HPC: ISC’10 SHOW PREVIEW
EPCC will exhibit at the event. Founded at The University of Edinburgh in 1990, EPCC is a leading European centre of expertise in advanced research, technology transfer and the provision of supercomputer services to academia and business. EPCC, as lead partner in the HPCx consortium, is proud to host and support the UK National Supercomputer, HECToR. The centre’s expertise and computing facilities underpin its long-standing research visitor programmes. EPCC is also a major provider of HPC training in Europe and it offers courses for academic and commercial users throughout the UK and Europe, including its popular MSc in HPC. In 2009 the EPCC Industry Hub was launched, which provides streamlined access to the centre’s supercomputing facilities and expertise.
www.epcc.ed.ac.uk
FluiDyna, based near Munich, Germany, will present two new in-house developments: LBUltra and Culises. LBUltra is GPU- based software that enables computational flow simulation tailored to the individual needs and demands of FluiDyna’s customers. Culises is a Cuda library for solving linear systems of equations, and offers various iterative solvers and interfaces to engineering applications as well as supporting unstructured meshes. As these algorithms are processed on GPU hardware, existing software solutions can be accelerated considerably and hence overall computing times are reduced.
www.fluidyna.de
IBM will present an overview of its HPC portfolio and associated network technology and storage solutions. IBM’s key topic at the event is ‘High performance computing for a smarter planet’. The company will present its latest solutions that help clients to stimulate innovations in an intelligent and lasting way. At its exhibitor booth, IBM will demonstrate high-end Unix-based Power7 servers, the latest x86 server technology with IBM eX5, the highly integrated scale out compute solution iDataPlex, and the IBM scale out storage system Sonas. IBM also plans to show a prototype of the IBM Power7 PERCS system (Productive, Easy-to-use, Reliable Computing System).
www.scientific-computing.com
Attendees at the show can find information on the new Exascale Innovation Center – a joint initiative of Jülich Supercomputing Centre (JSC) and IBM Research and Development Boeblingen – which aims to develop a supercomputer in the performance range of an exaflop.
On its booth, the JSC will demonstrate
a model of its supercomputer QPACE, which was developed by several European universities and research facilities and the German IBM research team.
www.ibm.com
LSI will
demonstrate its LSI Engenio 2600 storage system, based on next- generation 6Gb/s SAS technology, providing solid performance, high efficiency and cost savings.
Also on show will be the LSI Engenio DE6900 high-density SATA drive enclosure with InfiniBand or 8Gb/s Fibre Channel host connectivity, which enhances the LSI HPC system’s ability to handle the data requirements of HPC applications.
LSI has a rich and successful history of deploying solutions and supporting customers in HPC markets, including government, research, and energy, as well as media and entertainment. With LSI HPC and HEC high-bandwidth storage, end-customers can deploy InfiniBand and Fibre Channel data centre solutions with confidence.
www.lsi.com/hpc
NextIO will demonstrate the capability to pool up to eight double-wide GPUs in a single 4U top-of-rack vCore appliance, and dynamically reconfigure the server to GPU connections on the fly. The vCore GPU consolidation appliance allows any server connected via PCI Express to access any number of GPU’s depending on job requirements. Each GPU can slide into the chassis via a hot plug sled, providing ease-of-upgrade and serviceability without impacting running applications. The
Numascale will showcase the new NumaConnect technology in a live environment. This technology enables the HPC community to configure highly- scalable shared memory machines (SMP) at low cost, using commodity servers and NumaConnect. Numascale’s NumaConnect provides
shared memory and I/O capabilities to clusters. NumaChip is a combined cache coherence controller and distributed switching fabric interfacing to AMD’s coherent HyperTransport. NumaChip uses a directory-based cache coherence protocol that is far more scalable than broadcast or snooping-based protocols. The on-chip switch connects to neighbour nodes in 2D or 3D Torus topologies and scales to 4,096 nodes. A single system can address 256 terabytes of DRAM and the caches containing data from remote nodes are configurable up to 16 gigabytes per node.
www.numascale.com
PlanetHPC, a Framework 7 Support Action, will be exhibiting at the event. It aims to combine the experience of past successes in HPC, current research initiatives and the requirements and vision from industry to create a roadmap for HPC research in Europe. PlanetHPC is the opportunity for anyone involved in HPC to influence the direction of research. This is a time of change in HPC, with the benefits that the global HPC community has enjoyed for the past 20 years of increases in performance provided by faster microprocessor clock-speeds and core complexity seemingly at an end.
SCIENTIFIC COMPUTING WORLD JUNE/JULY 2010
41
to GPU connections, allowing multiple GPUs to be brought in and out of service without rebooting machines. vCore provides a dynamic GPU pool of GPU resources for all connected servers.
www.nextio.com
demonstration will show how applications like Moab can easily reconfigure the server
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60