Deploy GPU Accelerated at the Edge with NVIDIA Validated NGC-Ready for Edge Servers and NVIDIA EGX Edge Computing Platforms
Back in May we talked about edge computing and why edge computing should be on the minds of all companies. In that blog, we touched on the inclusion of AI and placing such high levels of computation closer to sensors and AI applications. With the rise of 5G, AI will be erupting at the edge. With this distributed approach, opportunities are emerging for AI applications, whether they are related to IoT, streaming sensor data, or cloud-native AI applications at the edge.
QCT grew with the advent of cloud computing that drove hyperscale data centers to run commercial off-the-shelf hardware years ago. With AI at the edge, we are also extending our experiences with CSPs to create scalable, GPU-accelerated platforms that can handle continuous integration and delivery to manage data remotely while providing decisions in real time.
Now, via the alignment with a new edge computing platform – NVIDIA EGX – we believe QCT and NVIDIA can work together to provide a more integrated architecture, from data center to edge, so our customers will enjoy optimized performance no matter where workloads are located.
Support of NVIDIA’s NGC-Ready Program with Edge Added
With NVIDIA’s expansion of its server certification program to include a new designation, NGC-Ready for Edge, which identifies systems capable of running today’s most demanding AI workloads at the edge, AI becomes easier to deploy. Because the NVIDIA driver, NVIDIA container runtime, Kubernetes NVIDIA device plug-in, and monitoring plug-in can all be managed with the cloud-native NVIDIA GPU Operator. It can automate the setup of a GPU-powered node with rapid installation via a Helm chart running on these QCT Systems. QCT will be validating additional server models with the latest NGC-Ready guideline, to be NGC-Ready for Edge, which is an advancement of QCT’s existing NGC-Ready product portfolio.
QCT NGC-Ready Computing Platforms
Each of these models can support NVIDIA T4 or V100 Tensor Core GPUs, and with validation as NGC-Ready, they will eliminate the time-consuming and cumbersome process of choosing a system, as customers can quickly identify that these are systems capable of running AI workloads whether in their data centers, in the cloud, or at the edge.
QCT currently offers the following NGC-Ready platforms*.
- QuantaGrid D52BV-2U
- QuantaGrid SD2H-1U
- QuantaGrid D43K-1U
- QuantaGrid S43KL-1U
*other platforms will be released through the end of 2019
And as we mentioned in May, the implications are endless, as every industry will have the opportunity to deliver automated intelligence be it retail, manufacturing, healthcare, or smart cities. 5G will be an additional driver for this roll out, and telco services and various retail and manufacturing use cases are already being developed (e.g., data analytics, video surveillance for safety, inspection, customer service, etc.). As a result, all verticals can take advantage of this cloud-native container deployment which enables moving AI workloads closer to where it is needed so that the goal to attain data with lower latency and in real-time can be achieved.