Nvidia GTC Fall 2021 with full-stack computing and new announcements
Another event hosted by Nvidia took place in the Nvidia GTC Fall 2021, where the company made big announcements unveiling improved AI, security, graphics processors and more. NVidia is a big name in technology as they make world-class graphics processors for large businesses and the consumer market. With crypto prices rising through the roof, mining for these cryptos is happening on a large scale. Hence the cost of their consumer-grade GPUs is out of reach for most of the customers. Although they implemented mining-block technology, it was lifted by miners, so it was of no use. Still, there is a crisis of graphics chips in the market, but the GTC Fall 2021 announcements came in within these.
The GTC is a large event where the company unveils their future technology, and this year was no different. It’s great to see companies are coming back in full swing as we are gradually defeating the pandemic. From lower interface GPUs to tiny semiconductors and high-speed networking platforms, all were unveiled in the event. As usual, the big news comes from Jensen Huang, CEO of Nvidia. Nvidia poses itself as a full-stack computing company that manufactures advanced silicon and software solutions. Huang said, “As we accelerate more applications, our network of partners experiences growing demand for Nvidia platforms.”
According to Huang, Nvidia now has over 150 software development kits that serve a wide array of development, including 3D workflow, 5G, quantum computing, robotics, AI, and many more. Being a chipmaker first, the priority is producing enough GPUs to sustain the market and feed increasing demand. Today, a GPU is more than essential even to non-gamer because intensive works become relatively efficient if a good graphics chip is present. Nvidia finally launched their next-generation A2 Tensor GPU for a low-power interface.
New generation (A2) GPUs has one low slot profile PCIe 4.0 with 10 Ray Tracing (RT) cores and 16 GB of GDDR6 memory. FP32 performance of these GPUs is 4.5 teraflops with Ampere’s sparsity feature enabled. That’s not all the tech giant had to offer us; Quantum-2 InfiniBand networking platform will serve as a 400Gbps networking provider for “extreme preference, board accessibility and a strong security.” It will provide double the speed than the previous generation and three times accelerated performance. It is much required for data centres as we are in the age of data, and companies handling their needs have all the power possible to serve a vast customer base.
Dell Technologies, Hewlett Packard Enterprise, IBM, Lenovo, Gigabyte, Supermicro will produce them under Nvidia at a larger scale. NVidia’s Omniverse Enterprise got a major revamp with newer paid software’s for premium work ethic. This tech allows 3D design teams to work together with much fluidity. Omniverse has three subscriptions. The Nucleus servers cost $1,000 annually, the second one is the creator costing $2,000 per floating license, and lastly, reviewers package with a $100 price tag. All the packages include Omniverse View and an enterprise support system.
Server-side infrastructure is something most users are not aware of, but the Triton Interface Server also got an update to serve its 25,000 customers. It helps run interface applications at a larger scale. The company is not popular for aggressive marketing for these underdog products, but they certainly are of the quality that matches competitors.
Security is a major concern of the era; NVidia also has a footprint on the section. The company launched a Zero-Trust Cybersecurity platform for developers. Dev’s can now use the Zero-Trust server solution, which is 600 times faster than those who do not use Nvidia Acceleration. Bluefield data processing and DOCA software development kit workloads together to improve security measures that even helps AI framework to stay secure from intruders.
There will be newer products coming out of Nvidia, which we are aware of. Still, the new Modules framework caught our attention as it further helps machine learning systems to understand physics for digital twin applications. Although we are in the waiting seat, these innovations surely are intriguing.