IoT

Following the acquisition of Mellonix, Nvidia is driving data centre changes as they become computing units. In what should have been a keynote at the company’s conference but which was altered to a virtual presentation due to the Covid-19 pandemic, CEO Jensen Huang introduced a GPU architecuture optimised for the new scale of data centre computing.

He announced the Nvidia A100 GPU, based on the Nvidia Ampere architecture, which is claimed to provide “the greatest generational performance leap” of Nvidia’s eight generations of GPUs. It is built for data analytics, scientific computing and cloud graphics, and is in full production and shipping to customers worldwide.

The A100 and the Nvidia Ampere architecture on which it is built, boosts performance by up to 20x over its predecessors. It is claimed to be the world’s largest 7nm processor with over 54bn transistors.

It also features third-generation Tensor Cores with TF32, a new math format that accelerates single-precision artificial intelligence (AI) training, and structural sparsity acceleration, an efficiency technique for AI maths. Another feature is multi-instance GPU (MIG), which allows a single A100 processor to be partitioned into up to seven independent GPUs, each with its own resources.

Also contributing to the increased performance is the third-generation NVLink technology, doubling high-speed connectivity between GPUs. The result is that A100 servers can act as one giant GPU, says Nvidia.

In the same announcement, Nvidia also announced a third generation of its Nvidia DGX AI system based on the A100. The Nvidia DGX A100 is believed to be the world’s first 5PetaFLOPS server. Each DGX A100 can be divided into as many as 56 applications, all running independently.

It allows a single server to either scale up to race through computationally intensive tasks such as AI training, or scale out, for AI deployment, or inference, Huang said.

The A100 will also be available for cloud and partner server makers as HGX A100.

A data centre powered by five DGX A100 systems for AI training and inference running on just 28kW can do the work of a typical data centre with 50 DGX-1 systems for AI training and 600 CPU systems consuming 630kW, Huang explained.

Another announcement was the next-generation DGX SuperPOD, powered by 140 DGX A100 systems and Mellanox networking technology. It offers 700PetaFLOPS of AI performance, the equivalent of one of the 20 fastest computers in the world.

The next-generation DGX SuperPOD delivers 700PetaFLOPS of AI performance.

http://www.nvidia.com

Smart manufacturing specialist, Foxconn, video and imaging SoC company, Socionext and artificial intelligence (AI) chip manufacturer, Hailo, have partnered to introduce Boxiedge AI edge computing for energy-efficient, standalone, AI inference nodes.

Foxconn’s high-density, fanless Boxiedge edge computer is combined with Socionext’s SynQuacer SC2A11, high-efficiency parallel processor and the Hailo-8 deep learning processor. According to the trio, the combination provides market-leading energy efficiency for standalone AI inference nodes, in smart cities, smart medical, smart retail applications and the industrial IoT.

It is designed to address the need for cost-effective multi-processing capabilities required in video analytics, image classifications and object segmentation in these compute-intensive applications. The robust, Boxiedge AI edge computing product is capable of processing and analysing over 20 streaming camera input feeds in real-time, at the edge. The result is a high-density, low-power, local video management system (VMS) server, ensuring top performance for video analytics and privacy, including image classification, detection, pose estimation, and various other AI-powered applications – all in real time.

Hailo’s specialised Hailo-8 deep learning processor features up to 26 Tera operations per second (TOPS). Its architecture enables edge devices to run sophisticated deep learning applications that could previously only run on the cloud. Its structure translates into higher performance, lower power and minimal latency, enabling enhanced privacy and better reliability for smart devices operating at the edge, said the company.

Gene Liu, VP of semiconductor sub group at Foxconn Technology, said: “We recognise the great potential in adopting AI solutions for a multitude of applications, such as tumor detection and robotic navigation. . . our edge computing solution combined with Hailo’s deep learning processor will create even better energy efficiency for standalone AI inference nodes to positively impact rapidly evolving [market] sectors.”

“We are very pleased with this joint effort by the companies, and to officially announce our strategic partnership with Hailo,” said Noriaki Kubo, executive VP at Socionext. “This collaboration will lead to more innovative solutions that specifically address the growing demand from our AI customers in multiple sectors. We are confident that this product will enable endpoint devices to operate with better performance, lower power, more flexibility, and minimal latency.”

The next generation of the Boxiedge AI computing is equipped with applications for a broader market relying on low latency, a high data rate, high reliability and quick processing at the edge. Smart retail and smart cities, for instance, require hundreds of cameras – either in-store or in traffic monitoring – to generate video streams that need to be processed locally, quickly, and efficiently with minimal latency.

Similarly, for industrial IoT, data acquiring, processing, inferencing, and presenting on the production floor rather than in the cloud translates into significant cost savings along with more efficient processing for tasks such as inspection and quality assurance.

http://www.socionext.com