SC11 is the international conference for High Performance Computing, Networking, Storage and Analysis
Advanced Cluster Systems (http://advclustersys.com) will present the Supercomputing Engine Technology (SET), a general-purpose library enabling quick adaptation of new and legacy mainstream software applications to multicore, clusters, and supercomputers without complex parallel compilers or parallel languages or 'breaking' such applications. The company's patent-pending technology provides infrastructure for the message-passing and parallel-computing patterns most useful for mainstream applications, especially those with modular structure. SET, adopting the paradigm of the industry-standard Message-Passing Interface (MPI), is designed for mainstream software writers to achieve practical parallelism quickly. Applying SET to Wolfram's Mathematica parallelised that large mainstream application, without changing any Mathematica code, in one man-month, creating an industry-disrupting product. The company will present SET's API, structure, and supporting Unix- and Linux-compatible technologies.
Advanced Clustering (advancedclustering.com) provides highly customised turnkey cluster systems to organisations with specialised computing needs. It develops systems for some of the most prestigious universities, national laboratories, military installations, and corporations in North America. In business since 2001, Advanced Clustering has earned a reputation as one of the nation's premiere providers of high performance computing systems.
Altair (www.altair.com), the digital simulation experts, will demonstrate how it solves the world's most complex problems through its creation of HPC simulation apps, the delivery of HPC management tools, and the use ofapps/tools to build real products. PBS Professional is the 'hub' for its HPC solutions, manages and schedules the process of big data, providing users with optimised resource utilisation, maximum security and reliability. Altair’s HPC simulation solver, Radioss, is powered by PBS Professional’s ability to minimise job turn-around time. Altair's user-centric portals increase productivity, allowing users to focus on work at-hand. Compute Manager allows users to run, monitor, and manage workloads on distributed resources remotely, while PBS Desktop, launches jobs from the workstation to any server running PBS Professional. Finally, HyperWorks On-Demand, the company's HPC cloud computing offering, is built on Altair's own dedicated and secured infrastructure.
Appro (www.appro.com) is a leading developer of innovative supercomputing solutions and continues to produce award-winning products. At this year’s SC’11, Appro will showcase its next generation Xtreme-X Supercomputer featuring latest cutting-edge cluster technologies and future processors from AMD, Intel and Nvidia wrapped with the latest release of the Appro’s Cluster Engine management software. The Xtreme-X supercomputer is a highly scalable architecture that groups HPC servers together into a unified Scalable Unit (SU) that can be managed as a standalone supercomputer. This system is ideal for medium to large-scale deployments demanding HPC applications requiring high availability and flexibility. ACE Management Software was designed from the ground up and is tightly integrated with the Appro Xtreme-X Supercomputer to provide complete lights-out remote management capability. ACE offers easy to use graphical and command line management interface allowing for precise control over the Xtreme-X from any location while not risking any loss of performance.
The Bergen Language Design Laboratory (BLDL) (bldl.ii.uib.no) at the Department of Informatics, University of Bergen is dedicated to experimenting with programming concepts, providing these as tool-supported language constructs, and show their utility by developing proper applications in the new language. It has a strong focus on the high performance domain.
Among its achievements are: programming models for controlled usage of resources, e.g., placement of computational steps in space (which core) and time (which time-step) on GPU, multicore and FPGA; and coordinate-free numerics for flexibility (portability and extensibility) in exascale software.
To get there, BLDL also has a strength in technologies for development and validation of APIs, for achieving precision in API descriptions thus avoiding misunderstandings and ill-conceived library interfaces. Its approach to provide policies for precondition checking and error handling helps separate these security issues from the 'business logic' of the software.
Boston (www.boston.co.uk) will be presenting its latest offerings for HPC. In partnership with Hardcore Computers, Boston are unveiling a liquid blade with total submersion technology, the Liquid Blade. The Liquid Blade offers industry-leading performance by harnessing the power of Intel's Xeon 5600 generation processors. By employing patented total liquid submersion cooling technology, the Liquid Blade allows the CPUs to be clocked to 4.7GHz. Heat generating components, including CPU, GPU and power supply are submerged in Core Coolant, which has a 1,350 times greater heat rejection capacity by volume than air, by volume; keeping components cooler and providing greater reliability and sustained optimal performance.
Boston will also be exhibiting its latest generation of SuperFlex Blade solutions, powered by HPC Links’ latest Vertex software. Vertex (Virtual Execution of Resource Tagged Executables) is an HPC system architecture for commodity platforms to deliver supercomputer-class performance. It enables the scalability of hundreds of thousands of standardised commodity components with a virtual execution environment that makes compute node execution transparent to applications running on the control nodes (Vertex nodes), providing a usable and scalable environment from personal supercomputers to exascale-class clusters of super-nodes.
Bull (www.bull.com) will be exhibiting its complete range of Extreme Computing solutions, based on bullx, a family of systems designed for uncompromised performance. The latest evolution of the bullx blade system will be in the limelight, since it is at the heart of the two production petascale systems installed by Bull in 2011. Bull will be unveiling a brand new addition to the bullx family, incorporating new levels of energy efficiency. Visitors will have the opportunity to meet extreme computing experts for advice on the best solution for high-performance computing needs.
CAPS (www.caps-entreprise.com) is a major supplier of solutions dedicated to application migration and deployment on manycore processors. CAPS global solutions for manycore lead the developer to performance by providing top-of-the-range technology (HMPP hybrid compiler and wizard), code porting methodology and ecosystem (third-party software tools, expertise, and training).
During SC11, CAPS will present HMPP 3.00: the latest release of its directive-based and multi-target HMPP compiler enables developers to safely move to hybrid CPU/GPU model and quickly get performance by leveraging the computing power of stream processors without the pain associated to GPU programming.
HMPP is offered within the new CAPS DevDeck package: an all-in-one multi-level suite for manycore application definition, porting and optimisation with tools (HMPP compiler, development tools such as HMPP Wizard, debugging and profiling software and scientific libraries), methodology and resources (such as tutorials and case studies).
Louisiana State University's Center for Computation & Technology (CCT) (www.cct.lsu.edu) will showcase research that demonstrates the diverse ways in which CCT advances parallel computing. The ParalleX group will provide a live demonstration of an adaptive mesh refinement (AMR) toolkit based on its HPX framework. The simulation will adapt interactively to changing power requirements as well as interactively killing or adding processors.
Poster displays at the booth will highlight advances by the Cactus Frameworks group including: the release of the new Einstein Toolkit (an open toolkit for relativistic astrophysics), computational fluid dynamics simulations using GPUs, and new simulation software for coastal science based on Funwave.
The Cornell University Center for Advanced Computing (CAC) (www.cac.cornell.edu) is a leader in high-performance computing systems, applications, and data solutions that enable research success. CAC will demonstrate Red Cloud, a new Infrastructure as a Service (IaaS), and Red Cloud with Matlab, a Software as a Service (SaaS), developed specifically for research computation and now available to scientists across the nation. CAC will also demo the Cornell Virtual Workshop, an online learning platform that CAC is using as an NSF Extreme Science and Engineering Discovery Environment (XSEDE) and Texas Advanced Computing Center partner to advance the computational science and engineering skills of researchers. In addition, Cornell breakthroughs will be described, including the atomic-level modelling of crystal ceramics such as perovskite that is providing insight into the possible discovery of the best materials to make low-power electronics at room temperature. Finally, CAC will show a new video of the Cornell campus.
DataDirect Networks (www.ddn.com) is a storage company and storage system supplier to 60 per cent of the world’s largest supercomputers. The scalability and performance of DDN storage systems, including the fastest storage systems in the world at more than 300GB/s, help customers optimise their supercomputing investment.
DDN will be exhibiting a complete HPC storage offering, including its new SFA10K-X, which features maximum performance efficiency while supporting the complete range of SSD, SATA and SAS media from a single, scalable system. SFA10K-X highlights include the ability to: accelerate big data and HPC workloads, with read-write speeds of up to 15GB/s; efficiently exploit solid-state drive technology to deliver more than 840,000 I/Os per second; support up to 1,200 drives (SSD, SATA, SAS) in a single system, while requiring only two floor tiles; and be capable of housing up to 1.8 petabytes (PB) in a single data centre rack.
The German Climate Computing Centre (DKRZ) (www.dkrz.de) is a national service centre for climate research, which depends on access to HPC systems. Climate modelling produces enormous amounts of data – at DKRZ annually about 10 Petabytes. Climate research has evolved into a data intensive science. DKRZ provides tools and services to deal with the typically large data sets. The interdisciplinarity of climate research requires data integration across different disciplines and data types. Therefore, DKRZ supports the entire data life cycle including data documentation, data quality assurance, long term archiving and network data access within international data federations.
DKRZ will present visualisations of intricate model experiments. Special focus is set on results of small spatial scales processes such as storms, cyclones or hurricanes in the atmosphere or eddies in the oceans.
The Scientific Computing research group at DKRZ will discuss high performance I/O optimisations, energy efficiency, and simulation of cluster infrastructure.
EPCC (www.epcc.ed.ac.uk) will be highlighting its work at the forefront of exascale and data intensive research; training (including its bespoke courses for industry, MSc in HPC and PhD positions) and its extensive collaborations with industry. Visitors can also find out about EPCC's HPC service provision, which includes machines such as HECToR, the UK national supercomputing service. EPCC is also represented elsewhere, for example on the booth of PRACE, the European supercomputing project. Finally, it is involved in the education side of SC’11 too, with staff assisting on a number of the OpenMP tutorials and providing a Co-array Fortran tutorial in collaboration with Cray.
Gnodal (www.gnodal.com) has developed an ultra-low latency 10/40 GbE switch family, free from current network limitations, offering unprecedented performance improvements and almost unlimited scalability. Gnodal switch products derive their performance from their architectural design, which minimises network congestion and delivers optimum performance, with minimal cost and power consumption.
When multiple switch products are combined, the latency increase across multiple switch units is minimised and full congestion management is maintained and the overall administration of the switch infrastructure is simplified. This allows Gnodal Ethernet to extend the capabilities of the Ethernet standard into highly-scalable configurations with minimal complexity, offering an overall more efficient use of capital equipment in the data centre. Gnodal products fulfill today’s high performance requirements in such markets as science and engineering, financial services, Web 2.0, cloud and hosting providers, video, life sciences, and oil and gas exploration.
Intel (www.intelserveredge.com) will be demonstrating how its server products can deliver quality, ingenuity and commitment. Its server boards, systems, chassis and RAID solutions are designed, manufactured and tested to meet the highest quality standards. Intel works with its Intel Technology Provider partners to provide high levels of customer service.
Visitors to the booth will discover why John Lee, vice president, ATS, Appro International, called Intel's Server Board S2600JF 'the best HPC server board in the industry'. Purpose-built for HPC and supporting the Future Intel Xeon processor E5 family, this platform is designed to put the 'high performance' back in computing.
Mellanox (www.mellanox.com) will be displaying the world’s fastest server and storage I/O: FDR 56Gb/s InfiniBand.
During the show, it will be hosting presentations every half-hour from leading server and storage OEMs, ISVs, end-users and academia, who will provide insight to the benefits and application performance improvements end-users achieve when using low-latency 56Gb/s InfiniBand interconnect technology.
More information, including Mellanox presentation time slots during the conference, live demonstrations, give-a-ways, and a special Wednesday night event, is available on the Mellanox website.
NextIO (www.nextio.com) will be featuring its recently updated vCORE Extreme GPU consolidation appliance. Now with the ability to deliver the Nvidia M2090 GPUs, the vCORE Extreme is a data centre-class GPU consolidation solution specifically designed to maximise density for the Tesla Fermi M-Class GPUs. A 3U box that supports up to 16 GPUs, vCORE Extreme 2090 enables customers to run GPUs outside of their servers providing denser server configurations, easier lifecycle management and a lower overall cost of ownership. The vCORE Extreme 2090 is the ideal solution for real-world applications such as seismic processing, biochemistry simulations, weather and climate modelling, signal processing, computationalfinance, CAE, CFD, and data analysis.
NICT (National Institute of Information and Communications Technology), Japan (www.nict.go.jp) conducts general research and development on information technology supporting the ubiquitous society, and supports businesses related to information communications.
Some of the exhibits and demonstrations on the booth include: a space weather forecast, a research project that uses computer simulation and informatics technology to monitor and model space environment; a joint research project with NICT, AIST and KDDI R&D laboratories, that will show network provisioning for cloud infrastructure across Japan, US and EU; a research project that will show the status of a multi-domain R&E networks on the Google Earth by using perfSONAR, PERFormance Service Oriented Network monitoring ARchitecture; and a joint research with NTT laboratories that will show measurement-based global-scale automated path setup technology using standardised protocol NSI in combination with video streaming processing nodes.
Numascale (www.numascale.com) is a leader in interconnect technology development, disrupting the computer server business by introducing a low cost scalable shared memory adapter for commodity servers. Using AMD's standardised HyperTransport processor-memory interface, Numascale has developed and commercialised an interconnect with the ability to build scalable high performance clusters with cache coherent shared memory. The NumaConnect SMP adapters enable the whole cluster to be utilised as one single system running one instance of the operating system. Standard applications can take advantage of all available processors, memory and I/O throughout the cluster with no software overhead or system memory loss. Building on decades of experience with high performance computer and interconnect architectures, the Numascale team will be present to showcase the technolog.
Numerical Algorithms Group (NAG) (www.nag.co.uk) delivers trusted, high quality numerical computing software and high performance computing (HPC) services. For four decades NAG experts have worked closely with world-leading researchers in academia and industry to create powerful, accurate and flexible software, which is relied upon by tens of thousands of users, companies, and learning institutions as well as numerous independent software vendors.
From multicore workstations to the latest supercomputers, NAG's robust software development products, parallel software engineering services, independent technology consulting and HPC training can support users' modelling and computing needs
The Ohio Supercomputer Center (www.osc.edu) empowers thousands of researchers through its HPC services. It forges partnerships to make manufacturers more competitive through the Blue Collar Computing programme, and leads the national effort to boost workforce competency through the Ralph Regula School of Computational Science. Its world-class booth partners this year include leaders in bioinformatics and digital manufacturing: The Alliance for High Performance Digital Manufacturing (AHPDM), The Research Institute at Nationwide Children's Hospital, Nimbis Services, The Ohio State University Department of Biomedical Informatics, and TotalSim.
Panasas (www.panasas.com) is a provider of high performance parallel storage solutions for technical applications and big data workloads. All Panasas ActiveStor scale-out storage appliances use the Panasas PanFS parallel file system to deliver superior performance, data protection, scalability, and manageability. ActiveStor linearly scales to 6PB and 150GB/s and is optimised for demanding HPC storage applications in the bioscience, energy, government, finance, manufacturing, and other core research and development sectors. The booth will feature an ActiveStor product demonstration and the latest information about: parallel NFS (pNFS), high performance parallel storage for private cloud implementations, and the future of exascale computing. During the week, Panasas will also be delivering a technical tutorial on 'Parallel I/O in Practice', and co-founder and CTO Garth Gibson will take part in an SC11 paneldiscussion on data intensive computing.
Platform Computing (www.platform.com), a leader in cluster, grid and cloud management software, has more than 2,000 of the world’s most demanding organisations as customers, including iconic names like Formula 1's Red Bull Racing, AMD, Citigroup, Harvard Medical School, and CERN. In addition to a number of announcements at the conference, Platform Computing will be delivering several presentations. During the Exhibitor Forum HPC Cloud Product Manager, Chris Porter will be asking the question 'To Burst or not to Burst?' and weighing the benefits of HPC applications in the cloud. Director of product management, William Lu, will be joined by Virginia Tech Transportation Institute's chief information officer, Clark Gaylord, to discuss 'Building a Scientific Data Warehouse, including Multi-boot HPC Capabilities'.
The Partnership for Advanced Computing in Europe (PRACE) (www.prace-ri.eu) forms the top level of the European HPC ecosystem. PRACE project members will be available on the booth. PRACE is also organising a BoF session entitled 'PRACE – The European HPC Infrastructure', on Thursday 17 November at 12.15pm, at which it will present some of its outstanding research results achieved with PRACE resources and give information on its Tier-0 and Tier-1 services. This will be an excellent opportunity to learn how to apply for resources and application support. PRACE representatives will be on hand to chat about everything PRACE and answer any questions visitors may have. Finally, PRACE is also organising a treasure hunt at the booth, with prizes on offer.
ScaleMP (www.scalemp.com) is a leader in virtualisation for high-end computing, providing higher performance and lower total cost of ownership (TCO). The Versatile SMP (vSMP) architecture aggregates multiple x86 systems into a single virtual x86 system, delivering an industry-standard, high-end symmetric multiprocessor (SMP) computer. Using software to replace custom hardware and components, ScaleMP offers a new computing paradigm. vSMP Foundation is a software-only solution that eliminates the need for extensive R&D or proprietary hardware components in developing high-end x86 systems, and reduces overall end-user system cost and operational expenditures. vSMP Foundation aggregates up to 128 x86 systems to create a single system with 4 to over 1,000 processors (that's over 8,000 cores) and up to 64 TB of shared memory.
Solarflare (solarflare.com) develops application-intelligent 10GbE networking hardware and software solutions, and will be showcasing its solutions for the HPC market. Solarflare's OpenOnload application acceleration middleware increases message rates for HPC applications two to threefold, while reducing application latency by 50 per cent. Its new SFN5322F dual-port 10GbE SFP+ server adapter extends the reach of precision time protocol and enables monitoring, auditing and analysing of HPC networks. In addition, Solarflare's low-latency, high-performance 10GbE server adapters are now available for IBM and HP blade chassis. The products expand Solarflare's ability to reach customers in scale-out computing environments by reducing the complexity and increasing the performance of high-density blade server environments. Solarflare’s networking solutions provide a flexible, direct model for accelerating applications, while maintaining support for network and operating system functions. Solarflare supports more vNICs and PCIe functions than competing solutions to relieve network I/O bottlenecks and deliver scalability in virtualised environments.
Source Support Services (www.sourcesupport.com) is a global provider of customised logistics and onsite service solutions for the HPC industry. For more than a decade, Source Support's service programmes have helped improve the outsourcing experience for IT manufacturers.
Through its training programmes, global logistics capabilities and network of locally-based field technicians, it provides same day and next business day onsite support in North America and more than 60 countries worldwide. Each of its service programmes is customised to user products, processes and requirements, no matter how specialised.
SysFera (www.sysfera.com) develops SysFera-DS, an open-source software solution for mutualising and managing heterogeneous and distributed HPC resources (workstations, supercomputers, grids, public or private clouds).
SysFera will present the many advanced features of SysFera-DS and how it provides a business-grade, user-centric HPC solution. It will be showcasing the new version of the SysFera-DS web interface. It allows users to simply run intensive applications or manage dataflows (that can be designed using an integrated GUI tool or an XML-like language). It also features data-management functionalities over the infrastructure (persistence, implicit and explicit replication). Users can access the power of the HPC infrastructure without worrying about its underlying complexity. SysFera offers training, support, maintenance and expertise for SysFera-DS.
Tech-X (www.txcorp.com) will be demonstrating GPULib, its library of mathematical kernels for GPU computing; Orbiter, our framework for web services; and Vorpal, our simulation application for modelling the interaction of matter with electromagnetic fields.
GPULib allows users to access high-performance computing with minimal modification to their existing IDL programs. GPULib can accelerate new applications or be incorporated into existing applications with minimal effort.
Orbiter is an end-to-end framework delivering fast and secure solutions through both thin-client web access and thick-client desktop application suites and modules. These applications leverage the information-sharing capabilities of Orbiter in providing powerful and personalised web-accessible components.
Vorpal enables researchers to simulate complex physical phenomena in less time and at a much lower cost than empirically testing design changes for electromagnetic and plasma problems. Vorpal offers a combination of physical models to cover the entire range of EM and plasma simulation problems.
Texas Instruments (www.ti.com) will be showcasing its TMS320C6678 Multicore DSP. TI’s multicore processors are well suited for a wide variety of HPC applications, delivering ultra high performance and significant power savings. TI will be demonstrating a quad-DSP, PCIe solution with >500 GFLOPs in 50W, along with key benchmarks from applications such as medical imaging, oil and gas exploration, and finance. TI will also demonstrate the flexibility of its multicore DSPs where the C6678 card can either be used as an accelerator card or a platform for running a complete solution on its own. The performance, flexibility and scalability of TI’s multicore DSPs allows customers to embed high performance computing into a wider variety of solutions than ever before.
VLSCI (Victorian Life Sciences Initiative) (www.vlsci.org.au) is an initiative of the Victorian State Government in partnership with The University of Melbourne and includes the world's first IBM Research Collaboratory for Life Sciences. Located in Melbourne and dedicated solely to life sciences research, VLSCI provides both expertise and facilities to Victorian researchers and their collaborators. Currently operating at a modest 46 teraflops, VLSCI will deliver petascale computing in 2013. VLSCI contributes to Victoria’s reputation as a global centre for excellence in life sciences research capabilities.
Dynamic collaborations between life scientists and computer scientists aim to advance human health by improving diagnostics, finding new drug targets, refining treatments for major diseases affecting our community: cancer, epilepsy, genetic disorders, infectious diseases and eye disease, among others.
The joint VPAC and VLSCI booth represents the leading edge of Australian HPC and research services. A wide variety of projects will be on display, accompanied by staff.