We need a 20MW 20,000-GPU-strong machine-learning supercomputer to build EU's planned digital twin of Earth
- Reference: 1614324459
- News link: https://www.theregister.co.uk/2021/02/26/eu_climate_supercomputer/
- Source link:
Starting mid-2021, the boffins will embark upon a seven-to-ten-year mission to create and deploy [1]Destination Earth , or DestinE for short, which will be part of a €1tn (£868bn, $1.2tn) investment in green technologies by the European Union.
[2]
And it is at the core of DestinE, we're told, you'll eventually find that GPU-stuffed supercomputer: a federated system capable of running artificial intelligence, data analytics, and other applications. Crucially, this super will bring together so-called digital twins of Earth, which are numerical models of our home world that simulate and forecast the weather and climate, ocean currents and polar caps, food and water supplies, the effect of humans on the environment, and so on.
These aim is to help scientists, politicians, and the public understand the role nature and humans will play in shaping the planet's future, and help the EU reach its goal of becoming carbon neutral by 2050 through policy decisions. By 2025, the team hopes to have four or five digital twins running, and by 2030, "a full digital twin of the Earth through a convergence of the digital twins already offered."
[3]
I can run through the data in my digital twin and check whether the dike will in all likelihood still protect against expected extreme events
"If you are planning a two-metre high dike in The Netherlands, for example, I can run through the data in my digital twin and check whether the dike will in all likelihood still protect against expected extreme events in 2050," [4]said Peter Bauer, deputy director for Research at the European Centre for Medium-Range Weather Forecasts (ECMWF) and co-initiator of Destination Earth initiative, this week.
Computer scientists will be able to tweak the digital twins to assess hypothetical scenarios, such as the outcome of adding wind farms in various locations across Europe or where crops should be best grown in the future as conditions change. The DestinE team wants to map all the processes unfolding on Earth’s surface, and be able to inspect the model down to a kilometre by kilometre scale. In order for the digital twins to be accurate recreations, they will need to ingest data from numerous sources.
“First and foremost, a digital twin would become a data assimilation instrument that continuously cycles real-time, highly detailed, high-resolution Earth system simulations and ingests observational information from all possible instruments — including novel observatories like miniaturized satellites, drones in the Arctic, undersea cables and buoys, smart sensor arrays in crop fields and mobile phones within the expanding internet of things — also to estimate uncertain model parameters and surrogate missing process details,” states a paper describing the project [5]published in Nature.
The sheer amount of data from these instruments and devices will require a non-trivial level of compute to process, Bauer and his colleagues at the ECMWF and ETH Zurich warned in a second study also [6]published in Nature.
More power!
The team figured primarily GPU-accelerated machines are the way to go versus purely CPU or a balance of CPU-GPU. After weighing up benchmarks, and considering what technology will be developed in the near future, they believe to run a full-scale digital twin of Earth, they will need something four times the scale of today's [7]Piz Daint 25-plus petaFLOPS supercomputer, which has 5,000 Nvidia Tesla P100 GPU accelerators.
“Extrapolating this to near-future technology produces an estimate of a remaining shortfall factor of four thus requiring about 20,000 GPUs to perform the digital-twin calculations with the necessary throughput,” they wrote. “This machine would have a power envelope of about 20MW.”
US Air Force boots up not one but two AMD-powered supercomputers after five years of Intel Haswell CPUs [8]READ MORE
For comparison, America's 10MW Summit, the second most-powerful-known super listed in the latest [9]top 500 rankings , has a peak theoretical performance of about 200 petaFLOPs and today [10]sports more than 27,000 Nvidia Tesla V100 GPUs. It was built by IBM for the Oak Ridge National Laboratory.
The irony of building an energy-intensive system to tackle climate change was not lost on the researchers. They noted in their paper that the future super should be built at a location where its nodes can run on more renewable energy sources: “According to the US Environmental Protection Agency, which accounts about 1,000 lb CO2 output per MWh, such a simulation machine, if it was built in places where only ‘dirty’ power is available, would produce substantial amounts of CO2 per year.”
In order to keep the digital twin running as efficient as possible, the model’s software will need to be boosted by machine learning, we're told. The boffins believe that it will require a mix of traditional climate modelling techniques and AI algorithms to operate the various physical models. It's likely neural networks will deal with processing the incoming data and crunching it before it is passed onto the mathematical models.
[11]
“The majority of the weather and climate community remains skeptical regarding the use of black-box deep-learning tools for predictions and aims for hybrid modeling approaches that couple physical process models with the versatility of data-driven machine learning tools to achieve the best results,” the study noted. ®
Get our [12]Tech Resources
[1] https://ec.europa.eu/digital-single-market/en/destination-earth-destine
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_datacentre/hpc&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2YDjU0abij7LdHwL7IOMZ9gAAAEc&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_datacentre/hpc&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33YDjU0abij7LdHwL7IOMZ9gAAAEc&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[4] https://ethz.ch/en/news-and-events/eth-news/news/2021/02/a-highly-accurate-digital-twin-of-our-planet.html
[5] https://www.nature.com/articles/s41558-021-00986-y
[6] https://www.nature.com/articles/s43588-021-00023-0
[7] https://www.cscs.ch/computers/piz-daint/
[8] https://www.theregister.com/2021/02/10/us_airforce_supercomputer_hpe_amd/
[9] https://www.top500.org/lists/top500/2020/11/
[10] https://blogs.nvidia.com/blog/2019/11/19/record-gpu-accelerated-supercomputers-top500/
[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_datacentre/hpc&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44YDjU0abij7LdHwL7IOMZ9gAAAEc&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[12] https://whitepapers.theregister.com/
Re: We all do...
Its a black box model. We use them in lots of things all the time. Very effectively. For some things you dont need explanations. IF AI can come up with a model that accurately models something accurately enough to match reality over a period of time then its probably going to be of value in prediction. While it would be nice to have an 'explanation' it may have found something it will take us a long while to understand.
It's not enough captain.
They should build an even bigger machine with even higher power consumption and then simulate what the effect on climate would have been if they didn't waste energy on the simulation. Much more useful data to be cleaned from there.
Who needs this?
The answer is 42. We already know this.
Re: Who needs this?
Ah, but what is the question?
Re: Who needs this?
That's what the supercomputer is for!
Question
Question - "will I be able to put the washing out tomorrow?"
Re: Who needs this?
I was going to post something along the same lines, but just want to add that I'd love to see one of these rigs produce schematics for an even bigger supercomputer that will be able to compute the question.
I can see where this is going
By the early '30s DestinE will have become a singularity and will begin remedying the prime causes that are threatening the Earth's climate and ecology.
Someone needs more FPS in Quake.
If you are planning a two-metre high dike in The Netherlands, for example, I can run through the data in my digital twin and check whether the dike will in all likelihood still protect against expected extreme events in 2050," said Peter Bauer, deputy director for Research at the European Centre for Medium-Range Weather Forecasts (ECMWF) and co-initiator of Destination Earth initiative
And for what one imagines would be likely to use far less expended energy and be significantly more practical use, one could just spend the extra time with a JCB and pike an extra meter or two on the dike if you had any concerns about what it was protecting getting wet.
One could buy a lot of JCBs for the price of a 20MW super computer.
And for what one imagines would be likely to use far less expended energy and be significantly more practical use, one could just spend the extra time with a JCB and pike an extra meter or two on the dike if you had any concerns about what it was protecting getting wet.
Of course if your "2metre dyke" is a €100m dyke programme surrounding a significant area, it's probably cheaper and more efficient to bid for a couple of days on DestinE to check your calculations than to "just build it" and find it doesn't work because of a fundamental error in your design.
Muckshifting and civils are expensive. Increasing a 2metre dyke to 3metres doesn't increase material use by 50%, you're actually getting closer to double the material as the footprint gets wider (a problem if you're building up to a river or a housing development as your dyke is now in the water or inside people's kitchens).
20MW
20MW is a shedload of power. It's the equivalent draw of a small town with 25k homes.
Equally though, the new GE Haliade-X turbines provide 12-14MW each with a 60-64% capacity factor. So if they sponsor one and a half turbines, that'll largely cover them (with some hydro as backup for wind drops). The new Dogger Bank farm is installing hundreds of them. Magnificent bits of engineering (albeit they need to work out how to recycle the things at EOL).
Re: 20MW
Surely somewhere like the Dogger Bank is the right location for all of this: all that waste heat could be used for a fish farm.
Re: 20MW
It sounds like the timeline is similar to when the US and Canada are looking to start deploying micro-reactors (not sure on the Russian and Chinese timelines). So a couple of these may also do the trick (typically these are specced around 10MWth, so around 4MWe). The spare heat capacity could also, potentially, drive heat engines to drive the cooling systems, although that's probably looking a bit further forward.
We all do...
“The majority of the weather and climate community remains skeptical regarding the use of black-box deep-learning tools for predictions" and projections, and anything else.
You have a hard time to tell why things should work the way the "model" says. Qutation makrs because for me a model should be based on understanding things like physics or dynamics or the maths/stats on the appropriate scale (yes, you will need to treat eddies on different scales differently, and you should include non-deterministic trajectories caused by the time and space scale separation of the small scale turbulence on large scale weather patterns) - no, that is not easy. But expecting "AI" to come up with "explanations" is... well... suboptimal.
Plus the deep learning can only analyse (and likely predict / project) only stuff it has observed in the past. I guess that high enough quality data is probably not available for more than 50 years (with sufficiently high spatial coverage)...