IBM, Nvidia to Partner on GPU-Accelerated Power Servers
IBM will begin using Nvidia's GPU accelerators in its Power-based servers, a move aimed at enabling organizations to better process and analyze massive amounts of data and one that will expand the reach of such accelerators into the enterprise.
IBM and Nvidia made the announcement Nov. 18 at the SC '13 supercomputer show in Denver.
Big Blue will develop enterprise software capable of leveraging GPU accelerators and will work with Nvidia in integrating capabilities in Nvidia's Tesla GPUs with IBM's Power processors. Doing so will enable companies to bring supercomputing capabilities into their data centers, which is important with the rapid growth in the amount of data being generated, and the increasing need to more quickly process and analyze that data.
"Companies are looking for new and more efficient ways to drive business value from Big Data and analytics," Tom Rosamilia, senior vice president of IBM Systems and Technology Group and integrated supply chain, said in a statement. "The combination of IBM and Nvidia processor technologies can provide clients with an advanced and efficient foundation to achieve this goal."
Sumit Gupta, general manager of Tesla accelerated computing products at Nvidia, said the partnership is a win for both his company and IBM.
IBM officials "are going to look for opportunities to accelerate whenever they can whatever they can," Gupta told eWEEK. For Nvidia, the partnership broadens the market reach for its GPU accelerators from supercomputers to enterprise systems.
"The growth opportunity is the enterprise part of this," he said, noting that Intel will see increasingly competition in the data center from the likes of IBM and ARM. "Over the next year, the data center will become very diverse. … It will become a pretty interesting time in the market."
Nvidia and Advanced Micro Devices for the past several years have promoted their graphics capabilities as accelerators that enable OEMs to build supercomputers with improved performance and greater parallel processing capabilities that don't drive up the power consumption of the systems. Intel last year introduced the first of its many-core Xeon Phi x86-based coprocessors, a key part of the company's neo-heterogeneity computing strategy of offering choices of processors that include common development tools and platforms.
ARM and its chip-making partners are expecting to begin offering 64-bit ARM-based server chips later next year and into 2015, challenging Intel in low-power, dense servers. At the same time, IBM in August unveiled its plan to license its Power technology in hopes of giving enterprises an alternative to Intel chips in cloud computing and hyperscale computing environments. Nvidia and Google were among the companies helping IBM launch the OpenPower Consortium, which aims to create an ecosystem around the effort.
IBM will combine its upcoming Power8 CPUs with Nvidia's GPUs, including the company's new Tesla K40 offerings, which the company also announced Nov. 18. The new products will help bring greater performance and power efficiency to such workloads in such areas as engineering, science, big data analytics and high-performance computing (HPC), according to the two vendors.
In addition, IBM Power systems will support applications developed with Nvidia's CUDA programming platform, and also will make its Rational enterprise software development tools available to supercomputing programmers.
"This partnership will bring supercomputer performance to the corporate data center, expanding the use of GPU accelerators well beyond the traditional supercomputing and technical computing markets," Ian Buck, vice president of accelerated computing at Nvidia, said in a statement. "It will also provide existing supercomputing and high performance computing customers with new choices and technologies to build powerful, energy-efficient systems that drive innovation and scientific discovery."