Canonical has been working closely with NVIDIA for many years to fuel innovation and support open source software with the power of accelerated processing. That already allowed us to jointly deliver GPU acceleration into Linux, OpenStack and container workloads on traditional datacenter servers.
We continued working together, with Ubuntu forming the base operating system for NVIDIA DGX systems, including the latest NVIDIA DGX A100 system. Today we are announcing the support of a new class of acceleration at the edge, on the NVIDIA EGX Edge AI platform, powered with Ubuntu.
With EGX, NVIDIA brings GPU-accelerated servers to the enterprise and the edge to the world’s largest industries. In combination with the NVIDIA BlueField-2 data processing unit (DPU), the next-generation SmartNICs, EGX is making the latest technologies in AI, encryption and storage acceleration accessible to enterprises to be deployed anywhere. Canonical is excited to collaborate with NVIDIA to provide the software platforms and infrastructure behind the new EGX Edge AI platform, solving some of the most challenging problems from the data center to the edge today.
NVIDIA’s certification program allows engineers to quickly ID edge servers pre-validated by NVIDIA for GPUs and DPUs, along with containers, pre-validated AI models, and collections that combine cutting-edge tools in one place. The NVIDIA CUDA and DOCA SDKs enable developers to build AI, crypto, encryption and more services across diverse verticals including retail and healthcare.
Ubuntu Core to extend security at the edge
The NVIDIA EGX AI platform will benefit from all the latest technologies from Canonical, starting with the latest Ubuntu Server 20.04 LTS release – offering the 5.4 kernel and all associated security enhancements – but also the edge/IoT ultra-secured version called Ubuntu Core. For a couple of years, Core has become the operating system of choice for Embedded 2.0, enabling millions of developers to use familiar tools from the cloud to deploy at the edge and reduce time to market. Ubuntu Core offers over-the-air atomic and delta upgrades, allowing reliable deployments in areas with patchy network connectivity. With MicroK8s, customers access their favourite container technologies with a pure-upstream Kubernetes that allows them to deploy within 60 seconds, with zero ops. Enterprises can also leverage the trusted and secure Snap Store ecosystem (apps and private stores) with 1,000s of applications, without compromising on security and control to focus on delivering their core value.
“AI has already become a must in modern data center architectures. NVIDIA EGX and SDKs bring the best of those datacenter acceleration capabilities to the highly distributed realm for enterprises with Ubuntu, the leading software platform for innovators in AI/ML. We are truly excited about the revolutionary opportunities this brings to solve complex problems such as smarter cities, safer transports and real-time speech recognition,” said Regis Paquette, VP Global Alliances at Canonical.
“Canonical has been working closely with NVIDIA for many years to ensure that their Ubuntu operating system benefits from the latest NVIDIA hardware and AI software advancements,” said Justin Boitano, vice president and general manager, enterprise and edge computing, NVIDIA. “Extending this collaboration to edge platforms and converged platforms allows developers to use the same operating system from cloud to edge.”