Because that’s the cooling system required to run the thing. It requires more toxic coolant, that will eventually end up in the ocean, than several hundred supercomputer megaclusters and sucks more power than a thousand suburbs.
I believe you may have misread your own source.
For example, the world’s fastest supercomputer, Frontier, draws 8 megawatts when it idles — a quantity that could simultaneously power thousands of homes
If this was the basis for your saying this…
several hundred supercomputer megaclusters and sucks more power than a thousand suburbs.
… then you misread AND misstated.
Misread: this “thousands of homes” energy use was in reference to Frontier, which is not a quantum computer but based on more conventional architecture, the kind the article goes on to say might eventually be improved upon by quantum computing. Eg:
Consequently, experts are looking to new strategies that can rein in energy use while continuing to improve computing performance. One proposed solution: quantum computing.
Misstated: “thousands of homes” != “thousands of suburbs.”
A suburb is not a home but a a collection of homes, a region of a city even. See definition:
an outlying part of a city or town. b. : a smaller community adjacent to or within commuting distance of a city. c. suburbs plural : the residential area on the outskirts of a city or large town.
So in your zeal to make your point you demonized quantum computers, which could be a solution to the problem you’re ostensibly so concerned about, and in the process you misstated a metric by at least one order of magnitude.
So yeah… I don’t know what to tell you. You really messed up here. Your problem is with LLMs and big compute, not necessarily quantum computers.
Because that’s the cooling system required to run the thing. It requires more toxic coolant, that will eventually end up in the ocean, than several hundred supercomputer megaclusters and sucks more power than a thousand suburbs.
Congrats on getting upvotes for this utter bullshit, none of which is substantiated in the article you linked.
Edit- this was at +15 when I first saw it.
Helium is not toxic and it sure as fuck isn’t going in the ocean after it escapes a quantum computer.
But it is difficult to acquire here on earth.
As uranium and thorium naturally decay underground, they produce some helium as well. That’s why you can literally make a helium mine. On earth it’s also a finite resource, because once released into the atmosphere, it will eventually escape the atmosphere and end up in space.
I’m prepared to be proven wrong on this, as my exposure to quantum computer cooling systems has been super brief, but as far as I know there are no toxic coolants.
The pre cooler is a Pulse Tube Refrigerator, and the main cooler is a Dilution Refrigerator. Those both use helium, and that stuff floats out into space. I doubt it’s going into the ocean. Here’s another article that talks about the operation.
Like I said though, my exposure was brief. Unfortunately we didn’t land any projects with the supercomputer people 😞. I’m always down to learn more about niche topics though. Makes me super fun at parties. If you have good sources shoot them my way. I couldn’t find anything in my 5 ish minutes of web searching.
Wrong. I breathed in some helium once and it made my voice all high pitched which threatened my fragile masculinity. Very toxic.
(/s…)
Never worked much with cryogenics, but the one thing I learned was to never get in an elevator with (large quantities of) liquid nitrogen — if the elevator stops it can displace the oxygen and that’s…kinda bad.
Yeah, totes. Scentless non-toxic gases can still be deadly by merit of not being oxygen.
The only recreational octave-shifting gas I indulge in is Sulfur Hexafluoride. Bolsters the ol’ baritone.
Wow, that is insane! But it’s also amazing that it has been able to solve problems that humans haven’t been able to solve in over 50 years.
What are you talking about? Most of these things are experimental and none of them have solved a single fucking thing.
From the linked article
In 2020, the artificial intelligence (AI) software AlphaFold demonstrated that it could predict the three-dimensional structure of proteins from their amino acid sequence2, a 50-year old grand challenge in biology.