On October 5th, NDT held its annual 2020 GTC (GPU Technology Conference). Huang Renxun once again delivered his opening speech in his own kitchen, which was divided into four parts: data center, edge artificial intelligence, collaboration tools, and healthcare. Core highlights include:
Data Center: Launch of Ntveda Blue Field-2 DPU and Ntveda Blue Field-2X DPU with all the features of Ntveda Mellanox;
Edge AI: Announced the launch of the EDX AI platform for organizations of all sizes to deploy AI quickly and efficiently at scale, and the Launch of the Jetson Nano 2GB development kit, which ranges from entry-level AI devices to advanced platforms.
Collaboration Tools: Open source beta version of Omniverse platform, streaming video AI platform Iveda Maxine, supports real-time collaboration among millions of designers.
Healthcare: In partnership with global healthcare company GSK and its AI group, supercomputers are being built for use by british medical researchers.
Next-generation DPU accelerates data centers.
‘The data center is defined by software, which makes the data center more flexible and also places a huge burden, and the data center infrastructure can run at 20 to 30 percent CPU core, so a new processor, the Data Processing Unit, is needed, ‘ Mr. Huang said.
What exactly does the DPU do? According to in Vedda’s official interpretation, the DPU is an SOC that combines three key elements:
Industry-standard, highest-performance, and software-programmable multi-core CPUs, usually based on a wide range of Arm architectures, work closely with their SOC components;
A high-performance network interface that parses, processes, and efficiently transmits data to GPUs and CPUs at wire speeds or speeds available in the network.
A variety of flexible and programmable acceleration engines that uninstall applications such as AI, machine learning, security, telecommunications, and storage, and improve performance.
These capabilities of the DPU are critical to enabling large-scale computing on the secure next-generation cloud.
Based on the latest research from Mellanox, acquired by Nweida in April, Nedda has launched two DPU products, the AfD BlueField-2 DPU and the BlueField-2X DPU.
According to NV, the BlueField-2 DPU has all the features of NV Mellanox Connext-6 SmartNIC, is connected to a powerful Arm core, and is fully programmable, providing data transfer rates of 200 gigabits per second, accelerating security, networking, and storage tasks in critical data centers.
The BlueField-2X DPU has all the key features including blueField-2 DPU, which can be enhanced by the AI capabilities of the Nweida Ampere GPU.
To enable developers to build applications on DPU-accelerated data center infrastructure services, INDS also launched a DOCA Software Development Kit, a GPU-accelerated application that is similar to the AUS CUDA programming model.
It’s worth noting that DOCA provides developers with a comprehensive, open platform, fully integrated into a software catalog, providing a convenient software environment for third-party application providers, using DPU to accelerate data centers, and developing, certifying, and distributing applications to customers.
Huang Renxun announced that a new BlueField-2 will be added to Nive Field-2 for Niveda AI computing.
In addition, Huang Renxun also announced the development roadmap of the Niveda DPU, which is expected to launch the BlueField-4 with 400 TOPS and 400Gbps in 2023, which means that in just a few years, Niveda DPU’s computing throughput will exceed nearly 1000 times.
The EGXAI platform deploys artificial intelligence, a democratized Jetson AI robot.
Huang Renxun believes that one day tens of thousands of artificial intelligence devices and robots will be spread across homes, office buildings, shops and other places across the globe, “to support this massive transformation, Avanta has created an accelerated computing platform to help companies modernize data centers and deploy artificial intelligence anywhere.” “
The accelerated computing platform referred to here is the IND EGX artificial intelligence platform, the platform’s system can be obtained through Dell, Lenovo, Wave, Ultra micro and other leading server manufacturers.
Nedda also announced that the EGX platform is expanding the capabilities of the combined NDUs and BlueField-2 DPU on a single PCIe card, providing enterprises with a common platform to build secure, accelerated data centers.
To deploy and manage EGX servers, IVEDA has launched a new service called Indida Fleet Command that combines the security and real-time processing power of edge computing with the ease of use of remote management and software services to centrally manage servers across a wide area, including factories, hospitals, retail stores and even city streets, through a unified control plane. It is reported that THE Group, the leader in global supply chain solutions, is using the NVDA EGX AI platform to develop AI applications for its smart warehouse systems to improve the throughput and efficiency of retail distribution centers.
Last year, InDida launched a $99 Jetson AI computer, and this year expanded jetson AI’s entry-level development kit on the Edge platform for just $59.
As a new addition to the Jetson family, the Jetson Nano 2GB developer suite is designed for people who teach and learn artificial intelligence, ranging from entry-level AI devices to advanced platforms, and, in Huang’s words, Jetson AI robots are democratized robots. To support AI teaching, IDA also announced Jetson’s free online availability, complementing many open source projects, with thousands of developers active in the Jetson community, training and AI certification.
“At Booz Allen, we’re looking to empower people to change the world,” said Drew Farris, director of analytics and artificial intelligence research at Booz Allen Hamilton. We’re using Affida Jetson to train new technology resources, because artificial intelligence is critical for businesses and people, using it to solve the most difficult global challenges. “
Ampere Architecture expands into a professional vision product line, and Omniverse builds metaverses.
In addition to the progress made in data centers and edge AI, AFD also announced the launch of an open beta version of the collaboration tool Omniverse platform and the AI data streaming platform Maxine to build a supercomputer called Cambridge 1 dedicated to AI medical research. In addition, ING has announced an ampere architecture extended to the professional vision product line at GCT fall.
Omniverse is the world’s first 3D simulation and collaboration platform based on Nvida RTX, which combines physical and virtual worlds, simulates reality in real time, and has realistic details. “Omniverse allows designers, artists, creators and even artificial intelligence to use different tools in different worlds, connecting and collaborating in a common world to create a world.” Huang Renxun said.
With Maxine, developers can reduce video bandwidth consumption, reduce vendor costs, and improve video conferencing experiences for end users with a smooth video conferencing experience.
In addition to improving the user experience for regular streaming media, IND announced on GCT Fall that it is working with AWS to release the Aweda CloudXR, which is available on Amazon EC2 P3 and G4. It is reported that the Nveda CloudXR platform includes cloudXR software development kit, Quadro virtual workstation software and AI SDK. With a cloud-streamed experience, professionals can easily set up, extend, and access immersive experiences from anywhere, free from VR environments.
On GCT Fall, Niveda also announced the launch of quadro RTX A6000 and Nedda A40 GPUs for professional designers, with the next generation of GPUs with new RT cores, Tensor cores and CUDA cores, speeding up graphics rendering and AI.
Finally, it is worth noting that in this speech, Huang Renxun also mentioned the relationship between Invida and Arm, announced major initiatives to advance the Arm platform:
First, Avda will complement Arm’s business with GPU, networking, storage, and security technologies to create a complete acceleration platform.
Second, NV is working with Arm partners to create platforms for HPCs, clouds, edges and PCs that require chips, systems and system software.
Third, NV is porting AI and RTX engines to Arm.
“Today, these features are only available on x86, and with this initiative, the Arm platform will also lead the way in acceleration and AI computing.” Huang Renxun said.