DGX H100 Programs Now Transport Worldwide


Prospects from Japan to Ecuador and Sweden are utilizing NVIDIA DGX H100 programs like AI factories to fabricate intelligence.

They’re creating companies that provide AI-driven insights in finance, healthcare, regulation, IT and telecom — and dealing to remodel their industries within the course of.

Among the many dozens of use circumstances, one goals to foretell how manufacturing unit tools will age, so tomorrow’s vegetation might be extra environment friendly.

Known as Inexperienced Physics AI, it provides data like an object’s CO2 footprint, age and power consumption to SORDI.ai, which claims to be the most important artificial dataset in manufacturing.

Green Physics AI demo accelerated by DGX H100
Inexperienced Physics AI lets customers mannequin how objects age.

The dataset lets producers develop highly effective AI fashions and create digital twins that optimize the effectivity of factories and warehouses.  With Inexperienced Physics AI, additionally they can optimize power and CO2 financial savings for the manufacturing unit’s merchandise and the parts that go into them.

Meet Your Good Valet

Think about a robotic that might watch you wash dishes or change the oil in your automotive, then do it for you.

Boston Dynamics AI Institute (The AI Institute), a analysis group which traces its roots to Boston Dynamics, the well-known pioneer in robotics, will use a DGX H100 to pursue that imaginative and prescient. Researchers think about dexterous cellular robots serving to individuals in factories, warehouses, catastrophe websites and ultimately houses.

“One factor I’ve dreamed about since I used to be in grad college is a robotic valet who can observe me and do helpful duties — everybody ought to have one,” mentioned Al Rizzi, CTO of The AI Institute.

That can require breakthroughs in AI and robotics, one thing Rizzi has seen firsthand. As chief scientist at Boston Dynamics, he helped create robots like Spot, a quadruped that may navigate stairs and even open doorways for itself.

Initially, the DGX H100 will deal with duties in reinforcement studying, a key method in robotics. Later, it can run AI inference jobs whereas related on to prototype bots within the lab.

“It’s a particularly high-performance pc in a comparatively compact footprint, so it gives a straightforward means for us to develop and deploy AI fashions,” mentioned Rizzi.

Born to Run Gen AI

You don’t must be a world-class analysis outfit or Fortune 500 firm to make use of a DGX H100. Startups are unboxing a number of the first programs to journey the wave of generative AI.

For instance, Scissero, with workplaces in London and New York, employs a GPT-powered chatbot to make authorized processes extra environment friendly. Its Scissero GPT can draft authorized paperwork, generate experiences and conduct authorized analysis.

In Germany, DeepL will use a number of DGX H100 programs to broaden companies like translation between dozens of languages it gives for purchasers, together with Nikkei, Japan’s largest publishing firm. DeepL not too long ago launched an AI writing assistant known as DeepL Write.

Right here’s to Your Well being

Lots of the DGX H100 programs will advance healthcare and enhance affected person outcomes.

In Tokyo, DGX H100s will run simulations and AI to hurry the drug discovery course of as a part of the Tokyo-1 supercomputer. Xeureka — a startup launched in November 2021 by Mitsui & Co. Ltd., one in every of Japan’s largest conglomerates —  will handle the system.

Individually, hospitals and educational healthcare organizations in Germany, Israel and the U.S. might be among the many first customers of DGX H100 programs.

Lighting Up Across the Globe

Universities from Singapore to Sweden are plugging in DGX H100 programs for analysis throughout a spread of fields.

A DGX H100 will practice massive language fashions for Johns Hopkins College Utilized Physics Laboratory. The KTH Royal Institute of Sweden will use one to broaden its supercomputing capabilities.

Amongst different use circumstances, Japan’s CyberAgent, an web companies firm, is creating sensible digital advertisements and movie star avatars. Telconet, a number one telecommunications supplier in Ecuador, is constructing clever video analytics for secure cities and language companies to help prospects throughout Spanish dialects.

An Engine of AI Innovation

Every NVIDIA H100 Tensor Core GPU in a DGX H100 system gives on common about 6x extra efficiency than prior GPUs. A DGX H100 packs eight of them, every with a Transformer Engine designed to speed up generative AI fashions.

The eight H100 GPUs join over NVIDIA NVLink to create one large GPU. Scaling doesn’t cease there: organizations can join lots of of DGX H100 nodes into an AI supercomputer utilizing the 400 Gbps ultra-low latency NVIDIA Quantum InfiniBand, twice the velocity of prior networks.

Fueled by a Full Software program Stack

DGX H100 programs run on NVIDIA Base Command, a collection for accelerating compute, storage, and community infrastructure and optimizing AI workloads.

In addition they embrace NVIDIA AI Enterprise, software program to speed up information science pipelines and streamline improvement and deployment of generative AI, pc imaginative and prescient and extra.

The DGX platform affords each excessive efficiency and effectivity. DGX H100 delivers a 2x enchancment in kilowatts per petaflop over the DGX A100 era.

NVIDIA DGX H100 programs, DGX PODs and DGX SuperPODs can be found from NVIDIA’s world companions.

Manuvir Das, NVIDIA’s vp of enterprise computing, introduced DGX H100 programs are transport in a chat at MIT Know-how Assessment’s Future Compute occasion as we speak. A hyperlink to his discuss might be out there right here quickly.


Please enter your comment!
Please enter your name here