Twitter Facebook Google Plus Linkedin email Print Pdf


Bull launches bullx supercomputer suite

Posted on 10 November 2010 by Pierre Picard
  • Comprehensive Extreme Computing software suite, designed to help customers effectively deploy and easily administer HPC clusters
  • Extreme scaling, from entry-level to very large supercomputers

 ‘bullx® supercomputer suite’, the new generation of Bull’s HPC software suite, delivers a comprehensive, Extreme-class solution that helps customers to effectively deploy and easily administer their HPC clusters. It also provides everything that users need to develop and run their HPC applications. The suite leverages Bull’s years of experience in delivering HPC software for the most demanding customers, combined with its significant investment and continuous efforts in R&D.

 Available as either a Standard or Advanced edition, with an additional Extreme Pack option, bullx supercomputer suite is designed to address a variety of needs: from those of small supercomputers based on a few hundred cores, where ease of use and flexibility are key, scaling right up to demands of the very largest supercomputers based on several tens of thousands of cores, where very high availability and reliability, self management, an optimized energy footprint, very high I/O data throughput, a whole development environment and a fine-tuned MPI library are all essential features.

 ”We have been testing bullx supercomputer suite for six months now,” said Jacques-Charles Lafoucrière, High-Performance Systems Division Leader at the CEA/DAM, the Military Applications Division of the French Alternative Energies and Atomic Energy Commission. “We have been impressed by the scalability and performance delivered by this new generation of HPC software. Thanks to both the advanced power management and automated system health management features, we have noticed a significant increase in productivity while reducing the overall operational cost.”

 ”We have designed the bullx supercomputer suite to meet customers’ requirements both in terms of extreme scalability and availability on the one hand, and flexibility and ease of use on the other,” said Eric Monchalin, Software Director at Bull’s Extreme Computing Business Unit. “Ranging in scale from entry-level clusters to the most powerful supercomputers in the world, the bullx supercomputer suite is a major step ahead in HPC software technology. It will help Data Centers drastically reduce their operational costs, simplify their daily systems administration tasks and efficiently run their development and production activities.”

 bullx supercomputer suite includes:

  • bullx Management Center (bullx MC) a comprehensive, Web-based range of supercomputer management software programs, featuring an Extreme-speed image-based provisioning engine, powerful rule-based energy management, automated system health management, SLA-driven maintenance support and a centralized multi-level monitoring dashboard
  • bullx Parallel File System (bullx PFS) is based on the bullet-proof Lustre parallel file management system specifically configured for Extreme scalability and better reliability. Users of bullx PFS benefit from the fact that the ‘Shine’ Lustre administration tool is fully integrated with bullx Management Center, providing advanced Lustre monitoring for silent deadlock and denied file access event detection
  • bullx Batch Manager (bullx BM) is based on the SLURM Open Source resource manager, which has been further enhanced by Bull to provide additional functionality including topology-aware resource allocation and advanced placement policies, for improved application execution performance and better overall quality of service
  • bullx MPI is compatible with the MPI 2.1 standard. It is built around OpenMPI, which has been enhanced by Bull with many new features such as effective abnormal pattern detection, network-aware collective operations, and multi-path network failover, to increase reliability and resilience. It has also been designed to significantly boost the performance of parallel MPI applications thanks to full integration with bullx PFS and bullx BM. bullx MPI is scalable up to several tens of thousands of cores
  • bullx Development Environment (bullx DE) is a collection of tools and libraries providing full support to users throughout the entire project lifecycle (from design to production). bullx DE features best-in-class products covering compiling, debugging, profiling and optimization activities, with specific focus on MPI companion tools.

 All these components will run under bullx Linux®, which is a Linux distribution compatible with Red Hat® Enterprise Linux® and specifically adapted for the demands of Extreme Computing. Linux-basic services configurations are optimized and features such as CPU-SET and GPU-SET, and Lazy Page Migration, as well as OS jitter reduction, have all been incorporated to deliver a significant performance improvement for HPC applications. bullx Linux will also include an enhanced and scalable OFED™ (OpenFabrics Enterprise Distribution)-based InfiniBand® stack, enriched with management tools tailored for supercomputers with over 10,000 nodes.

 Support and availability

bullx supercomputer suite customers will benefit from Bull’s highly professional support offering, including personalized and pro-active services. bullx supercomputer suite is scheduled to be released by end Q1 2011 on bullx Linux and Red Hat Enterprise Linux, and at a later date on Suse® Linux® Enterprise Server.


For more information >>>

Comments are closed.