New largest prime number discovered by former Nvidia software engineer
New largest prime number discovered by former Nvidia software engineer
New prime just dropped. The Great Internet Mersenne Prime Search (GIMPS), a group of volunteers who search for prime numbers that are (2 to the...
According to GIMPS, this is the first time a prime number was not found by an ordinary PC, but rather a “‘cloud supercomputer’ spanning 17 countries” that utilized an Nvidia A100 GPU chip to make the initial diagnosis. The primary architect of this find is Luke Durant, who worked at Nvidia as a software engineer for 11 years
I don't know why Nvidia is mentioned at all, except the hardware. That's cool that this person found the number, but Nvidia didn't do anything except employ them once upon a time and make a product that does a thing. It's not justified to celebrate the maker of a stove when a soup kitchen feeds everyone.
This is a win for Luke and GIMPS in general, and I'm happy for them.
This is the first such prime that was discovered using GPU cloud computing. It's not just an incredible new discovery, but also a demonstration of what this type of hardware network may be capable.
They mention nvidia because that’s the hardware used to find/prove.
I find it quite relevant to have the person/ group, the strategy or method, and the device used (including chipsets). Most articles on prime number discovery will mention all these things.
The fact that he worked there seems pretty irrelevant tho.
also the person apparently spent 2 million dollars to find the number. and the money is probably from stock compensation from nvidia
I wonder if he wrote some of the CUDA code or anything like that.
We define people by their labor value in capitalist societies. It only makes sense headlines would refer to people thru the lens of their previous employer.
Wake up babe, new prime number just dropped.
Holy hell
Oontz oontz oontz oontz! Priiiiiiiime numbuh, bay bee!
Is it 3?
I bet it's 3.
Half-Life 2136279841-1 confirmed.
Almost certainly a multiple of 2 minus one
24 is the highest number.
To save you a click: 2136,279,841-1
Formatting is off.
2^136,279,841 - 1
2 to the power of something, then subtract one to make it an odd number.
Babe wake up, new prime number just dropped.
this is the first time a prime number was not found by an ordinary PC, but rather a “‘cloud supercomputer’
The first time since the 90's, before that all computer assisted Mercel primes found were found by super computers.
No first time ever. This isn’t a supercomputer, it’s a distributed cloud network that they’re referring to as a supercomputer because it has a lot of power. It’s not a supercomputer in any other sense of the word, as it’s set up on cloud providers around the globe rather than in one location in the same room.
Legit question: how is that different conceptually to the random individuals all running Prime95? Could that not also be referred to as a cloud supercomputer?
I don't know if this is a common feature of large primes, but the digits in the exponent (136,279,841) themselves represent a prime number.
that does happen to be one of the defining characteristics of mersenne primes.
And searching for mersenne primes happens to be the easiest known way to find extremely large prime numbers (via the Special Number Field Sieve I believe)
🤯
Me wondering why I haven't been able to deploy cloud instances with the A100 for an actual useful purpose for the past month
Already on the H100 bud
He must've been very anxious in order to count up to that...
It feels like people are celebrating this but hating on ai developments. not sure if these people are hypocritical or if that's two different groups of people.
You can dislike corporate hype around ai and celebrate someone finding a legitimate use case for ai.
Yeah. Stuff like this, work in medical treatments and new drugs, I'm on board.
Using it to replace human workers or steal their hard work to train them?
Fuck you sideways with a cactus, you corporate fucks.
Size does matter, after all?
I don't understand this and therefore it's stupid and pointless. Fuck you math elitist assholes with your so-called "large" prime numbers spending billions of dollars that could be used to make my life better. I don't comprehend this at all and there it does not matter. The end.
Yeah, fuck those assholes that pursue science for the benefit of humanity! I do not see why anyone should be allowed to be creative if I do not see the benefit for me in particular.
Is this Poe's law? I genuinely thought this was satire but the downvotes and responses are very serious!
I think it's satirizing the other guy in these comments.
If Jesus had wanted us to use prime numbers why did he turn the water into wine and not numbers? Checkmate atheists. /s
This is a very dangerous way of thinking. You cannot tell at the time of discovery if specific research will be useful or not down the line. You need to advance the research in all directions, even if some of them seem silly or useless, or else you will handicap your progress in other fields which you didn't see the connection with at first.
If you want it to be useful for the economy and industry in order to warrant funding, I've got news for you:
The majority of modern encryption relies on prime numbers. It is currently speculated but not known, that the number of prime numbers is infinite.
Should it be proven, that there are only a finite amount of prime numbers, all encryption would become vulnerable.
It is easily provable that there is an infinite number of prime numbers.
Many encryption algorithms rely on the assumption that the factorizations of numbers in prime numbers has an exponential cost and not a polynomial cost (I.e. is a NP problem and not P, and we don't know if P != NP although many would bet on it). Whether there are infinite prime numbers or not is really irrelevant in the context you are mentioning, because encryption relies on factorizing finite numbers of relatively fixed sizes.
The problem is that for big numbers like n=p*q (where p and q are both prime) it's expensive to recover p and q given n.
Note that actually more modern ciphers don't rely on this (like elliptic curve crypto).
There are infinite prime numbers. This has been known for thousands of years. You can find numerous proofs of this online, and go through them until one makes sense to you.
Also, quantum computers are on track to make division-based cryptography useless in the next decade or two. (Note that this only affects public key cryptography, and not shared key cryptography. So your online backups should be safe as long as you have a password for them.)
And this helps humanity and mathematicians by...nothing. It's absolutely nothing.
If we can analyze larger primes, we can generate larger primes which has applications in math, particularly cryptography and other areas, not even beginning to look at number theory. Specifically being able to verify them over a cloud is useful, we can generate them quicker and worry about their safety less. https://en.wikipedia.org/wiki/Hensel's_lemma has uses in physics actually.
Oh, you mean you don't understand it, gotcha.
Yes, and Bayesian statistics are useless too, they're all about things that have already happened!
All knowledge is good knowledge
Finally! Someone who understands! If it's not something we can understand, it's not really math.
