System News for Sun Users

Sockets Direct Protocal Gives InfiniBand the Advantage in HPC Java Applications

2011-10-31T16:00:00+01:00

In his post Enabling HPC with InfiniBand in Java Applications Ricardo Ferreira argues that the maturity and e bandwidth advantages of InfiniBand give it an edge over Fibre Channel and Ethernet as the choice for a network fabric. He argues that InfiniBand's capability to move data directly from the memory of one computer to another, bypassing the operating systems of both in the operation, gives it a performance advantage in HPC settings. It is the enabling of Sockets Direct Protocol (SDP) support in InfiniBand, he concludes, that is the key to maximizing the performance of Java applications.

Lustre Is the Logical File System for Exascale

2011-07-05T16:00:00+01:00

The clock is running on the goal to stage an exascale technology demonstration capable of handling a peak of 400 petaflops by 2015. This is more data than a single stream can deliver, so the search is on in the HPC community for a solution to these I/O demands. The hardware approach, writes Brent Gorda, Whamcloud CEO and President, in his HPCWire article "Why Lustre Is Set to Excel in Exascale," -- boosting performance on the single unit and going widely parallel -- will work until power limitations and parallelism issues become an impediment. Gorda enthusiastically nominates the Lustre file system to fill the software side of the equation. Lustre, he continues, based on an object store, has the necessary fundamental architecture for exascale, which the HPC file system technologists think will be based on an object store approach. It might work, Gorda continues, to introduce new file types to Lustre that will provide exascale object storage semantics internally. While this will require development of the underlying object model, Gorda points out that it holds the promise that the same file system will be able to support the full range of applications from (legacy) POSIX through to exascale. Finally, under development since 1999, Lustre has the maturity and stability that developers seek as the starting point for rapid and diverse development, Gorda concludes.

Fujitsu/RIKEN "K computer" Tops TOP500 List

2011-06-20T16:00:00+01:00

The "K Supercomputer," being jointly developed by RIKEN and Fujitsu, has achieved a "world's best" performance on the LINPACK benchmark program," logging 8.162 petaflops (quadrillion floating-point operations per second), and a computing efficiency rating of 93.0% to place it at the head of the TOP500 list.

The K computer system, currently in the configuration stage, has 672 computer racks equipped with a current total of 68,544 CPUs. When it goes into full shared-use service in November 2012, as a part of the High-Performance Computing Infrastructure (HPCI) initiative led by Japan's Ministry of Education, Culture, Sports, Science and Technology (MEXT) the K computer will comprise over 800 computer racks'each equipped with ultrafast and energy-efficient CPUs'that access into a network capable of an immense amount of interconnectivity.

Sun Blade Servers Provide Foundation for Red Sky/Red Mesa Supercomputer

2011-04-28T16:00:00+01:00

RedSky/Red Mesa one of the world's fastest and most energy efficient supercomputers, housed at Sandia National Laboratories, is powered by Oracle's Sun Servers that include Sun Blade Modular Systems that employ the Sun Cooling Door System and Oracle's network fabric as foundational technologies. With this architecture in its supercomputer datacenter, Sandia has realized reductions in energy consumption and cost by 77 percent and achieved a peak performance of more than 500 trillion mathematical operations per second. The Sun Cooling Door removes 90 percent of the heat load on servers and saves 5 million gallons of water per year. Using the Lustre file system running on Sun storage servers and on Sun disk arrays, Sandia increased their storage capacity and IO performance using an Infiniband connected file system with over 6PB in capacity that delivers more than 20GB/Sec in throughput.

Future Direction for the Oracle Grid Engine (OGE)

2011-01-28T17:00:00+01:00

The future of the Oracle Grid Engine product includes three major goals, according to Dan Templeton, whose blog contains a link that leads viewers to a replay of the webcast he did with Oracle's marketing director Moe Fardoost. The three goals articulated in the webinar are core infrastructure and feature improvements; tighter integrations with other Oracle products; and a richer cloud feature set. Registration and sign-in required.

Oracle Grid Engine: Changes for a Bright Future at Oracle

2011-01-06T17:00:00+01:00

With the coming of the new year, a new chapter is opening in Oracle Grid Engine's life. Oracle is handing off maintenance of the open source code base to the Open Grid Scheduler project hosted on SourceForge. Dan Templeton reports that the transition will allow the Oracle Grid Engine engineering team to focus their efforts more directly on enhancing the product.

Oracle Grid Engine Podcast

2010-11-15T17:00:00+01:00

A podcast conversation between Daniel Templeton and Bob Rhubart about the Oracle Grid Engine is available on the OTN Architect Community's ArchBeat. Rhubart also lists a number of links that reflect Templeton's long history as an active member of the Grid Engine community.

An Orphan No Longer, Lustre Gets Backing of Non-Profit Corporation

2010-10-25T16:00:00+01:00

Lustre has been adopted by a consortium of some of the biggest names in HPC computing, reports Michael Feldman in HPCwire. Open Scalable File Systems, Inc. (OpenSFS), backed by Cray, Data Direct Networks, Lawrence Livermore National Laboratory (LLNL) and Oak Ridge National Laboratory (ORNL), is the adoptive entity. The organization's mission is to bring together the stakeholders for high-end scalable file systems and provide a formal structure for moving the associated software forward, Feldman writes.

The Latest Intel Xeon and AMD Opteron Processors

2010-04-06T16:00:00+01:00

HPCwire Editor Michael Feldman gives an overview of the latest Intel Xeon and AMD Opteron processors - Intel's 6-core Westmere EP and 8-core Nehalem EX CPUs, and AMD's 12-core Magny-Cours processor - and examines the different approaches each company is using to appeal to the high-end server market.

HPC News Bites

2010-03-04T17:00:00+01:00

HPC in 2010
Lustre File System Update
Oracle to Participate in ISC and HPC Consortium
NFS Tuning for HPC Streaming Applications
HPC Application Performance with Flash Storage
How Flash Storage Fits into HPC

http://blogs.systemnews.com/

New HPC Installation Goes Online at University of Strathclyde

2010-02-12T17:00:00+01:00

With its 1,088 computing cores, 100TB of data storage tied to a quad data rate Infiniband network and a quoted performance of 13 Teraflops at peak -- equivalent to up to thirteen trillion operations per second -- the new High Performance Computer at Scotland's University of Strathclyde, implemented by supplier Esteem Systems using Sun technology, enables the university's Faculty of Engineering to address complex problems in materials, fluid dynamics and design.

A Solution to HPC Application Performance Improvement

2010-02-02T17:00:00+01:00

The Sun Storage Solutions white paper "Turbo-charging HPC Application Performance with Flash Storage: Changing the Focus from Processing to Moving Data Faster" contends that, with ever more CPU cores and more GFLOPS per core, HPC application performance will be determined by the ability to move data in and out of fast processors. This paper examines how Flash technology can be deployed and the effect it will have on HPC workloads. It examines various Flash technologies and compares their performance on leading HPC application benchmarks. Finally, the performance characteristics of the Flash-based Sun Storage 7000 Unified Storage System is described.

The Virtues of Virtualization

2010-01-29T17:00:00+01:00

Following on his talk "HPC Trends and Virtualization," a presentation given at Sun's HPC Consortium in Hamburg recently, Josh Simons goes further with his blog post "Virtualization for HPC: The Heterogeneity Issue," a subject to which he takes an even-handed approach, conceding at the outset that " ... while heterogeneity is either desirable or to be avoided, depending on your viewpoint, virtualization can help in either case."

Sun HPC Solution Ushers in Teraflop-scale Applications for Brazilian Computing Grid

2010-01-27T17:00:00+01:00

High-performance computing (HPC) initiative Petroleo Brasileiro (BR) Network was created to handle computer simulations to aid oil production in Brazil's pre-salt basin. An HPC grid was required to handle the complex project with the goal being a high-performance, affordable solution that could expand easily over the next few years. Sun, IBM, Dell, Bull, SGI, and HP were consulted. After an extensive review process, the group decided that Sun would best meet its goals for performance, scalability, energy efficiency, and cost, the latter of which was 30% less than an offering from the closest competing vendor.

Sun HPC Consortium 2009 Videos and PDFs

2010-01-11T17:00:00+01:00

Videos and PDFs are now available from the Sun HPC Consortium meeting held Nov. 14-15, in Portland, Ore. The 20-plus presentations include the opening talk from Sun VP of HPC and Cloud Computing Marc Hamilton, panel discussions on meteorology and climate, as well as high performance data management, HPC software sessions, and much more.

Sun Constellation System at the Heart of CLUMEQ HPC Implementation

2009-12-29T17:00:00+01:00

CLUMEQ is a supercomputing consortium of universities in the province of Quebec, Canada. It includes McGill University, Universite Laval, and all nine components of the Universite du Quebec network. CLUMEQ supports scientific research in disciplines such as climate and ecosystems modeling, high energy particle physics, cosmology, nanomaterials, supramolecular modeling, bioinformatics, biophotonics, fluid dynamics, data mining and intelligent systems. Sun is providing the technology for this HPC implementation.

"Sun Business Ready HPC for Altair RADIOSS"

2009-12-23T17:00:00+01:00

In his Sun BluePrints paper, "Sun Business Ready HPC for Altair RADIOSS: Best Practices for Altair RADIOSS Deployments," Dr. Gunter Roeth, of Sun ISV Engineering provides insight into constructing an optimal configuration for Altair RADIOSS car crash simulations using a high-performance cluster of Sun systems equipped with x86/x64 processors. The paper points out how HPC solutions often consist of servers equipped with Intel Xeon or AMD Opteron processors that are clustered together using high-speed, low-latency interconnects. In many cases, Roeth writes, these high-performance clusters can match and even surpass the throughput capabilities of vertically-scalable proprietary solutions.

Configuring Sun Ops Center for the Sun HPC Software Framework

2009-12-22T17:00:00+01:00

Anyone interested in an integrated open-source software solution for Linux-based high-performance computing (HPC) clusters running on Sun hardware may want to read the Sun BluePrints article "Using Sun Ops Center with Sun HPC Software, Linux 2.0." Authors Mike Berg and Zhiqi Tao describe how to set up a Sun Ops Center server, provision the HPC cluster nodes (head node, Luster file system servers, and compute nodes), and configure system services.

South African Government Implements 2nd-Gen Sun Constellation System

2009-12-10T17:00:00+01:00

When the South African government launched its largest (both in South Africa and the continent as a whole) supercomputer at CHPC in Cape Town, Marc Hamilton, Sun VP of Americas Systems Practice was there to interview Naledi Pandor, South African Minister of Science and Technology, on the system and the Sun technology at its core - the second generation Sun Constellation System.

Sun HPC Recent Developments

2009-11-18T17:00:00+01:00

The latest Top500 list published in November shows three of the top five supercomputers based on Intel Nehalem CPUs were Sun Constellation Systems. Sun also doubled its number of entries since the June 2009 list with a total of 11 deployments providing nearly 2 PetaFLOPS (PFLOPS). Additionally, Sun announced new products and technologies for high performance computing (HPC), an upgrade to the Sun Storage 7410 Unified Storage System, new HPC benchmarks and new HPC customers.

HPC Profiling with the Sun Studio Performance Tools

2009-11-10T17:00:00+01:00

Marty Itzkowitz and Yukon Maruyama describe how to use the Sun Studio Performance Tools to understand the performance issues in single-threaded, multi-threaded, OpenMP and MPI applications, and the techniques used to profile them. Presented at the Third Parallel Tools Workshop held in Dresden, Germany back in September, this technical paper outlines techniques implemented in the Sun Studio Performance Tools to profile HPC applications. The 30-page PDF is hosted on Sun's developer's site.

Optimizing Sun HPC Hardware for ANSYS FLUENT

2009-10-19T16:00:00+01:00

The Sun BluePrints article "Sun Business Ready HPC for ANSYS FLUENT: Configuration guidelines for Optimizing ANSYS FLUENT Performance" by William Aiken explores the parameters to be considered when defining a processing configuration for ANSYS FLUENT. The number of cores (independent computing elements), amount of memory, and interconnect options are discussed. For situations where a significant amount of file I/O will occur, an overview of parallel and serial file system operations and storage subsystems is also provided.

Bechtolsheim Shares His View on the Rise of 10GbE

2009-09-30T16:00:00+01:00

Andy Bechtolsheim's new firm, Arista Networks, is well positioned to capitalize on the burgeoning demand for 10GbE ethernet products that HPC users find essential to their operations. He told The Register's Timothy Prickett Morgan that "it is possible for some 25 percent of all servers in the installed base to have 10GbE ports by the year 2011." Said Bechtolsheim, "It is easier to forecast 10GE ports than revenues. This is one of the few areas in IT with predictable growth."

Iowa State's Newest Supercomputer Cystorm

2009-09-16T16:00:00+01:00

Sun, along with the National Science Foundation and Iowa State University, has facilitated the institution's second supercomputer known as Cystorm that boasts a peak performance of 28.16 trillion calculations per second. The 3,200 computer processor cores powering Cystorm provide five times the peak of the university's first supercomputer CyBlue, an IBM Blue Gene/L, that uses 2,048 processors to do 5.7 trillion calculations per second.

Sun Grid Engine Inspect

2009-09-15T16:00:00+01:00

Inspect is a new monitoring tool included in Sun Grid Engine (SGE) Update 3. It is a flexible interface for viewing current and historical data on SGE cluster(s) and the Service Domain Manager. The new GUI-based monitoring tool is developed purely in Java and uses JMX. This means an SGE installation has to enable JMX. Users interested in using Inspect must have version 6+ of the following software: Sun JDK, OpenJDK, IcedTea, and Apple JDK.

HPC News Bites

2009-09-08T16:00:00+01:00

Free Training on Parallel Programming
HPC Virtual Conference
Differences Between MPI and OpenMP
eSeminar Panel Discussion: Flash in HPC
Sun HPC Community Wiki
Sun Studio Software: Developing OpenMP 3.0 Applications
RADIOSS Public Benchmark Test Suite: QDR vs DDR

http://blogs.systemnews.com/

Replacing HDDs with SSDs to Improve HPC Systems I/O Performance

2009-08-11T16:00:00+01:00

An argument is made for the use of SSD devices in benefiting HPC applications with large I/O components in a Sun BluePrints Online article by Lawrence McIntosh in Sun's Systems Engineering Solutions Group and Dr. Michael Burke in Sun's Strategic Applications Engineering. The two compare traditional hard disk drives (HDDs) and the newer solid state drive (SSD) technology in high-performance computing (HPC) applications. Application-based benchmarking and storage performance benchmark testing are used and demonstrate significant benefits in the use of SSD devices for HPC applications with large I/O components, the authors conclude.

Sun HPC Software, Linux Edition 2.0

2009-08-10T16:00:00+01:00

The Sun HPC Software, Linux Edition 2.0 is an integrated, open-source software solution for Sun HPC clusters. A ready-made framework of software components turns a bare-metal system into a running HPC cluster. This latest edition features support for SUSE Linux Enterprise Server (SLES) 10sp2, Sun supported cluster software (Sun ClusterTools, Sun Studio, Sun Grid Engine), Lustre 1.8.0.1, MPI libraries pre-built with Pathscale and Sun Studio x86 compilers and more.

Sun HPC ClusterTools 8.2 and 8.1

2009-07-29T16:00:00+01:00

Sun HPC ClusterTools 8.2 is the latest version of the integrated toolkit for Message-passing Interface (MPI) applications that runs on high performance clusters and SMPs. New features available in version 8.2 include additional compiler support, suspend / resume support, improved intra-node shared memory performance and scalability, plus more. A video presentation on the prior version (8.1) is available which covers some of the general features of Sun HPC ClusterTools and provides information on the community along with resource links.

Configuring Systems for Very High Bandwidth I/O

2009-07-16T16:00:00+01:00

The Sun BluePrints Online paper "Configuring Systems for High Bandwidth I/O" by Kevin Colwell and Carolyn Bumatay describes a recipe for configuring systems to support the very high bandwidth I/O capabilities needed by such data-intensive applications as operational intelligence and surveillance, epidemic trend analysis and prediction, failure analysis of aircraft and ships, predictive traffic management, weather and ocean forecasting, virtual design, astronomy, human genomics, and other scientific disciplines that are data-intensive, and depend on streaming read/write I/O performance.