The NVIDIA-Certified Systems program has seen tremendous uptake since it launched last year. More than 20 partners offer over 100 certified systems of all types, from the highest end servers built for high performance computing to lean, yet powerful laptops and desktops that can run high-quality visualization and state-of-the-art data science algorithms.
Now the program has expanded to include edge systems — one of the biggest areas of growth, with computing tied to devices for applications such as video analytics, robotics and 5G communication.
Edge computing is driven by the requirement for low-latency, real-time results as well as the desire to contain data transit costs without resorting to compromises such as reducing image resolution. Edge systems run in locations physically close to where data is collected or processed, in settings such as retail stores, factory floors and cell phone base stations.
NVIDIA-Certified edge servers are validated to run a range of accelerated workloads with the best performance, and must also meet specific requirements due to their deployment location.
Security in Edge Systems
NVIDIA certification includes tests to ensure that key technologies, such as the Trusted Platform Module, function properly in systems. Because edge systems are often in easily accessible locations, such as a retail store’s IT closet, malicious actors could take them off site and try to extract data or install malware.
TPM counters physical threats like these by ensuring that the system can only boot with firmware and software that has been digitally signed and unaltered. Additional security checks such as signed containers ensure that applications haven’t been tampered with, and disk volumes can be encrypted with keys that are securely stored in the TPM.
Other features that improve zero-trust security at the edge, such as network encryption acceleration offload, can be leveraged in certified systems deployed with NVIDIA ConnectX-6 network adapters or NVIDIA BlueField-2 data processing units.
NVIDIA certification also includes a specific industrial edge category for systems deployed in places where environmental conditions are more extreme.
Systems on a factory floor for automation control or in an enclosure next to a telecommunications antenna tower must perform well under potentially harsh conditions. They must pass all tests while running within the environment for which they were designed, such as elevated temperatures.
Advantech is the first NVIDIA partner to certify a system in this category, with its USM-501 system, a medical-grade computer for diverse hospital applications. NVIDIA-Certified Systems for the industrial edge will be available soon from other system providers, including Dynics, Lenovo, Mercury Systems and Prodrive Technologies.
For systems deployed in more controlled edge environments, such as retail stores, the enterprise edge category would apply. These would be tested in data-center-like conditions and will be available soon from system providers including GIGABYTE and Lenovo.
NVIDIA-Certified edge systems provide the acceleration needed by intelligent edge applications, such as video analytics and robotics, where total system power is often limited.
The primary workload run on most edge servers is AI inferencing, and NVIDIA-Certified Systems equipped with the NVIDIA A30 Tensor Core GPU deliver leading inference performance for edge. These systems allow AI applications to be deployed with fewer servers and less power, resulting in faster insights with dramatically lower costs.
This was conclusively demonstrated by the latest results of the MLPerf Benchmark, in which NVIDIA-Certified Systems set records for both performance and energy efficiency.
With its Multi-Instance GPU, or MIG, technology, the A30 also allows up to four workloads to be run simultaneously with their own guaranteed quality of service. This helps maximize mainstream compute utilization and efficiency.
The recently announced NVIDIA A2 Tensor Core GPU delivers entry-level, low-power, compact acceleration for edge AI and inference, and offers the smallest footprint of the NVIDIA enterprise GPU portfolio. Systems with this GPU are expected to become NVIDIA-Certified in the near future.
Build and Deploy Edge AI Models
NVIDIA has developed a whole set of software for managing NVIDIA-Certified Systems for the edge, as well as frameworks for developing and supporting edge applications.
The NVIDIA Metropolis video analytics framework gives organizations the tools needed to build complete, unique computer vision solutions without having to start from scratch. Metropolis provides pretrained models and transfer learning, as well as a full software development kit to help developers get started. Its rich ecosystem of computer vision partners provides organizations with Metropolis-certified applications that are ready to be deployed in edge environments with little to no configuration needed.
For deploying applications at the edge, organizations can rely on NVIDIA Fleet Command, a cloud service that securely deploys, manages and scales AI applications across distributed edge infrastructure. Purpose-built for AI, Fleet Command offers streamlined deployments, layered security and detailed monitoring so organizations can go from zero to AI in minutes.
A Comprehensive Certified Portfolio
The NVIDIA-Certified Systems portfolio also now includes a category for high-density virtualization, which will feature systems optimized for virtual desktop and virtual workstation using the NVIDIA A16 GPU. With the addition of edge systems, enterprises have a full range of choices when it comes to hardware-accelerated infrastructure.
Whatever the use case, customers can confidently choose to power their accelerated workloads.
Discover more about NVIDIA-Certified Systems:
- Whitepaper: Accelerate Compute-Intensive Workloads With NVIDIA-Certified Systems
- Find an NVIDIA-Certified System
- NVIDIA-Certified Systems Configuration Guide
To learn more about NVIDIA edge computing solutions, check out Deploying and Accelerating AI at the Edge With the NVIDIA EGX Platform.
Watch NVIDIA CEO Jensen Huang’s GTC keynote: