SC12 brings together the world of HPC in Salt Lake City, Utah, from 12-15 November
Altair (www.pbsworks.com) will be running live demonstrations of its HPC portals at SC’12. A leader in digital simulation, the company creates HPC simulation applications, delivers HPC management tools, and uses the applications/tools to build real products. The hub for its HPC solutions, PBS Professional, optimises the world’s largest supercomputers. Altair states that it is the workload management software of choice by many supercomputing sites within the Top500, providing maximum security, reliability and energy-efficiency.
Compute Manager enables users to run, monitor and manage workloads on distributed resources remotely, while the Results Visualization Service offers advanced job monitoring capability through inspecting and visualising data from the solver log files. This allows the visualisation of animation and plot data for running and completed jobs. Altair’s Access Management Service enables users to share application profiles with specific users and/or groups, and the company’s HPC solver, Radioss, is powered by PBS Professional’s ability to minimise job turn-around time.
Boston (www.boston.co.uk), a leading manufacturer of high-performance, low-powered server, storage and clustered solutions, will present its Viridis server at this year’s Supercomputing exhibition. The Viridis platform is the first server of its kind to enter general availability. Based on technology from ARM, the Boston Viridis is driven by the low-powered Calxeda EnergyCore SoCs (System on Chip) that consume as little as five Watts each, providing data centre performance while consuming power comparable to a mobile device.
Clustered within a 2U enclosure, each system holds up to 12 EnergyCards that contain four Calxeda EnergyCore SoCs each. With up to 48 nodes per enclosure, this highly-dense solution contains the equivalent of up to 900 servers per industry standard 42U rack. Supported by Ubuntu Server 12.04 LTS and Fedora v17+ distributions, the Calxeda EnergyCore SoC has been shown to run popular software platforms including cloud management software from Openstack and big data technologies such as Hadoop, Cassandra, applications built in Java, Ruby on Rails, Python and the popular LAMP stack.
At SC12, Bull (www.bull.com) will be showing its broad range of extreme computing solutions, including its bullx family of systems designed specifically for HPC, and its HPC on-demand solution. The latest evolution of the bullx blade system, the B515 accelerator blades, boosts high-performance computing with either two Nvidia Tesla GPUs or two Intel Xeon Phi co-processors. They are suited for configurations of all sizes and can be combined with standard CPU-based blades within the same chassis.
Bull will also be exhibiting its B700 Direct Liquid Cooling blades, which the company says deliver drastic savings in energy consumption. Their revolutionary technology brings cooling to the heart of the blades and makes it possible to use warm water for cooling, while the systems remain easy to maintain.
Convey Computer (www.conveycomputer.com) will be introducing a new platform that accelerates high-performance analytics applications. Based on a new Scalable Threaded Architecture (STA), the platform directly addresses performance limitations found in today’s commodity systems. STA leverages Convey’s hybrid-core, reconfigurable compute elements, a highly parallel memory subsystem and an industry-standard development environment to provide dramatically increased performance of big data analytics applications.
STA supports extreme multi-threading, allowing parallelisation that efficiently spans thousands or tens of thousands of threads. The high degree of parallelisation is supported by a new, automatic hardware-based thread scheduling mechanism that reduces overhead and further increases performance.
Convey Computer breaks power, performance and programmability barriers with the world’s first hybrid-core computer – a system that marries the low cost and simple programming model of a commodity system with the performance of customised hardware architecture.
The Cornell University Center for Advanced Computing (CAC) (www.cac.cornell.edu) is a leader in high-performance computing systems, applications, data and cloud solutions that enable research success. The centre will highlight Red Cloud services, a new National Science Foundation award focused on managing cloud usage allocation and accounting. In addition to that, CAC will introduce a cloud use survey that attendees are encouraged to participate in to help the NSF plan for the integration of cloud resources into national cyberinfrastructure. CAC will also highlight its roles in the Extreme Science and Engineering Discovery Environment (XSEDE) project, including user training, campus bridging and architecture and design coordination.
The German Climate Computing Centre (www.dkrz.de) is a unique national service centre for climate research which depends on access to HPC systems. Climate modelling produces enormous amounts of data and at DKRZ more than 20 Petabytes of climate model data is stored. DKRZ provides tools and services to deal with the typically large data sets. The interdisciplinary nature of climate research requires data integration across different disciplines and data types, and so DKRZ supports the entire data lifecycle, including data documentation, data quality assurance, long-term archiving and network data access within international data federations.
At SC’12, DKRZ presents the visualisations of intricate model experiments, which are a German contribution to the next IPCC assessment report. The extensive climate simulations were carried out by DKRZ and Max-Planck-Institute for Meteorology with the Earth system model MPI-ESM and occupied one quarter of total computing capacity over a period of two years.
EMI (www.eu-emi.eu) is a joint effort between the major European distributed computing middleware providers gLite ARC, UNICORE and dCache. EMI delivers distributed, secure compute and data management services to support and evolve research infrastructures and enable academic and industrial researchers to access resources, data and applications across the world. It targets the needs of users with huge computational and storage demands. EMI improves the existing middleware services and harmonises them, realising a common framework with the result of rendering the middleware to be simpler and easier to use, while reducing the interoperability problems faced by the distributed computing infrastructure providers.
EPCC (www.epcc.ed.ac.uk) will be highlight ing its work on Exascale research, training in the form of its MSc in HPC and PhD positions, and its extensive collaborations with industry. The organisation will also be displaying details of its HPC service provision, where it hosts machines like HECToR (the UK national supercomputing service) and the DiRAC consortium's BlueGene/Q. However, the booth is only a small part of its involvement in the conference, with EPCC involved in other booths such as PRACE (the European supercomputing project), and joining up with its CRESTA partners to showcase European Exascale research. EPCC is also participating in a number of the SC’12 workshops and birds-of-a-feather sessions.
Eurotech (www.eurotech.com), a HPC manufacturer and solutions provider, and leading embedded electronics company, will introduce Tigon, the new hybrid supercomputer of the Aurora family. In the configuration presented at SC’12, Tigon will use Nvidia GPUs and Intel Xeon E5 series processors to get to a peak performance of 300 Tflops per rack with 120Kw of peak power. The Tigon rack and nodes with the hot water cooling plate and a demo system showing the Aurora HPC 10-10 supercomputer architecture and liquid cooling will all be on display on the company’s booth. In addition, there will be the presentation of the DEEP architecture, based on the concept of ‘booster’, a cluster of accelerators, as well as an introduction to the Eurotech cloud computing solutions.
Extoll (www.extoll.de) is a developer and vendor of high-performance computing network technology that realises direct networks, with each node having up to six neighbours. At SC’12, the company will showcase its in-house developments, such as a nine-node system equipped with Extoll network cards, a direct GPU-GPU communication between another two nodes and a 12 times high-density Active Optical Cable (AOC). The latter is appearing for the first time. In addition to observing live demonstrations, attendees will be able to obtain further information at the Exhibitor Forum Talk on 14 November at 15:30 local time. There will be the opportunity to register as early adopter or to participate in a lottery drawing with the chance of winning computing time on an Extoll cluster.
Georgia Tech (hpc.gatech.edu) is one of the US’ premier public research universities, consistently ranked in the top ten by the US News & World Report. Georgia Tech’s Institute for Data and High-Performance Computing is a leader in real-world computing innovation, and its School of Computational Science & Engineering creates new expertise, technologies and practitioners in areas like HPC and cloud computing, modelling and simulation, visualisation and data analysis and mining. At SC’12, Georgia Tech HPC faculty and students will be participating in numerous presentations, workshops, posters, panels and birds-of-a-feather sessions.
IGI-Italian Grid Initiative (www.italiangrid.org) is a joint research unit that develops, manages and supports the National Grid Infrastructures for the National Scientific Research. IGI provides a set of interoperable grid and cloud services with several customisations that meet the specific needs of many disciplines and a large variety of communities. IGI enables the access and sharing of data and compute facilities thanks to its flexibility and a lower entry barrier.
Part of EGI, the biggest European Grid Infrastructure, IGI is heavily involved in all relevant EU and worldwide grid-related projects, with substantial contributions. Moreover, it is moving towards environments other than research and is committed to the realisation of the Smart Communities and Cities initiative by offering a wide set of services, tools and support for the solution of everyday life issues. Open source IaaS and PaaS cloud platforms where developers essentially match with users is the main strategy of IGI Cloud Computing Technologies for a real Smart Government.
The mission of the InfiniBand Trade Association (IBTA) and the OpenFabrics Alliance (OFA) (www.infinibandta.org) is to drive RDMA and high-performance I/O technologies forward in the industry. RDMA continues to be a thriving, crucial technology with increasing relevance and importance, and at SC’12, the IBTA and OFA will highlight recent adoptions in areas such as cloud and virtualisation, in addition to HPC. Members of both organisations will participate in a panel titled ‘Exascale and Big Data I/O’ which will take a closer look at the drivers for future I/O architectures, and discuss whether they should be open or proprietary. The IBTA and OFA will make experts available throughout the show, focusing on topics such as HW testing/interoperability capabilities, alternatives to sockets, RDMA over WAN, storage protocols, OpenSM and much more. To meet with the IBTA or OFA at SC’12, please contact firstname.lastname@example.org.
The broad range of solutions for supercomputing will be on display from the Intel Server Products Division (www.IntelServerEdge.com) at SC’12. In addition to presenting its family of high density half-width and full-width boards and systems for the Intel Xeon E5 Processor – products that featured in the top 30 of the latest Top500 (www.top500.org) list – Intel will unveil its new four socket boards and systems (S4600LH) for the Intel Xeon E5-4600 Processor. The company’s server boards, systems, chassis and RAID solutions are designed, manufactured and tested to meet the highest quality standards, and along with its Technology Provider partners, Intel is committed to delivering an unmatched customer experience.
Mellanox Technologies (www.mellanox.com) will be showcasing, along with its partners and customers, Connect-IB, the industry’s only adapter card with >100Gb/s throughput on a single device, as well as the full line of end-to-end FDR 56Gb/s InfiniBand and 10/40GbE interconnect solutions. Along with several live demonstrations, attendees can come every half hour to the booth to hear presentations where they can learn from leading server and storage OEMs, ISVs, end-users and academia, who will provide insight to the benefits and performance and scalability improvements when using low-latency FDR 56Gb/s InfiniBand and 10/40GbE interconnect solutions.
Nallatech (www.nallatech.com) will be showcasing its PCIe-385N network processing card. The PCIe-385N low profile network processing card provides a powerful PCI Express Gen 3 computing platform for FPGA development and deployment across a range of application areas, including network monitoring, filtering, analytics, data retention and algorithm acceleration.
The PCIe-385N features two SFP+ connectors supporting 1GbE, 10GbE, SONET/SDH and OTN via the PCI backplate. These network interfaces are directly coupled to an Altera Stratix V FPGA with high bandwidth DDR3 SDRAM memory. The PCIe-385N is the same form factor as most standard network cards, providing compatibility with almost all high density server and blade centre platforms.
NASA's exhibit (www.hec.nasa.gov) highlights science and engineering results enabled using the agency's world-class supercomputing resources. Cutting-edge work in aeronautics, Earth science, human space exploration and study of the universe will all be showcased at SC’12. This year’s slate of talks and demonstrations reveal some of the innovative technologies behind projects such as the Mars Science Laboratory’s thermal protection system; flight formation for reducing aircraft carbon emissions and fuel consumption; research to better understand the formation of stars, planets and galaxies; the upcoming Space Launch System vehicle and launch environment; and advanced methods for code optimisation, data analysis and visualisation. Visitors to the booth can experience spectacular scientific visualisations displayed on a new, state-of-the art hyperwall and talk to the agency’s experts.
The Numerical Algorithms Group (NAG) (www.nag.com) delivers trusted numerical computing software and high-performance computing (HPC) services. NAG experts work closely with 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 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 modelling and computing needs. At SC12, NAG will show how routines from the NAG Library, running on various platforms including the latest co-processors, can improve applications in many areas of scientific computing. NAG’s consultants will also be available to help attendees solve their specific numerical computing problems.
Panasas (www.panasas.com) is a premier provider of high-performance parallel storage solutions for technical computing applications and big data workloads. All Panasas scale-out storage appliances leverage the Panasas PanFS parallel file system to deliver superior performance, data protection, scalability and manageability. The company introduced initial parallel storage products based on its first generation blade architecture in 2004 and will be launching the fifth generation of ActiveStor technology at SC’12.
Panasas systems are optimised for highly demanding applications in the four key big data segments: Design – facilitating engineering collaboration; Deposit – providing secure data warehousing; Discover – enabling complex simulations; and Decide – powering analytics across a range of disciplines. Visitors to the booth can find out more about ActiveStor 14, the scale-out NAS appliance for big data workloads that is easy to deploy, use and manage.
The Parallel Computing Institute (parallel.illinois.edu) draws on the collective strength of the University of Illinois' College of Engineering to build powerful and efficient infrastructure for computation-heavy applications. Visitors to the booth will see how PCI research is enabling the ability to read MRI results in real-time, weather and disease prediction tools, individualised genomic analysis tools for targeted treatments, design tools for new nano-scale materials, ultra-realistic video gaming and video conferencing, and other applications. One highlight at SC’12 will be the teleimmersion demos, where attendees can interact with a user in a different location in a single virtual space.
PRACE (www.prace-project.eu), the Partnership for Advanced Computing in Europe, is an international non-profit association based in Brussels. It has 24 member countries providing persistent world-class high-performance computing services for researchers in academia and industry. The PRACE infrastructure is well integrated into the European HPC ecosystem and PRACE is supported by The European Commission under grant agreement n° RI-261557, RI-283493 and RI 321763. www.prace-ri.eu
At SC´12, PRACE will be presenting its Tier-0 infrastructure of petascale systems and its services in training and supporting users in porting, scaling and optimising key applications to make effective use of the three billion core hours granted to scientists through a rigorous peer review process.
Rogue Wave Software (www.roguewave.com) is the largest independent provider of cross-platform software development tools and embedded components for the next generation of HPC applications. Rogue Wave products reduce the complexity of prototyping, developing, debugging and optimising multi-processor and data-intensive applications.
At SC’12, the company will be featuring TotalView, a highly-scalable debugger that provides troubleshooting for a variety of applications. A GUI-based source code defect analysis tool, TotalView gives developers what Rogue Wave states is unprecedented control over processes and thread execution and visibility into program state and variables. It allows developers to debug one or many processes and/or threads with complete control over program execution.
ThreadSpotter is an advanced and easy-to-use performance optimisation product for single and multi-threaded applications that helps eliminate performance issues by analysing memory bandwidth and latency, data locality and thread communications/interaction, to pinpoint performance issues. It then guides the developer to the location of the issues in the source code.
RSC Group (www.rscgroup.ru), a leading Russian full cycle HPC solution provider, will demonstrate the RSC Tornado at SC’12. RSC Tornado is a cutting-edge, energy-efficient and scalable solution with liquid cooling from 50 Tflops miniDC to large multi-Pflop supercomputers, and a computing efficiency ratio of more than 92 per cent on Linpack. It is a flexible and green design based on widely available standard server motherboards with Intel Xeon E5-2690 processors, Intel SSD, high-speed Infiniband FDR interconnects and co-processor support. In addition, RSC BasIS, the company’s HPC software stack, manages clusters in an energy-efficient way, complements proven liquid cooling to deliver a low total cost of ownership (TCO) and more than 60 per cent power savings with an industry-record PUE of less than 1.06.
Silicon Mechanics (www.siliconmechanics.com), a leading manufacturer of rackmount servers, storage and high-performance computing hardware, will highlight its new server products and strategic partnerships at SC’12. Silicon Mechanics was the August recipient of the Inc. 500/5000 fastest growing companies, and is proud to sponsor the Massachusetts Green High-Performance Computing Consortium in this year’s Student Cluster Competition. The company is also looking forward to the announcement of the 2nd annual Research Cluster Grant Competition at SC’12.
Silicon Mechanics and its partners, including Intel, AMD, Nvidia and Supermicro, offer customers a line of products and services that are comprehensive, competitively priced and customisable. These include high-performance workstations, department and workgroup clusters, and energy-efficient data centre solutions.
Spectra Logic (www.spectralogic.com/bluewaters) will present the Spectra Logic T-Finity, a modern design that offers what the company says is industry-leading scalability with the speed necessary to meet the requirements of the most data-intensive environments in the world. T-Finity redefines what customers should expect from modern tape archive and backup solutions, with comprehensive features and significant CapEx and OpEx savings. Its unique design can scale to 3.6 EB of compressed capacity with TS1140 technology tape drives. This extreme capacity can be stored without giving up the density and ease of management that are hallmarks of Spectra enterprise libraries. Visitors to the booth can discover how the T-Finity library is getting the NCSA Blue Waters project to 380 Pbytes.
SysFera (www.sysfera.com) is an editor of software with a strong experience in the fields of HPC, SaaS, cloud and big data. For software editors, the company builds scalable, reliable SaaS solutions to move their software offering to the cloud or to deploy it inside an intranet. For industrials, computing centres, research labs and universities, SysFera develops SysFera-DS, a collaborative computing portal that makes intensive applications, often usable only by computer experts, accessible to every user from a regular web browser, no matter their skill level. In addition to its modules for managing applications, users and information, SysFera-DS is now receiving a new module to handle remote visualisation and additional, more precise, metrics for resource usage.
This year marks the 8th time that The MathWorks (www.mathworks.co.uk) has exhibited at the US Supercomputing show. In addition to running demonstrations of Matlab GPU computing capabilities, the company will show several Matlab toolboxes, covering a variety of application areas that now incorporate GPU-enabled algorithms. Experts will be available to answer questions and show users how to use the GPU-enabled toolboxes to accelerate their Matlab code with only minor modifications.
The MathWorks will also explain how users can harness cloud computing resources available from Amazon EC2 for running computationally intensive Matlab programs and Simulink simulations. Demonstrations will illustrate how to launch computations from Matlab sessions and scale up, on-demand, to a large number computers available through EC2.
The University of Illinois and its National Center for Supercomputing Applications (www.ncsa.illinois.edu/extreme-scale) are addressing the challenges and opportunities of parallel computing at all scales, from portable personal devices of ever-growing power to petascale supercomputers like Blue Waters. Attendees will be able to learn more about the 11+ petaflop Blue Waters system, NCSA’s big data capabilities (including a 380 PB archive), and how these resources will be harnessed to improve understanding of severe weather, astrophysics, the molecular mechanisms of disease, and much more. Visitors to the booth will also get the chance to play Flops Fever, the mobile game designed and programmed by University of Illinois students.