New OpenPOWER Servers Accelerate Deep Learning with NVLink

Print Friendly, PDF & Email

s822lcToday IBM unveiled a series of new servers designed to help propel cognitive workloads and to drive greater data center efficiency. Featuring a new chip, the Linux-based lineup incorporates innovations from the OpenPOWER community that deliver higher levels of performance and greater computing efficiency than available on any x86-based server.

Collaboratively developed with some of the world’s leading technology companies, the new Power Systems are uniquely designed to propel artificial intelligence, deep learning, high performance data analytics and other compute-heavy workloads, which can help businesses and cloud service providers save money on data center costs.

The three new systems are an expansion of IBM’s Linux server portfolio comprised of IBM’s specialized line of servers co-developed with fellow members of the OpenPOWER Foundation. The new servers join the Power Systems LC lineup that is designed to outperform x86-based servers on a variety of data-intensive workloads.

Early testing with Tencent, one of the world’s largest Internet service providers based in China, has shown that a large cluster of the new IBM OpenPOWER servers was able to run a data-intensive workload three times faster than its former x86-based infrastructure. Moreover, this result was achieved while reducing the total number of servers used by two-thirds. Given the significant cost benefits of using fewer servers to deliver faster performance, the company is now integrating the new servers into its hyperscale data center for big data workloads.

The user insights and the business value you can deliver with advanced analytics, machine learning and artificial intelligence is increasingly gated by performance. Accelerated computing that can really drive big data workloads will become foundational in the cognitive era,” said Doug Balog, General Manager of POWER, IBM Systems. “Based on OpenPOWER innovations from partners such as NVIDIA, our new OpenPOWER Linux servers with POWERAccel set a new standard for these workloads compared with x86 processor-based servers.”

Industry’s Most Advanced Acceleration Platform to Fuel Cognitive Era

Industry advancements including artificial intelligence, self-driving cars, real-time fraud protection and the development of life-saving drugs require new levels of computing power. To address this need, IBM has taken an open development approach to system design to more fully take advantage of acceleration technologies.

IBM has collaborated with fellow technology industry leaders through the OpenPOWER Foundation to radically re-design the platform at the chip and system levels by incorporating the use of a wide range of accelerators to achieve greater levels of performance than available on traditional commodity servers. Through a family of interconnect innovations collectively known as POWERAccel, contributors to the OpenPOWER ecosystem will continue to develop systems and other solutions on the POWER platform that are optimized for accelerated applications.

A major achievement stemming from open collaboration is the new IBM Power System S822LC for High Performance Computing server. Featuring a newly designed processor, the IBM POWER8 with NVIDIA NVLink, the new system delivers an industry-first innovation available only on POWER’s open architecture. The new system directly connects the new IBM POWER8 processor with NVIDIA Tesla P100 Pascal GPUs through NVIDIA NVLink, a high-speed, energy-efficient bidirectional interconnect. NVIDIA NVLink is embedded at the silicon level and incorporated into the overall system design. This tight coupling of IBM and NVIDIA technology enables data to flow 5x faster than on an x86-based system.

The open and collaborative model of the OpenPOWER Foundation has propelled system innovation forward in a major way with the launch of the IBM Power System S822LC for High Performance Computing,” said Ian Buck, VP of Accelerated Computing at NVIDIA. “NVIDIA NVLink provides tight integration between the POWER CPU and NVIDIA Pascal GPUs and improved GPU-to-GPU link bandwidth to accelerate time to insight for many of today’s most critical applications like advanced analytics, deep learning and AI.”

Anticipating the introduction of this industry-first technology, several businesses, research organizations and government bodies have pre-tested early systems and placed their orders. Among those first in line to receive shipments are a large multinational retail corporation and the U.S. Department of Energy’s Oak Ridge National Laboratory (ORNL) and Lawrence Livermore National Laboratory (LLNL).

ORNL and LLNL will use the new systems as a development platform for optimizing applications to take advantage of the built-in NVIDIA NVLink interface technology. The systems will serve as an early-generation test bed for developing demanding applications for Summit and Sierra, the next generation supercomputers that IBM expects to deliver respectively to ORNL and LLNL in 2017.

The combination of IBM POWER8 and NVIDIA Pascal via NVLINK will accelerate our mission-critical applications,” said Bronis R. de Supinski, the Chief Technology Officer of Livermore Computing (LC) at Lawrence Livermore National Laboratory. “The significant reduction in programming complexity that they support was a major contributor to our decision use these technologies in systems that will serve as precursors of our planned Sierra system.”

“The NVLink technology on the POWER platform provides coherency among the multiple memory hierarchies in the CPUs and GPUs,” said Arthur S. (Buddy) Bland, OLCF Project Director at the U.S. Department of Energy’s Oak Ridge National Laboratory. “As a long-time user of GPUs, we believe that this will improve the performance of our applications and make it easier for the users to deliver great science.”

The two additional LC servers available today – the IBM Power System S821LC and the IBM Power System S822LC for Big Data – can also leverage GPU acceleration technology to increase system performance levels on a variety of accelerated applications. Clients can attach NVIDIA Telsa K80 GPU accelerators by PCIe.

Additionally, clients can tap into the POWERAccel Coherent Acceleration Processor Interface (CAPI) available on the S821LC and S822LC for Big Data models for a high-speed interconnect with FPGA accelerators.

Lower Costs, Less Server Sprawl

Fully compatible in Linux-based cloud environments, IBM’s Power LC servers are optimized for data-rich applications and can deliver superior data center efficiency. The new servers can be acquired at a lower price, and can deliver 80% more performance per dollar spent, over x86-based systems.(1) This efficiency enables businesses and cloud service providers to lower costs and combat data center sprawl.

Pricing, Specifications and Availability

The new Power LC lineup is price-advantaged over comparatively configured Intel x86-based servers, costing 30% less in some configurations. Online pricing begins at $5999. Additional models with smaller configurations and lower pricing are available through IBM Business Partners.

Sign up for our insideHPC Newsletter