|
 |
 |
|
 |
 |
|
 |
 |
 |
|
|
 |
Computational Fluid Dynamics (CFD) is an integral part of the engineering design and analysis process for many companies and research organizations. CFD offers the ability to accurately predict the performance of new designs or processes before they are ever manufactured or implemented. CFD applications work by dividing the region of interest into a large number of cells or control volumes. This is referred to as the "mesh" or "grid". Within each cell, the partial differential equations describing the fluid flow, known as the Navier-Stokes equations, are rewritten as algebraic equations that relate the pressure, velocity, temperature and other variables to the values in the neighboring cells. These equations are then solved numerically yielding a complete picture of the fluid flow down to the resolution of the grid. The resulting image is an accurate depiction of the fluid dynamics surrounding the grid.
PSSC PowerWulf Clusters are custom configured, turn-key solutions that are optimized for CFD applications. Our PowerWulf Clusters are used by several CFD application developers for inhouse testing and development including Metacomp Technologies and Flow Sciences. According to Sukumar Chakravarthy, Metacomp Technologies President, "PSSC keeps tabs on all changes in technology so we don't have to, including hardware, chip, interconnect and operating system developments."
FEATURE OVERVIEW
TURN KEY SOLUTION
UNLIMITED LIFETIME SUPPORT
PERFORMANCE TUNING
EXPANDABILITY
FLEXIBILITY
STABILITY
What do you need out of your Linux Supercomputer? Reliability, performance, ease of use and excellent support are probably on the top of your list. PSSC Labs understands the significance of your work and your need for the most powerful, reliable Linux Supercomputer possible. PowerWulf Clusters are the answer to your Linux Supercomputer needs. PowerWulf Clusters are elegantly configured, easy to use Linux Supercomputer solution that is ready for your use right out of the box.
You will not need to consume valuable time configuring your Linux Supercomputer. Instead PSSC Labs Linux Experts will deliver your supercomputer with CBeST, the Complete Beowulf Software Toolkit. CBeST is much more than just cluster management. CBeST is a detailed process designed to deliver exactly what you need out of your Linux Supercomputer: reliability, performance, ease of use and excellent support. Our industry recognized CBeST software toolkit includes all cluster management tools necessary to manage, monitor, maintain and upgrade your cluster. All PowerWulf Clusters are fully configured with CBeST and thoroughly tested before leaving our facility. Your PowerWulf Cluster includes unlimited liftetime support. We are available to help answer your questions regarding PowerWulf cluster hardware and CBeST software. For complete CBeST details please visit (this will link to the product - software - cbest page)
Before beginning CBeST installation, PSSC Labs Cluster Technicians will work with you to properly configure the network settings, file system configuration and cluster nomenclature. This allows PSSC Labs to deliver the most TURN KEY solution possible.
All PowerWulf Clusters include unlimited lifetime support at NO additional cost. You are welcome to call or email any question you have regarding cluster hardware and software directly to the PSSC Labs Cluster Technician responsible for configuring your cluster. This ensures that you receive the best possible technical support in an immediate fashion.
PowerWulf Clusters are optimized for maximum performance. Our superior experience allows PSSC Labs Cluster Technicians to tune performance for maximum results. BIOS settings, network configurations, memory configuration, file system and all other possible areas for performance improvement are carefully reviewed and scrutinized. This ensure that each PowerWulf Cluster provide the highest possible computing power. Before any PowerWulf Cluster ships it must pass our internal performance standards.
Easily add more nodes, memory, network interconnects to your PowerWulf cluster. With CBeST software tools expanding the size of your cluster is fast and easy. PSSC Labs Cluster Technicians will be available to help every step of the way at NO additional cost.
PSSC Labs offers the most hardware and software choices available. You can customize your PowerWulf Cluster to meet your needs and budget. Chassis, processor, memory and networking options are all available. Select the operating system, MPI version and file system you need.
PSSC Labs testing procedure is considered one of the most stringent in the industry. Each PowerWulf Cluster undergoes a minimum of seven days continuous testing and retesting. For complete testing detail procedures please see below.
- Assemble the system, update Bios to latest version, set BIOS to match the rest of the hardware. (approx 1 hour)
- Run Memtest to find MEM, CPU or MOBO errors. (approx 1-2 days)
- Run STRESS to find MCE errors, (approx 3 days)
- Power down, unplug, cool system. (approx 3 hours)
- Power up, run Memtest. (approx 1 hour)
- Power down, unplug, cool system (approx 3 hours)
- Power up, run Memtest. (approx 1 hour)
- Power down, unplug, cool system. (approx 3 hours)
- Power up, run Memtest. (approx 1 hour)
- Power down, unplug, cool system. (approx 3 hours)
- Install HPC Linpak and run over entire cluster (approx 3 days)
- Install OS of customer choice and run benchmark test (approx 3 days)
- Power down, unplug & ship to end user.
Some companies may take testing shortcuts or not test the cluster as a complete solution. Ultimately these shortcuts will cost you a significant amount of time, energy and money. PSSC Labs eliminates all of the headaches through our testing procedure.
|
COMPUTING IN THE FAST LANE WITH METACOMP TECHNOLOGIES AND AMD OPTERON™ PROCESSORS
"High performance computing is essential for designing and supporting our Computational
Fluid Dynamics and other programs, and 64-bit Dual-Core AMD Opteron™ processor-based
clusters help us to reduce development time and improve turnaround for support issues.
AMD's 64-bit processor technology gives us the ability to support large computational models
which demand high performance, large memory addressability, and economical space, power,
and cooling requirements. In addition, HyperTransport™ technology has enabled us to link two
motherboards together to create a 16-core stand-alone system with128GB of RAM for a fast
and highly efficient pre-processing system."
-- SUKUMAR CHAKRAVARTHY, PRESIDENT, METACOMP TECHNOLOGIES
CHALLENGE
Complex, performance-intensive computational application development and support required higher-performance 64-bit
processing Company needed the ability to scale processor cores and memory in cluster-based systems without exceeding power,
space, and cooling budgets.
SOLUTION
Working with system integrator PSSC Labs, Metacomp Technologies upgraded to AMD Opteron™ processor-based servers in 2004, and has continued to build new, AMD- based clusters using 64-bit, dual-core technologies.
IMPACT
Metacomp Technologies has cost-effectively built clusters of up to 64 cores that enable it to support even the largest and very
compute-intensive customer applications with quick turnarounds, and without exceeding its power, space, and cooling capabilities.
ORGANIZATIONAL PROFILE
Metacomp Technologies is a leading provider of Computational Fluid Dynamics (CFD) and other analytical software tools for major
aerospace, defense, and automobile manufacturers as well as other companies that require complex modeling during research
and development.
Metacomp Technologies' CFD ++ product is setting the pace in the CFD marketplace with its unique combination of accuracy, reliability and performance..
 |


Metacomp Technologies relies on PSSC Labs for all their high performance computing needs. PowerWulf Clusters and various PowerServe Servers are used by Metacomp Technologies for inhouse application development and client support.
|
 |
- 3U Rackmount Chassis with Redundant UL Approved Power Supply
- (12) Hot Swappable SATA II Drive Trays / Carriers
- Dual Opteron Motherboard with AMD 8000 Chipset / 8X Dimm Sockets / Dual Core Ready
- (2) AMD Opteron 2356 Quad Core Processors with Heat Sink and Fan (8 Processor Cores)
- 32.0 GB DDR II PC3200 ECC Registered System Memory
- (2) 200GB SATA II 7200 RPM Hard Drives (Configured for Raid 1: Cluster Tools and OS)
- 3Ware 2 Port SATA Raid Controller
- (12) 1.0TB SATA II 7200 RPM Hard Drives (Configured for Raid 5 : Approx 10TB Usable Space)
- 3Ware 12 Port SATA Raid Controller
- 3Ware Battery Back Up Unit
- DVD RW + / - / Rom / CD RW / CD Rom Combo Drive
- Mellanox Infiniband DDR Network Adapter / PCI Express x 8
- (2) 10/100/1000 BaseT Gigabit Network Adapter
- SVGA Video Adapter
- 104 Key Keyboard (non rackmount)
- Logitech 3 Button Mouse (non rackmount)
- Rackmount Sliding Rails
- Redhat Linux HPC Latest PSSC Labs Supported Version Installed to Your Specifications and Configured for Beowulf Cluster
- Qty of Head Nodes: 1
- 1U Rackmount Chassis with UL Approved Power Supply
- Dual Opteron Motherboard with AMD 8000 Chipset / 8X Dimm Sockets / Dual Core Ready
- (2) AMD Opteron 2356 Quad Core Processors with Heat Sink and Fan (8 Processor Cores)
- 32.0 GB DDR II PC3200 ECC Registered System Memory
- 80 GB SATA 7200 RPM Hard Drive
- Mellanox Infiniband DDR Network Adapter / PCI Express x 8
- 10/100/1000 BaseT Gigabit Network Adapter on Motherboard
- SVGA Video Adapter
- Rackmount Sliding Rails
- Redhat Linux HPC Latest PSSC Labs Supported Version Installed to Your Specifications and Configured for Beowulf Cluster
- Qty of Slave Node: 23
Custom installation of CBeST v. 3.0 Beowulf cluster software including:
- Setup of OS including Custom Kernel for Performance Optimization
- Message Passing Libraries (MPI CH)
- Open PBS Batch Scheduling System or Sun Grid Engine
- GNU Scientific Compilers
- LAPACK Libraries, BLAS Libraries
- Ganglia cluster monitoring utility and Graphical User Interface tools
- System Imager for remote installation/update of software across the network
- Custom Script for Node Reboot
- Custom Script for Node Shutdown
- LM Sensors monitoring for cpu temp and fan speed (when supported)
- Security patches including Preconfigured Port Mapper, IP Chains/Tables and Security Errada Upates (Redhat) are included for protection
PSSC Labs CPR Complete PowerWulf Recovery System includes PSSC Labs custom configured hardware and software for immediate recovery of PowerWulf cluster to factory installed condition. CPR includes full documentation and technical support from PSSC Labs.
PSSC Labs Custom Operation Manual includes installation instructions, hardware testing guide, MPI operation instructions, Node recovery/addition instructions using System Imager, Ganglia cluster management operation guide and user account setup information.
- 24 Port Infiniband Network Switch with DDR Support
Qty: 1
- Infiniband Network Cable
Qty: 24
- 24 Port 10/100/1000 BaseT Gigabit Network Switch
Qty: 1
- CAT 6 Network Cable
Qty: 25
- 42U Rack Unit (on casters)
Qty: 1
- 6 Outlet 20 Amp / 110V PDU
Qty: 6
|
|
 |
 |
|
 |
 |
|
 |
 |
 |
|
 |
|
|
 |
|