InfiniBand and OpenFabrics Software (OFS) Continue to Lead the TOP500 List in Efficiency and Overall Systems
With today’s supercomputers pushing nearly 100 petaflops, the need for efficient, low latency performance is higher than ever. TOP500, the organization that provides statistics on the world’s 500 most powerful supercomputers, released its latest report and, as with previous lists, InfiniBand leads the way. The 47th edition of the bi-annual list shows that 205 of the fastest commercially available systems are accelerated by InfiniBand. InfiniBand-based systems outperformed competing technologies in overall efficiency, scoring an 85 percent list average for compute efficiency – with one system even reaching an incredible 99.8 percent.
View our official announcement for more details.
InfiniBand-based Supercomputer to Power New Discoveries at the Library of Alexandria
In 2002, Bibliotheca Alexandrina was constructed to commemorate the Library of Alexandria’s remarkable history and lasting notoriety. To honor the original library’s legacy of preserving knowledge and pursuing new discoveries, Bibliotheca Alexandrina is building an advanced supercomputer. The new supercomputer will feature high-density servers powered by high-performance InfiniBand interconnects. Once completed, it will support a variety of research fields, including bioinformatics, data mining, physics, weather forecasting, resource exploration/extraction and cloud computing.
For additional information on the library’s supercomputing project, check out our blog.
InfiniBand Accelerating Africa’s Fastest Supercomputer
The Council for Scientific and Industrial Research (CSIR) in Cape Town, South Africa, recently unveiled the new Lengau supercomputer, which will be the fastest on the African continent. The Lengau supercomputer, fittingly named after the Setswana word for cheetah, clocks in at one petaflop of peak performance. Sporting a 19-rack system using EDR InfiniBand interconnect, Lengau is more powerful than CSIR’s other supercomputers yet it occupies less physical space. The new South African system was built in hopes to not only drive new research, but to help the country’s economy as well.
To learn more about the CSIR’s new supercomputer, read ZDNet’s coverage.
NVMe over Fabrics is New, Fast and Gaining Popularity
NVMe over Fabrics (NVMeF) may be young, but the implications of this new technology are far-reaching for all flash arrays. NVMeF allows a data center to aggregate and abstract a pool of NVMe flash storage, enabling remote access to NVMe devices over RDMA fabrics. The NVMeF standard, as defined by NVM Express Inc., is almost complete. Although all flash arrays hold great promise, they need RDMA adapters to unlock their full potential.
Read Virtualization Review’s article for more on NVMeF products and use cases.
IBM’s Summit Supercomputer Will Be the Fastest in the World
Shortly after the TOP500 List named China’s Sunway TaihuLight the fastest supercomputer in the world, IBM revealed details about Summit, the company’s newest supercomputer that will unseat its Chinese competitor. The system will use a dual-rail Mellanox EDR InfiniBand interconnect to provide a high performance I/O throughput. Originally planned to be capable of 150 petaflops, the machine will instead achieve up to 200 petaflops, beating Sunway TaihuLight’s capability of 93 petaflops. Summit will be deployed in the U.S. Department of Energy’s Oak Ridge National Laboratory in early 2018.
For further details on the Summit supercomputer, read BetaNews’ coverage.
Big Data and the Need for Speed
Today, there is an unprecedented amount of data creation, especially with the growing prevalence of connected devices and the Internet of Things. As the rate of data creation increases, so does the need to quickly process that information. This is putting an excessive strain on the existing hardware of enterprise networks and data centers. Data acceleration technologies, such as RDMA over Converged Ethernet (RoCE), can step in to make an entire data center more efficient by enabling higher CPU utilization when moving data between serves via hardware offloads. This increases overall system performance, lowers CPU idle time and saves overall costs by eliminating the need for additional servers.
Read more on RoCE’s role within the world of Big Data on our blog.
Mellanox Delivers ConnectX-5, the Next Generation of 100G InfiniBand and Ethernet Smart Interconnect Adapter
Mellanox announced its new ConnectX-5 interconnect adapter that allows smart offloading and enables the highest application performance while maximizing data center ROI. The new intelligent ConnectX-5 100G adapter enables advanced real-time in-network computing engines to drive business opportunities and new technological developments.
View Mellanox’s official announcement to learn more.
Oracle and Mellanox to Collaborate on High Speed Cloud Networking Standards
Oracle and Mellanox announced a partnership to drive interoperability between their products and to jointly develop further industry standards for InfiniBand. Solutions designed to the InfiniBand specification not only provide extreme bandwidth and ultra-low latency for fast, agile and secure cloud infrastructure, but they are also based upon an open industry standard and available from multiple vendors. Oracle and Mellanox InfiniBand products are backward compatible with the previous generations of InfiniBand, protecting customers’ existing investments in InfiniBand-enabled applications.
Read Oracle’s announcement for more details.
Mellanox Solutions Accelerate the Fastest Supercomputer in the World
Mellanox announced that its interconnect solutions accelerate the world’s fastest supercomputer at the supercomputing center in Wuxi, China. The new leading supercomputer delivers 93 Petaflops, with nearly 41 thousand nodes and more than ten million CPU cores. InfiniBand interconnects now accelerate nearly 50 percent of the TOP500 List’s Petascale systems.
For more on the announcement, check out the official press release from Mellanox.
HPE Unveils Software Defined High-Performance Computing Platform
HPE announced a new portfolio of High Performance Computing (HPC) solutions, including enhancements to its Apollo servers with next-generation EDR 100Gb/s InfiniBand technology. The EDR upgrade will provide customers with improved response times and alleviate bottlenecks that impact application performance. HPE states that the new developments will simplify the deployment and management of HPC solutions, ensuring companies of all sizes can accelerate their HPC projects and create competitive differentiation for their business.
View HPE’s press release for more on the Apollo server upgrades.
Flash Memory Summit (FMS) 2016, August 9-11, 2016
Attending Flash Memory Summit 2016 in Santa Clara, CA? Be sure to take advantage of two educational opportunities from InfiniBand experts. First, representatives from IBTA member companies will be participating on a panel to discuss the value of RDMA interconnects to the NVMe over Fabrics standard. The panel will be held Tuesday, August 9 at 9:45 a.m. Also, stop by the “Beer, Pizza and Chat with the Experts” on Tuesday, August 9 at 7:00 p.m., where Bill Lee, IBTA Marketing Work Group Co-Chair, will host a table on RDMA interconnects. This informal event gives attendees a chance to sit and “talk shop” with a diverse group of experts on the latest trends happening within the flash industry.
For more information on the event, visit the website.
Intel Developer Forum (IDF) 2016, August 16-18, 2016
Held annually in San Francisco, CA, the Intel Developer Forum 2016 is aimed at innovators and developers who want to stay on top of the latest technological trends from Intel. The agenda includes keynotes from leading Intel experts, hands-on technical sessions and demos, and in-depth technology showcases. The program includes a variety of InfiniBand and HPC related tracks, including Data Analytics, Cloud and Software Defined Infrastructure, and the Internet of Things.
For more information visit the event website.
For questions or comments, please contact firstname.lastname@example.org