluster architecture • Master (head) node • Connected to LAN • KVM Switch • Compute nodes • optimized for application • Communication backbone • Fast Ethernet, Myrinet ... • Switch • other • Software • System Administration • Queue System • MPI

  7. MSC.Nastran Configuration • Master Node – HP Visualize Personal Workstation • 2 Intel Pentium III Processors (866 MHz currently) • 512MB (1 GB recommended) • RDRAM,133 MHz FSB • 3x10k RPM disks • Ultra2 (or faster) SCSI disk controller • 2 Network Cards (100Mbs Ethernet) • HP’s Performance Leading FX6 Graphics subsystem • CD-ROM, Floppy, Keyboard, Monitor, Mouse

  8. MSC.Nastran Configuration • Compute Node – HP Visualize Personal Workstation • 2 Intel Pentium III Processors (866 MHz currently) • 1GB Memory • RDRAM 133 MHz (or faster) FSB • 3x10k RPM disks • Ultra2 (or faster) SCSI disk controller, RAID-0 • Network Card (100Mbs Ethernet) • Elsa Synergy II Graphics board • CD-ROM drive (optional), Floppy

  9. MSC.Nastran Configuration • Other Hardware • Network switch and cables • User interface switch • (Optional) Uninterruptible Power Supply (UPS) for the master node

  10. Linux Clusters • LINUX Cluster Performance

  11. MSC.Nastran 70.7 DMP PerformanceDirect Frequency: SOL108 (XLTDF): Car Body Problem XLTDF (Car Body: SOL108 (XLTDF): 529,257 Ndof: 450 Mb: SCR Disk 5000 Mb: Total I/O 209 Gb: 32 Frequency Increments) Linux Cluster: 1 CPU=1 NodeSGI Origin 2000: 2 CPU=1 Node HP N4000: 8 CPU=1 Node

  12. MSC.Nastran 70.7 DMP PerformanceDirect Frequency: SOL108 (XLTDF): Car Body Problem XLTDF (Car Body: SOL108 (XLTDF): 529,257 Ndof: 450 Mb: SCR Disk 5000 Mb: Total I/O 209 Gb: 32 Frequency Increments) Linux Cluster: 1 CPU=1 NodeSGI Origin 2000: 2 CPU=1 Node HP N4000: 8 CPU=1 Node

  13. MSC.Nastran 70.7 DMP PerformanceLinear Statics: SOL101 (XLRST): Engine Problem XLRST (Engine: SOL101 (XLRST): 739,651 Ndof: 80 Mb: SCR Disk 4000 Mb: Total I/O 18 Gb) Linux Cluster: 1 CPU=1 NodeSGI Origin 2000: 2 CPU=1 Node HP N4000: 8 CPU=1 Node

  14. MSC.Nastran 70.7 DMP PerformanceLinear Statics: SOL101 (XLRST): Engine Problem XLRST (Engine: SOL101 (XLRST): 739,651 Ndof: 80 Mb: SCR Disk 4000 Mb: Total I/O 18 Gb) Linux Cluster: 1 CPU=1 NodeSGI Origin 2000: 2 CPU=1 Node HP N4000: 8 CPU=1 Node

  15. MSC.Nastran 70.7 DMP PerformanceLinear Statics: SOL101 (Model DF3): Axle Problem DF3 (Axle: SOL101: 321,138 Ndof) HP VPW running Linux: 1 CPU

  16. MSC.Nastran 70.7 DMP PerformanceLinear Statics: SOL101 (FF3): Axle Problem FF3 (Axle: SOL101: 429,118 Ndof) Linux Cluster: 1 CPU=1 Node

  17. Performance Summary • Excellent Price/Performance compared to traditional UNIX server solutions • Excellent solution for Frequency Response and Optimization problems - high scalability • Fair solution for Linear Statics • Traditional Unix Server solutions still provide fastest time to solution • Traditional Unix Server/Workstation solutions can handle larger problems and more users due to more capable proprietary architectures

  18. Why MSC.Linux? • Expertise in: • Numerical performance • Distributed & parallel memory • Application tuning/porting • Kernel tuning • Software Configuration Mgmt. & Testing • Extreme performance computing presence • Strong HP hardware partner relationship

  19. Why HP Visualize Personal Workstations? • Designed for Technical Computing: fastest architecture design • Best 3D graphics systems: HP FX graphics fastest in industry and with CAE applications • Same high reliability as all HP computing systems • Part of a complete solution portfolio • Experienced supplier: Support for Linux from the beginning • Best CAE partner relationships

  20. MSC.Linux Business • Services • Porting • Training and Support • Web Portal • Software • Applications • High Performance Computing • Appliances • Engineering Tools • High Performance Clusters • New Opportunities

  21. MSC.Linux Services • Third Party Application • Porting • Tuning • Parallel Processing • Training • Support

  22. HP and MSC.Linux Cluster Solution Partnership Benefits • Simplified ordering: Customer places one order with MSC. • Immediate productivity: MSC delivers turn-key MSC Linux Cluster platform based on HP Linux workstations • HP VPW’s are specifically designed and tuned for high-performance computing • One stop support: MSC supports full turn-key solution. • HP/MSC work as a single team: HP provides support to MSC for issues/problems

  23. HP and MSC.Linux Cluster Initiative Agreement • HP and MSC entered into a unique agreement! • All MSC primary development and testing is done on the latest HP provided, Intel-based, Visualize Personal workstations • MSC delivers a complete turn-key solution only on HP VPW’s. This ensures customers get a fully supported, production quality, Linux Cluster environment

  24. HP VPW/MSC.Linux: Customer Benefit • Compelling price-performance alternative • Tier II Supplier example: • Desire improved throughput, limited budget • Intel-based cluster versus Unix cluster • 5X reduction in h/w costs, comparable throughput • Most customer applications on Unix today • Linux port easy versus NT • Retain benefits of Unix • Solid • Network performance • Throughput

  25. For More Information on turn-key HP/MSC Linux Clusters….. • HP Contacts • Chris Smith: c_smith@hp.com: +1 (303) 649-5263 • Didier Large: didier_large@hp.com: +33 (1) 69 64 84 63 • Wolfgang Kocher: wolfgang_kocher@hp.com: +49 (7031) 14 7793 • Jan Perry: jan_perry@hp.com: +44 (1344) 36 5935 • MSC.Linux Contacts • Bob Jones: bob.jones@mscsoftware.com: +1 (714) 444-5074

More Related