NVIDIA Announces Eos, the World’s fastest AI Supercomputer

NVIDIA is back with another edition of the GPU Technology Conference, the NVIDIA GTC in March 2022. Data Scientists, Machine Learning Engineers, Researchers all around the world are curious to know the most recent hardware and software advances in Machine Learning and Data Science from NVIDIA. During GTC, NVIDIA’s CEO Jensen Huang announced about NVIDIA’s plans to create Eos, the world’s fastest AI Super Computer.

Image of NVIDIA DGX pod which will be making up NVIDIA Eos
Image of a NVIDIA DGX pod taken from NVIDIA. Source

Eos will be the first NVIDIA AI Factory based on the Hopper Architecture of GPU’s. Eos is expected to begin operations later this year and will have 18.4 Exaflops of AI computing performance which will make it the world’s fastest AI Super Computer in the world. NVIDIA Eos will have more than 4 times faster AI processing as compared to Fugaku Supercomputer in Japan, the current fastes AI Super-Computer in the world.

NVIDIA Eos will also be a behemoth in traditional scientific computing with an expected 275 petaflops of performance. This will make it 1.4 times faster than the fastest science computer in the United States.

Eos will be consist of 18DGX Pods, each of which consist of 32 DGX’s, with each DGX being powered by 8 H100 GPU’s. The H100 GPU by NVIDIA is the latest GPU by NVIDIA based on the Hopper architecture and is being regarded by NVIDIA as the most significant leap ever made in GPU’s.

Eos will be used NVIDIA researchers for advancing climate science, digital biology and the future of AI.

The NVIDIA Press Release for Eos can be download from here.


Other Articles you might be Interested In

5 NVIDIA GTC 2022 Sessions to attend for Beginners in Machine Learning

Private Investment in Artificial Intelligence soars to 93.5 Billion Dollars in 2021, more than double from Previous Year

Trends in Artificial Intelligence in 2021

Ministry of Electronics and Information Technology launches TIDE 2.0 Challenge Grant for Emerging Technologies

Qiskit opens applications for Quantum Error Correction Summer School 2022

ConvNext: The ConvNet for the Roaring 20s