SGI extends High Performance Computing innovation on path to Exascale

SGI has announced new next generation supercomputers, enhanced software, and heightened services to accelerate time to discovery at greater scale and lower cost. SGI highlights areas of continuing innovation in high performance computing (HPC) that bring exaflop-capable systems into view by 2020.

“Exascale is the next frontier in high performance computing and will enable computational breakthroughs spanning climatology, energy, security and the speed of human thought,” said Jorge Titinger, president and CEO of SGI. “SGI is on a path to deliver systems capable of an exaflop by 2020. While these systems will be designed for the most extreme HPC environments, our progression will bring practical solutions to tera- and petascale problems seen today.”

Additions to SGI’s high performance computing (HPC) portfolio introduced today include:
· SGI ICE XA, a 6th generation distributed-memory supercomputer enabling breakthroughs in science, engineering, and government at greater speed, scale and efficiency.
· SGI UV 300 and SGI UV 30EX, in-memory supercomputers that enable complex data analytics, visualization and real-time streaming
· SGI Management Center 3.0, advanced HPC software that simplifies and improves system, health, and power management to reduce operating costs
· SGI Remote Services that help maximize system uptime, improve operational efficiency, and increase productivity

“Beyond faster processors, HPC system vendors must redefine how supercomputers are powered, cooled, and managed to reach exascale,” said Earl Joseph, program vice president, High-Performance Computing at IDC. “Reflected in SGI’s new offerings and aspiring to exaflop-capable systems, SGI remains at the forefront of HPC innovation, helping customer solve today’s business challenges and future-proofing for tomorrows.”

SGI ICE XA – Leading performance and cooling efficiency at greater scale
New SGI ICE XA is the sixth generation of the world’s most powerful distributed-memory supercomputer. The Linux-based system features new Intel® Xeon® E5-2600 v3 processors, with which SGI holds the performance world record, enabling larger, more detailed problems to be analysed. Extending SGI’s leadership in energy efficiency, ICE XA also features second generation E-cell warm water cooling technology, providing at least a 30% increase in cooling efficiency compared to competing technologies, saving customers millions of dollars in energy costs. Additionally, ICE XA can scale out to tens of thousands of nodes and 100s of thousands of cores running SUSE® or Red Hat Linux®, providing customers the ability to scale from teraflops to tens of petaflops

SGI UV 300 – The most powerful in-memory supercomputer for data-intensive workloads
Building on 20 years of in-memory computing expertise and extending SGI’s UV line of servers, the new SGI UV 300 and SGI UV 30EX are designed for complex data analytics, visualization and real-time streaming. Running SUSE Linux® Enterprise Server or Red Hat® Enterprise Linux, SGI UV 300 delivers up to 32 sockets (480 cores) and 24TB of coherent shared memory in a single system. Powered by Intel® Xeon® E7-8800 v2 processors and the ability to add Intel® Xeon Phi™ or NVIDIA® GPU accelerators, seventh generation SGI NUMAlink® ASIC technology provides extreme bandwidth and ultra-low latency network interconnects. Nodes can be further augmented with Intel® Xeon® Phi™ Coprocessors or NVIDIA GPU Accelerators to meet the needs of diverse environments. Designed for data-intensive environments, SGI UV 30EX is a 4-socket server providing up to 3TB of in-memory computing power, and can be upgraded to the scalable UV300 if your infrastructure needs change.

SGI Management Center 3.0 – Advanced HPC software to reduce complexity, saving time and resources
An integral component of the SGI Management Suite, the new SGI Management Center 3.0, leverages SGI’s experience delivering enterprise compute solutions for environments with extreme web traffic. Advanced capabilities include provisioning operating systems across clusters automatically–even with tens of thousands of nodes–in a fraction of the time of competing technologies. Power provisioning can be limited at a node level to conserve power on idle nodes and fit total system use within an available power envelope. With Management Center 3.0, event logs capture time-stamped history for analysis to proactively maintain system performance.

SGI Remote Services – 24x7 SGI system monitoring
New SGI Remote Services maximize system uptime, improve operational efficiency and increase productivity. With 24x7 systems monitoring, administrators can benefit from ongoing data gathering, component failure and change alerts, and help enable preemptive action to avoid downtime. Optional secure system access further enables SGI support personnel to diagnose and resolve issues quickly. SGI Remote Services are included in the SGI Management Suite which supports SGI HPC Compute solutions worldwide.

SGI on A Path to Deliver Exascale
On the path to exascale, SGI will continue to focus on innovation through:
• Bringing storage closer to processors coupled with smart data placement to accelerate I/O and reduce power consumption.
• Managing power consumption at a job level.
• Using higher facility water temperatures and full liquid immersion to further reduce cooling costs.
• Predicting hardware failure with node location through continual analysis of system data.
• Utilizing in-memory technology to provide “computational steering” for massive jobs on HPC to avoid time and resource loss.