Share

IBM unveils new Linux-based servers to boost AI, deep learning

As you might suspect with the use of NVLink, IBM’s latest Linux servers are GPU-focused, and because of that, IBM says that they can offer up to 80% better performance-per-dollar than exclusively x86-based servers.

Advertisement

IBM on Thursday rolled out its latest Power8 processor, which is created to move data faster, and new servers with OpenPower features. The system runs on revamped Power8 chips and will give organizations an alternative to servers powered by Intel’s x86-based Xeon processors, according to IBM officials.

The interconnect is based on Nvidia’s homegrown NVLink technology, which has been in the works for years. The system embeds Nvidia’s NVLink high-speed interconnect technology inot the silicon to connect the Power8 chips with Nvidia’s new Tesla P100 Pascal GPU accelerators. Tencent is one of the largest ISPs in China, and after testing with IBM’s new servers, it found that it could churn through data-intensive workloads three times faster than its previous setup.

IBM has launched a new range of Linux-based servers which have been specifically engineered for high performance in tasks related to artificial intelligence, deep learning and advanced analytics – with a central mission to increase data centre efficiency.

The US Department of Energy’s Oak Ridge National Laboratory and Lawrence Livermore National Laboratory will be among the first to use the high performance computing variant, and IBM noted that these systems will likely build a foundation for new supercomputers IBM is now developing. IBM designed the servers to meet the needs of data-intensive enterprises and cloud service providers. “Accelerated computing that can really drive big data workloads will become foundational in the cognitive era”, said Doug Balog, general manager of POWER at IBM Systems, in a statement. These systems have tight Nvidia integration as well as high-speed interconnects. The ORNL systems will be a test-bed to develop applications for its Summit, an IBM-built supercomputer being delivered in 2017.

A good example of an application that will benefit is Kinetica, an in-memory database application that harnesses the computing power of GPUs. Nvidia said that NVLink enables 5x faster data transfers between CPUs and GPUs compared to PCIe.

The servers were developed through collaborations with the OpenPOWER Foundation, and IBM said they can deliver better performance than X86 competitors.

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

Advertisement

The S822LC for Big Data and the S821LC are available now, but the S822LC for High Performance Computing will ship on September 26, 2016.

IBM's new chip and servers are designed to accelerate artificial intelligence deep learning and advanced analytics. With built-in NVIDIA NVLink technology the new servers deliver higher levels of performance and greater computing efficiency than