80 billion transistors on a chip, each chip provides 4,000 teraflops of computing power: Introducing the latest release from NVIDIA

Profile photo of the author

This week, NVIDIA began shipping its DGX H100 systems to customers around the world to garner the incredible amount of computing power needed for AI workloads. Announced last year, the DGX system includes 8 Hopper H100 data center chips. “And this chip is blazingly fast. Featuring an all-new GPU architecture built on a 4-nanometer process with as many as 80 billion transistors, each chip can deliver up to 4,000 teraflops of computing power,” writes Sander Hofman in his weekly newsletter.

ASwhat is a teraflop?

Sander Hofman explains: A teraflop is the ability of a chip to compute one trillion floating point operations per second. It is also often abbreviated to TFLOP or TFLOPS. Because of the sheer amount of computing power, teraflops are often used specifically for supercomputers, where you’ll also find petaflops (1,000 teraflops) and exaflops (1,000,000 teraflops, for exascale computing). An electronic device with 1 teraflop or TFLOP means that the chip inside can handle an average of 1 trillion floating point calculations every second. To give you an idea of ​​what that means in practice, Apple’s entry-level M2 chip offers 3.6 teraflops. The PlayStation 5 has over 10 teraflops.
And so NVIDIA’s Hopper H100 can do a whopping 4,000 teraflops! This makes it an amazing data-cruncher. As you can imagine, this chip is intended to offer industrial computing power for Artificial Intelligence, Machine Learning, Deep Neural Networking, and other High-Performance Computing (HPC) related workloads.

Revolutionize industries

NVIDIA’s DGX H100 systems are now shipping worldwide, bringing advanced artificial intelligence (AI) capabilities to various industries. Customers around the world are using these AI supercomputers to transform industries such as finance, healthcare, law, IT and telecommunications. Green Physics AI predicts aging factory equipment for greater efficiency, while Boston Dynamics AI Institute uses the DGX H100 to develop dexterous mobile robots. Startups like Scissero and DeepL harness the power of generative AI through DGX H100 systems for legal processes and translation services. Healthcare organizations, academia, and various industries are leveraging these systems to accelerate research, optimize data science pipelines, and create innovative AI-powered solutions. The DGX H100 boasts eight NVIDIA H100 Tensor Core GPUs, NVIDIA NVLink, and NVIDIA Quantum InfiniBand ultra-low latency 400Gbps, making it a powerhouse for AI innovation.

Green Physics AI: predict the aging of factory equipment

Green Physics AI aims to improve the efficiency of factories of the future by predicting the aging of factory equipment. Manufacturers can develop powerful AI models and create digital twins by adding information such as an object’s carbon footprint, age and energy consumption to SORDI.ai, the largest synthetic dataset in manufacturing. These digital twins optimize factory and warehouse efficiency and energy and CO2 savings for factory products and their components.

Boston Dynamics AI Institute: Development of dexterous mobile robots

The AI ​​Institute, a research organization rooted in Boston Dynamics, is using DGX H100 systems to develop capable mobile robots that can perform useful tasks in factories, warehouses, disaster sites, and eventually homes. Al Rizzi, CTO of The AI ​​Institute, imagines a robot waiter to follow people and perform tasks for them. The DGX H100 will initially tackle reinforcement learning tasks, a key technique in robotics, before performing AI inference work while directly connected to prototype robots in the lab.

Start-ups riding the generative wave of AI

Startups are using DGX H100 systems to explore the potential of generative AI[1]. Scissero, a legal tech startup, uses a GPT-based chatbot to streamline legal processes by drafting legal documents, generating reports, and conducting legal research. DeepL, a language translation company, uses several DGX H100 systems to expand its services, offering translations between dozens of languages ​​for clients such as Nikkei, Japan’s largest publishing house. DeepL has also released an AI writing assistant called DeepL Write.

Improving healthcare and patient outcomes

Many DGX H100 systems are used to advance healthcare and improve patient outcomes. In Tokyo, DGX H100s are running simulations and artificial intelligence to accelerate the drug discovery process as part of the Tokyo-1 supercomputer project. Xeureka, a startup of Mitsui & Co. Ltd., runs the system. Hospitals and academic healthcare organizations in Germany, Israel and the United States are among the first users of the DGX H100 systems.

Universities and research institutes embrace DGX H100

Universities around the world are adopting DGX H100 systems for research in various fields. The Johns Hopkins University Applied Physics Laboratory will use a DGX H100 to train large language models, while Sweden’s KTH Royal Institute of Technology will supply state-of-the-art computer science programs for higher education using the system. Other use cases include Japan’s CyberAgent which builds smart digital ads and celebrity avatars and Telconet, a leading telecommunications provider in Ecuador, which builds smart video analytics for safe cities and language services to support customers across Spanish dialects.

An AI innovation engine

Each NVIDIA H100 Tensor Core GPU in a DGX H100 system delivers approximately six times the performance of previous GPUs. The eight H100 GPUs connect via NVIDIA NVLink to create one giant GPU. Organizations can connect hundreds of DGX H100 nodes to an AI supercomputer using ultra-low latency 400 Gbps NVIDIA Quantum InfiniBand, twice the speed of previous networks.

#billion #transistors #chip #chip #teraflops #computing #power #Introducing #latest #release #NVIDIA

Leave a Comment