In an effort to help bring artificial intelligence to edge devices, Nvidia has announced its new Nvidia EGX platform that has the ability to understand and act on data in real time without having to send it to the cloud or data center first.
Up until now, AI algorithms were so complex that they required powerful machines to run them in cloud data centers. However, this required a great deal of bandwidth and resulted in higher latency which is why Nvidia and other hardware manufacturers have been trying to bring AI to the edge of the network instead.
Ideally AI should be run where data is created whether that be in retail stores, manufacturing facilities or warehouses.
- Nvidia closes in on Mellanox deal in data center push
- Michael Dell: Why edge computing could be the next big thing
- Companies who aren't investing in AI may risk falling behind
Nvidia's entrance to the AI edge market makes a great deal of sense as its graphics processing units are known for how well they can handle AI with the Tesla V100 being used for deep learning and the Quadro GV100 creating realistic images in real time using ray tracing.
Nvidia EGX
One of the best aspects of Nvidia's new EGX platform is the fact that it is scalable and can be used in a number of configurations.
For instance, EGX starts with the tiny Nvidia Jetson Nano which can provide one-half trillion operations per second (TOPS) with just a few watts for tasks such as image recognition and spans all the way to a rack of Nvidia T4 servers to deliver over 10,000 TOPS for real-time speech recognition and other real-time AI tasks.
Vice president and general manager of Enterprise and Edge Computing at Nvidia, Bob Pette explained how the EGX platform was designed to meet the needs of enterprise customers in a press release, saying:
“Enterprises demand more powerful computing at the edge to process their oceans of raw data — streaming in from countless interactions with customers and facilities — to make rapid, AI-enhanced decisions that can drive their business. A scalable platform like NVIDIA EGX allows them to easily deploy systems to meet their needs on premises, in the cloud or both.”
The platform also integrates security, storage and networking technologies from Mellanox which Nvidia plans to acquire by the end of the year.
Nvidia is working with 13 different server manufacturers to sell its EGX platform including Cisco, Dell-EMC, HPE, Lenovo and others. EGX is also compatible with AI applications running on AWS and Microsoft Azure and the platform can even connect to IoT services from both providers.
Via Silicon Angle
from TechRadar - All the latest technology news http://bit.ly/2JJfhEU
No comments:
Post a Comment