This page contains a Flash digital edition of a book.
HPC products HPC Products

MaxCloud Maxeler Technologies has launched MaxCloud, a cloud implementation of a high-performance datafl ow computing system.

MaxCloud offers businesses a scalable, on-demand, high-availability off-site resource to eliminate upfront hardware costs and deployment time while benefi ting from the high performance of datafl ow computing. MaxCloud provides a pool of accelerated compute nodes running an industry standard Linux

distribution that combine multi-core CPUs with multiple Maxeler datafl ow engines. Each compute node typically provides the performance of 20-50 standard x86 servers.

MaxCloud has multiple compute nodes each with 12 Intel Xeon CPU cores, up to 192GB of RAM for the CPUs and 4 MAX3 datafl ow compute engines. MAX3 uses Xilinx Virtex-6 FPGAs directly connected to up to 48GB of DDR3 DRAM, giving a total of 384GB in a single 1U server.

Datafl ow engines within the same compute node are directly connected via PCI Express and also via Maxeler’s high-bandwidth MaxRing interconnect.

Maxeler provides a complete

service to migrate applications to the MaxCloud and a comprehensive suite of software tools to develop, accelerate and maintain applications for Maxeler systems, whether deployed in the cloud or on-site. Maxeler systems are currently

NextComputing’s NextStream dense streaming blade server now supports Nvidia Tesla C-class GPU computing processors, including the recently released Tesla C2075. At two rack units (2U) high, the NextStream is optimised for high-performance computing tasks that require a mix of traditional multicore CPU processing with GPU co-processing in a dense form-factor. Traditional blade servers, while optimised for density in large-scale installations, require a signifi cant investment in rack real estate. Typical blade servers start at 5U high and those that are optimised for heavy GPU computing are often 7U high or greater. While this makes practical sense for large cluster deployments

NextStream supports Nvidia Tesla C-class GPUs in data centres, the combination of

CPU and GPU density is now a requirement across multiple industries. This includes projects in smaller environments like remote research labs or in vehicles such as aircraft or ships.

The NextStream high-density blade server solves this problem by loading up a 2U chassis with maximum CPU processing, support for the latest Nvidia Tesla GPU computing processors, integrated Ethernet switching, and a dedicated management PC module for in- chassis direct management of blade nodes. The NextStream 2U chassis can house up to three independent

Supercomputing Engine Technology (SET) Advanced Cluster Systems (ACS) has introduced its Supercomputing Engine Technology (SET), a software solution that parallelises modular sequential software applications, enabling optimal performance on multi-core platforms and clusters.

SET adopts the parallel paradigm In a GPU- optimised setup, the dual- processor blades, each with dual Intel Xeon

5500 series processors (up to six cores per CPU) and 96GB memory, for a total of 36 processing cores and 288GB of memory per chassis. Each blade has its own PCI Express 2.0 x8 expansion slot and internal Gigabit networking between blades, providing further space savings by eliminating external Ethernet switches.

NextStream can

be confi gured with two blades, each with dual

six-core Intel Xeon processors, 96GB RAM, and a dedicated Nvidia Tesla GPU computing processor on a full PCI Express 2.0 x16 slot. Additional expansion space is available for a third Tesla processor on one of the two blades, or for other expansion cards and storage.

deployed by investment banks to speed up pricing and risk calculations for complex portfolios. Its systems are also used to optimise reservoir modelling and seismic imaging programs used in oil and gas exploration. Other application areas that will benefi t from MaxCloud and Maxeler datafl ow computing technology include computational fl uid dynamics and bioinformatics.

of Message Passing Interface (MPI) while allowing programmers to think sequentially. This makes parallel programming accessible to a much larger pool of programmers. Rather than restrict parallel programming to the elite, SET makes it possible for applications to become parallelised

far more quickly and inexpensively than before. Target markets for the solution include software companies developing for legacy migration, cloud computing and high-performance applications; small and medium enterprises with high-performance

computing (HPC) requirements; military software development projects that need to manage legacy code; and HPC companies that require better software solutions for their hardware platforms. www.AdvancedClusterSystems. com


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