NVIDIA to Build Earth-2 Supercomputer to See Our Future
The Earth is heating up. The last seven years are set to be the warmest on record. Greenhouse gas emissions from human activity are causing average warming of about 1.1 degrees Celsius since the 1850-1900 period.
What we are experiencing today differs greatly from the historical norm. We are seeing extreme weather: record droughts, unprecedented heat waves, strong hurricanes, storms and catastrophic floods. Climate disasters have become the new normal.
“We need to fight climate change right now. Even so, we won't feel the returns of our efforts for decades. It is hard to mobilize action for something in the future. But we must know our future today, see it and feel it in order to act urgently”
To make our future a reality today, we should turn to simulation. Developing the best mitigation and adaptation strategies requires climate models that can predict climate in various regions of the globe decades ahead.
Unlike weather forecasting, which primarily models atmospheric physics, climate models are multi-decade simulations that cover the physics, chemistry and biology of the atmosphere, water, ice, land and human activity. Climate simulations today are run at resolutions of 10 to 100 km.
However, higher resolution is needed to model changes in the global water cycle, the movement of water from the ocean, sea ice, land surface and groundwater through the atmosphere and clouds. Changes in this system lead to stronger storms and droughts.
To simulate clouds that reflect sunlight back into space, meter-scale resolution is required. Scientists estimate these resolutions will require from millions to billions of times more computing power than is available today. It would take decades to achieve this with the usual pace of computer advances, which speeds up tenfold every five years.
“For the first time, we have the technology to model ultra-high-resolution climate at the speed of light and predict changes in extreme weather in a region decades ahead. We can achieve a million-fold speedup by combining three technologies: GPU-accelerated computing, deep learning and breakthroughs in physics-informed neural networks, and AI supercomputers, together with the huge amount of observed and model data to learn from”
With ultra-high-resolution methods, humanity may be able to make the billion-fold leap needed for ultra-high-resolution climate modeling. Countries, cities and towns will be able to get early warnings to adapt and strengthen infrastructure resilience. And with more accurate forecasts, people and countries will be able to act more quickly.
Huang and the company's scientists plan to dedicate themselves and significant resources to apply NVIDIA's scale and expertise in computational science to join the global climate science community. The company announced plans to build the world's most powerful AI supercomputer dedicated to predicting climate change. The system, named Earth-2 or E-2, will create a digital twin of Earth in Omniverse.
This system will be the counterpart of Cambridge-1, the world's most powerful AI supercomputer for healthcare research. Earlier in 2021, NVIDIA unveiled Cambridge-1 in the UK, and it is being used by several leading medical companies.
“Every technology we have invented to date is needed to make Earth-2 possible. I can't imagine a greater or more important use”
The company called Cambridge-1 the first big bet on the digital biology revolution. The UK's most powerful supercomputer will advance research at AstraZeneca, GSK, King's College London and Oxford Nanopore.
Presenting the supercomputer, Huang said: "We are at the beginning of the digital biology revolution, and scientists need a powerful rocket to do this work. Cambridge-1 brings together decades of our work in accelerated computing, artificial intelligence and life sciences in one computing center."
NVIDIA's partners in the UK will be able to tap the technologies behind Cambridge-1 with incredible speed. It took just 20 weeks to build, instead of the two years usually needed to create most supercomputers of this scale.
NVIDIA said that amid the global pandemic, combining advances in genomics and computing is more important than ever. The coronavirus pandemic has made accelerating healthcare a top social and economic imperative.
Cambridge-1 gives the company a platform to work with world-class British universities, biomedical companies, healthcare providers and more than a thousand startups.
