At its recent GPU Technology Conference (GTU), NVIDIA unveiled a swathe of solutions that reinforce the company’s commitment to becoming a provider of high performance computing (HPC) solutions to industries such as healthcare, automated vehicles and robotics.
The solutions announced at GTU also illustrate NVIDIA’s strategy to make its HPC solutions more accessible and available to a larger number of customers, and in different geographical locales, including traditional and cloud data centres, edge locations, and in the Omniverse, a platform developed by NVIDIA for hosting virtual 3D simulations of real physical world environments.
NVIDIA goes full stack
Although graphics processing units (GPUs) for the gaming industry continue to be NVIDIA’s largest business, the company has set its sights on becoming a major provider of computing capabilities across the IT industry. At GTU, NVIDIA co-founder and CEO, Jensen Huang, described the company as a “full stack platform vendor” that is able to support the accelerated computing requirements of specific industries.
Solutions announced at GTU include a new supercomputer, Quantum-2, a 400Gbps Infiniband networking platform that includes NVIDIA’s BlueField-3 data processing unit (DPU), and which is designed to provide flexible-to-consume, multi-tenant supercomputing capabilities for artificial intelligence (AI), data analytics and HPC applications. The launch of NVIDIA’s Quantum-2 comes at a time when the hyperscale cloud companies, including Amazon Web Services and Microsoft Azure, are also rolling out more flexible ways of delivering HPC to enterprise customers.
In addition to making its supercomputing and HPC solutions accessible to a broader market, NVIDIA is tailoring its solutions to the needs of specific industries. In June 2021 NVIDIA announced Cambridge-1, a supercomputer designed to support the UK’s Healthcare and Life Sciences industries. At GTU, NVIDIA unveiled Earth Two (E-2), a powerful supercomputer that will run a digital twin of the earth in order to simulate and predict the effects of climate change.
Although many of the details of E-2 will not be released until some time in the future, it is understood it will leverage NVIDIA’s Omniverse platform. The initiative further illustrates the extent of NVIDIA’s ambitions in the dynamic market for computing solutions.
Reaching the edge
Another new solution announced at GTU, the Jetson AGX Orin, represents NVIDIA’s latest effort to target the growing market for computing across distributed edge locations. Built on NVIDIA’s Ampere architecture, Jetson AGX Orin delivers a claimed 6 times the processing power of its predecessor, Jetson AGX Xavier, equivalent to up to 200 trillion operations per second. Edge computing opportunities are being targeted by a raft of IT service and solution providers with an equally broad range of solutions.
However, NVIDIA is differentiating its approach by visibly tying its solutions to specific use cases and by providing additional software to help organisations apply edge computing to their own specific requirements. These include NVIDIA Isaac Sim on Omniverse for robotics, NVIDIA Clara Holoscan software development kit (SDK) for healthcare, and NVIDIA DRIVE for autonomous driving.
Differentiation will be crucial as NVIDIA doubles-down on efforts to target emerging opportunities in supercomputing and edge computing. NVIDIA’s ambitions have not gone unnoticed by competitors such as Intel, Qualcomm and AMD, all of which are, in different ways, targeting opportunities in NVIDIA’s traditional GPU business, and in the emerging markets for supercomputing and edge computing.