Paving the Way for AI-enabled Supercomputing
By Chris Bergey, SVP & GM, Infrastructure Line of Business, Arm
It’s been a huge year for Arm and our ecosystem of infrastructure innovators as we work together to deliver the technologies and solutions that will ultimately enable the trillions of devices coming online by 2035. Arm continues to be the computing foundation for the convergence of AI, 5G, and IoT, fueling a massive change to today’s data consumption models from the cloud to the edge. Just a few years ago, an Arm-based HPC was difficult to fathom for many. Today, Arm is becoming established as the compelling platform for AI and HPC innovation to build upon, scaling from the largest clusters to the emerging edge.
Earlier this year our strategic partner NVIDIA announced their intent to more broadly support Arm CPUs and make its full stack of AI and HPC software available to Arm partner platforms. This was a significant announcement for both companies. NVIDIA GPUs power some of the fastest supercomputers in the world, including the fastest, Summit at Oak Ridge National Labs. These systems will continue to play an important role in future exascale deployments where supercomputers will run at 5x – 10x the speed of today’s best, but achieving this level of performance goes beyond the hardware.
Both Arm and NVIDIA know ecosystems and are great at building the software stacks needed for the success of our partners and end-users. Enabling NVIDIA GPU’s to run seamlessly in conjunction with Arm processors provides access to new markets for our HPC partners and opens further market opportunities for NVIDIA’s market leading ML/AI technology. This important collaboration highlights our similar vision for enabling efficient compute from the datacenter to the edge and draws our companies closer in terms of collaboration on ecosystem goals.
Building on this momentum, today at SC’19, Jensen Huang, CEO of NVIDIA, introduced a technology blueprint for companies to quickly and easily build GPU-accelerated Arm-based servers. The reference platform, which consists of hardware and software building blocks, aims to enable more high-performance computing in a fast-growing range of applications across science and research. Furthermore, NVIDIA, Arm, and our collective ecosystem are already demonstrating new proof points of solutions with Arm-based CPUs and NVIDIA GPUs.
Source: NVIDIA
This next step in delivering high-performance capabilities is significant because the combination of NVIDIA’s CUDA-accelerated computing and Arm’s energy-efficient CPU architecture will give the HPC community the advanced tools and technologies they need to enable compute from the edge all the way through to the largest HPC deployments. We are working diligently to deploy Arm technology all of the HPC space, addressing some of the world’s most complex research challenges. We are very excited about the progress that has been made this year on joint enablement of the combined Arm and NVIDIA ecosystem. But, this is just the beginning.
Congratulations to the Arm and NVIDIA teams for a very exciting milestone achievement!
Any re-use permitted for informational and non-commercial or personal use only.