Perimutter?
Grumbling edge computing?
Nvidia on Thursday unveiled what it called the world’s most powerful AI supercomputer yet, a giant machine named Perlmutter for NERSC, aka the US National Energy Research Scientific Computing Center. “Perlmutter’s ability to fuse AI and high performance computing will lead to breakthroughs in a broad range of fields from …
Speaking of someone who uses computer modelling in engineering quite a lot, I find projects like this mind boggling for two reasons;
Firstly the concept of managing to simplify the observable universe to the point where a small scale model can be built of it that is manageable for a super computer but still gives useful results
and secondly, the massive amount of work that goes into creating the hardware and software so that the model can be implemented.
The results of the modelling might be really interesting, but there's a whole load of interesting work that's gone on before the model even runs.
My computer isn't "optimized" for what it does, either, but throw a program at it and it runs at rapacious speed regardless.
I think a few gigamips of processing will compensate for any shortcomings in it not being a "special purpose encryption cracker", especially with those Tesla cores to crunch on the keys in parallel. A few thousand Tesla cores is just a TAD more powerful than any home machine.
Nah, encryption is dead. We've had machines that can brute force it on behalf of governments for some time now; they just don't like the fact to be known.
Since we, on our "utterly insignificant little blue green planet" at the "unfashionable end of the Western spiral arm" (thanks Douglas Adams) of just one of gazillions of galaxies, don't even know how much about the Universe we still haven't found out, it seems a teensy bit optimistic to expect any model of it we create to be "detailed" in any realistic sense.
Of course it has been postulated that the entire Universe is merely a computer game running on a higher plane, but if true that could lead to some interesting recursion management problems for a model such as this.
constructing the largest known three-dimensional simulation of the universe to date
So that will be better than
Today's Guardian story about Einstein being 'wrong'
Enough computer power to (dis)prove Einstein.
Can't get much more 'high powered' than that.
It was supposed to be online last year ... Here's an article on the subject written by a name that long-term ElReg commentards will either love or hate, from back in 2018. Needless to say, the specs have changed a trifle in the last couple years.