Upcoming events

OPEN WORLD FORUM: The premier global summit on Open Source Software
Paris, GEORGE V Convention Center EUROSITES, 30 september - 1st October

More information and registration

Breakthrough new High Performance Computing designs on show at SEG 2010

Bull’s Extreme Computing BU will be exhibiting at the Society of Exploration Geophysicists’ 80th Annual Meeting, to be held this year at the Colorado Convention Center  in Denver, Colorado, USA.

The Bull booth  at SEG 2010 will feature both ‘thin-node’ HPC blade and GPU blade servers and ‘fat-node’ SMP servers, capable of running all major seismic and reservoir applications with maximum performance, density and power efficiency. The booth will also showcase the all-weather, all-climate sealed HPC container facility.

The bullx 500 series blade system, which incorporates both Intel x86 CPU and x86/double GPU hybrid blade systems. This family of super-dense HPC blades won the HPCwire Editor’s Choice award for ‘Best HPC product’ when first shown at the Supercomputing conference in November 2009.  

The bullx 500 blade systems incorporate the latest dual-socket Intel ‘Westmere’ 6-core processors, twelve memory DIMM slots and both SSD and SATA local disks. Eighteen blades fit into a 7U chassis with full non-blocking QDR connections to an integrated 36-port QDR Infiniband switch, making this a true high-performance blade system, suitable for demanding technical applications such as seismic processing, imaging and reservoir simulation.

The bullx 505 GPU blade system incorporates two NVIDIA Fermi ® GPUs inside each blade, together with double QDR InfiniBand. The architecture provides two full GPU-to-host PCIe-16x bandwidth per blade and dual QDR interconnect bandwidth, which has been specifically designed for parallel 3D seismic imaging such as Reverse Time Migration.  Bull can deliver a comprehensive service to migrate and parallelize codes, including native CUDA programming and HMPP GPU compiler middleware, with specialist expertise in GPU parallel programming and geophysical algorithms.  From initial proofs of concept through to production codes, Bull can help to deploy GPU applications which are robust, portable and able to generate stunning results.

Bull will also show the newly-announced bullx ‘supernode’ cc-NUMA (cache-coherent Non Uniform Memory Access ) shared-memory system, which can scale to 128 cores and 128 memory slots, achieving over one Terabyte of cache-coherent single memory address space per node. This technology will enable high performance in leading-edge seismic processing and imaging, (RTM, tomographic inversion, FWI, SRME), as well as memory-hungry basin modelling and reservoir simulation applications.

It is also a powerful OpenGL graphics server when using NVIDIA GPU co-processors, able to render ultra-large 3D seismic volumes and 3D models, for interpretation, well planning and field development purposes. Bull offers a remote OpenGL visualization solution based on this architecture, able to deliver secure interactive 3D OpenGL applications from this server to multiple desktop clients (see also portal-based HPCoD solution below).

Bull can deliver innovative platforms for seismic imaging, interpretation, basin modelling and reservoir simulation, and can offer different deployment possibilities, from fixed and temporary processing facilities. Bull’s mobull containerised HPC facility is a sealed and highly-dense container which can operate in extreme conditions of heat, humidity and cold.

Please come and visit us at Hall C, Booth 2366, and see how we can help you to reach your imaging and modelling goals !

For further details in advance, please contact hpc@bull.net.

More about SEG 2010