NVIDIA has shared some details of the Earth-2 supercomputer that it is planning to build in order to simulate and predict the effects of climate change.
The tech giant’s CEO, Jensen Huang, first announced the project at the end of his keynote speech at the online GPU Technology Conference (GTC) last week.
Huang spoke about a number of new innovations at the conference, but, as the Cop26 climate summit continued in the UK, he said that he “can’t imagine greater and more important news” than Earth-2.
Now, NVIDIA has shared some more details of the plans via the company’s official blog.
Current climate simulation models use resolutions configured between 10km and 100km, but NVIDIA says that we need much higher resolutions in order to make accurate predictions.
The chipmaker says that the potential for ultra-high-resolution climate modelling is now possible due to the advent of three key technologies.
The combination of GPU-accelerated learning, neural networks with deep learning, and AI-powered supercomputers could help achieve million-x speedups (increases in the speed and performance of computing by a factor of a million or more).
Add in huge quantities of training data and super-resolution techniques, and NVIDIA believes that a billion-x leap could be within grasp.
Accurate prediction could allow for better adaption and mitigation
This would allow for the type of ultra-high-resolution simulations that could model the physics, biology and chemistry of atmospheric systems, land, oceans, ice and human activity over the course of multiple decades.
The company argues that being able to model climate changes in different parts of the world over these timescales would allow for the best strategies to adapt to and mitigate these changes.
Earth 2, also known as E-2, is described as a “digital twin” of the Earth and will be built in NVIDIA’s Omniverse – an open platform designed for virtual collaboration and real-time physical simulations.
It aims to be the climate-focused equivalent to Cambridge-1, which the company unveiled earlier this year and claims to be the world’s most powerful AI-fuelled supercomputer used for research in the healthcare sector.
In his speech at GTC, Huang also said that E-2 would be the most energy-efficient supercomputer ever built.
He added that if you thought of the Earth as a physical thing, E-2 would be “the engine of alternate worlds”.
Today’s news was brought to you by TD SYNNEX – the UK’s number one solutions distributor.
