eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
A couple of weeks ago at the Mobile World Congress event in Los Angeles, graphics processing unit market leader NVIDIA unveiled its EGX Edge Supercomputing platform. The product is designed for all edge applications including, but not limited to, artificial intelligence and internet of things. This was the first time NVIDIA has been part of a Mobile World Congress, and the announcement of the EGX certainly warranted its presence at the event.
EGX Brings GPU Computing to the Edge
As is the case with most NVIDIA platforms, EGX is a reference design that other server manufacturers can use to roll out a validated design for edge computing use cases that include NVIDIA GPUs. At the time of announcement, Cisco, Fujitsu, Lenovo, HPE and others were confirmed partners. This is a beast of an edge server because it is powered by NVIDIA’s CUDA Tensor Core GPU, has cryptographic acceleration for encryption, networking for Mellanox, NVMe over TCP and RDMA for storage, an industrial-strength cloud native stack and hardened AI software.
Go here for more eWEEK articles on software-defined networking.
Why so beefy? Well, with all due respect to the cloud vendors, the edge is going to be where the action is in the future. Computing is moving to where the data resides, and this will take power.
It makes sense that NVIDIA would announce this at MWC as 5G is one of the key enablers of edge computing. 5G brings multi-gigabit speeds to wireless and enables more data to be processed in more places. One school of thought is that all data can be processed in the cloud, but that doesn’t work for all workloads.
The ultimate example of this is a self-driving car. The decision about whether to stop or not stop, for example, should be done in the car itself, not by sending the data all the way to the cloud and back. Another use case is facial recognition at an airport. The ability to scan a face and determine whether the person is a threat or not needs to be done in real time. Even a delay of a few minutes could mean the difference between catching a bad guy or having him go free in the airport.
IoT drives massive volumes of data to the edge
Another factor driving edge computing is the rise of the IoT. We live in a world were literally everything is connected, and that is creating massive amounts of data to be analyzed. Most traditional compute platforms don’t have the horsepower for AI on large data sets, hence the requirement for NVIDA to go built its own optimized reference architecture.
The NVIDIA Edge Stack software is optimized for real-time AI uses cases, such as video analytics, video, speech and audio as well as Red Hat OpenShift for Kubernetes container orchestration. EGX also has support for NVIDA Metropolis, which is an IoT-optimized ecosystem for smart cities, smart factories, traffic engineering, retail and other places where large amounts of sensor data are being pulled in and analyzed.
NVIDIA Aerial virtualizes the radio access network
In conjunction with the launch of EGX, NVIDA also announced its new Aerial software developer kit, which enables mobile operators to move to a cloud-native architecture. Mobile operators are suffering the same fate as traditional telcos, which is that their infrastructure is so rigid that cloud companies and newer service providers will eat them alive and digest them before they can even react.
Aerial enables them to build software defined wireless radio access networks (RANs) that deliver on two critical capabilities: low-latency data transmission between Mellanox network cards to GPU memory and a 5G physical layer signal-processing engine that keeps all data within the confines of the GPU’s high performance memory.
During his keynote, CEO Jensen Huang discussed how Aerial allows mobile operators to deploy completely virtualized 5G RANs that open the door to a number of new adjacent markets including cloud gaming, augmented reality, IoT and virtual reality. Ariel running on EGX brings accelerated computing benefits to service provider hardened Kubernetes infrastructure running at the edge.
The edge is where the emerging action is today and where the mainstream action will be soon. The cloud certainly isn’t going away, but more and more businesses will deploy edge computing nodes to analyze more data in more places. The GPU based EGX is ideally suited for many of the advanced edge use cases.
Zeus Kerravala is an eWEEK regular contributor and the founder and principal analyst with ZK Research. He spent 10 years at Yankee Group and prior to that held a number of corporate IT positions.