EuroPVM/MPI 2006

Sponsoring Companies talk about their Products and Visions

[print version]
The vendor session at EuroPVM/MPI has a long tradition of not being marketing-hype, but typically offer solid technical information and discussion on a vendor's products. For this reason, it is in fact not an annoyance, but a highlight of the conference.

Platinum Vendor Session

Time: 18th Sep 2006, 17:30 - 17:50

Location: K1/K2

IBM: Extending High Performance to High Productivity

David Klepacki, IBM T.J. Watson Research Center

slides of the presentation [PDF]

Intel: Intel Cluster Tools - your way to optimal experience on Intel cluster platforms

Alexander Supalov

We will provide an update on the status and plans of the Intel Cluster Tools product line. The focus of the presentation will be on the Intel MPI Library and related technologies as well as performance and scalability issues for upcoming cluster systems. We will give an overview of the Intel Cluster Toolkit enabling effective development and highest performance of applications on Intel platforms.

slides of the presentation [PDF]

NEC: NEC High Performance Computing Technology

Jörg Stadler

slides of the presentation [PDF]

Etnus: Debugging Parallel Programs for Multi-Core Chip Architectures

Christoper Gottbrath, Totalview Product Manager

Adapting parallel applications to multi-core processors often involves moving to program architectures that are both multi-process and multi-threaded. These hybrid applications, which incorporate both shared memory and message passing, are especially challenging to debug. This talk will describe how the parallel architecture and asynchronous thread control capabilities of the Total View debugger will help you quickly identify and resolve bugs in these otherwise challenging applications.

slides of the presentation [PDF]






































Gold Vendor Session

Time: 19th Sep 2006, 10:00 - 11:00

Location: K1/K2

Dolphin: Take the Expressway

Dr. Werner Butscher

Werner will present the current state and the defined roadmap of Dolphin's interconnect technology which provides low-latency, high-bandwidth remote memory access and RDMA. Results from current applications will illustrate the advantages of the current products. These advantages will increase with the upcoming product DolphinExpress. DolphinExpress is a PCI-Express-based switching interconnect, designed from scratch based on more than 10 years of high-performance interconnect experience. It offers extremely low latency and advanced RDMA features combined with full PCI-Express interoperability.

slides of the presentation [PDF]

Sun: Sun and Open MPI: investing in innovation and the community

Leonard Wisniewski

After ten years of developing its own ClusterTools product, Sun recently joined as an active member of the Open MPI open source team.  Sun believes in sharing innovation and growing communities and has a long history of sharing from NFS in 1984 to Solaris and now Open MPI today.  In their first release as an Open MPI member, Sun has made a number of contributions, including a plug-in to enable launching Open MPI jobs via Sun N1 Grid Engine as well as investing resources in supporting the quality and efficiency needed by a production-ready MPI implementation such as Open MPI.  Supported by a growing Open MPI community, an impressive line of Sparc and x64 servers, its Solaris software stack, and large successful customer installations such as Tokyo Institute of Technology, Sun is committed as a major player in the HPC market.

slides of the presentation [PDF]


HP: HP-MPI - a Universal Approach

Henry Strauss

slides of the presentation [PDF]


Microsoft: Compute Cluster Server 2003

Dennis Crain, Microsoft HPC Program Manager

For customers solving complex computational problems, Microsoft Windows Compute Cluster Server 2003 accelerates time-to-insight by providing a High-Performance Computing (HPC) platform that is simple to deploy, operate, and integrate with existing infrastructure and tools. Prescriptive setup procedures simplify network configuration, remote loading of nodes using Remote Installation Services, node configuration, and security setup.  The integrated Job Scheduler can be accessed via command-line interface, or through several API's is provided for submitting and managing cluster workloads.  The job console allows you to manage jobs in much the same fashion that you queue and manage print jobs today.  Active Directory integration provides end-to-end user and security management, while the Microsoft Management Console supports extensible snap-ins and integration with Microsoft Operations Manager.

slides of the presentation [PDF]

Platinum Sponsors

IBM
NEC
Etnus
Intel

Gold Sponsors

Dolphin ICS Microsoft HP Sun Research Centre Jülich
 
 
Proceedings by
Springer
 
Organized by
NEC
Research Centre Juelich
 
This site uses Etomite CMS