Thread 16761622

I just found out there’s a class of practical problems that can be solved efficiently by computers but they’re widely believed to be inherently sequential. That means massively parallel systems like supercomputers aren’t very effective at speeding up their solution and end up being pretty resource-inefficient. Ironically these problems are best tackled by fast single-threaded processors but those are cucked by physical constraints like the clock speed ceiling. So how do computer scientists and programmers cope with the fact that you just can’t scale up processing power for certain problems?

>>16761622 (OP)
The analogy most people use is that you can dig a long hole faster with more people shoveling dirt but you can't dig a deep hole faster with more people shoveling dirt.

>>16761622 (OP)
start looking for different types of physical computers using different material structures

accept the boundaries set by physics and ignore the retarded extrapolators

I was also thinking doesn't it put a limit on how powerful a single AI can be?

For anyone unknown to this topic
https://en.m.wikipedia.org/wiki/NC_(complexity)

>>16761987
All an AI can do is permute bits in a way that makes you more or less happy with the result. Any claim to power adheres absolutely to whether you can silence something, not whether you can say something.

>>16761995
https://en.m.wikipedia.org/w/index.php?title=P-complete&wprov=rarw1

>>16762007
This is an equivalently stupid answer.

>>16761622 (OP)
ASICS, if there's money in solving them I guess (there are limits, of course)
Clock speed is not everything

>>16761622 (OP)
Anyone using computers to dig holes is wasting their time.

>>16762346
Automated front end loader operators btfo

>>16762346
computer controlled drill bros its over

>>16762406
>b-but front end loader
>>16762423
>b-but drill bros
Might as well by using soldiers and theor pocket Internets to dig those holes.

>>16761622 (OP)
>So how do computer scientists and programmers cope with the fact that you just can’t scale up processing power for certain problems?
Probability and statistics help a lot.

>>16762456
How do either of those things help? Probability? Lol what. Statistics? Lol what.
Probability and Statistics are both sales garbage that smell like a rotting vegetable and should be disposed of in the same way

>>16762460
Oh, you're one of those kinds of retards. Have you ever actually spent any time learning probability theory/statistics, or are you just ideologically opposed to the idea that there are uncertainties worth trying to understand?

>>16762427
Hey man, you can think what you want. I'm just surprised you were going after my boy and his remote control dig-a-hole machines. He's just trying to mind his own business and is catching strays

>>16762466
Both. I get how poker works and I'm ideologically opposed to letting the house be the house forever. Tell me why you think the house should be.

>>16762468
Dig Dug was an Army AI trainer for trench warfare robots.
Prove me wrong.

>>16762469
There's certainly a lot of people who have made a living selling shit and gambling. Those are definitely not the only legitimate uses for probability theory or statistics. In fact, I'd argue most people in finance and economics programs have basically no understanding of fundamental probability theory or statistics. Once it gets beyond a linear Markov chain the speculative finance types tend to run from the real thing they fear, the need for effort.

>>16761622 (OP)
lookup table

>>16762472
I can't. You're 100% right. My dad was the commander of the army's special forces mole squadron, and I saw his training manuals once. It was all dig dug and Spelunky.

>>16762474
So, then we spend NP time searching through the table to speed up the slow but polynomial sequential algorithm? Sick solution bro.

>>16762469
>Both. I get how poker works and I'm ideologically opposed to letting the house be the house forever. Tell me why you think the house should be.
Anon, this one little trick is gonna flip your shit.
https://www.americascardroom.eu/how-to/poker-terms/rotation/

>>16762473
It's pretty easy to prove by markov chains that black scholes is lipstick on a pig. The only question is can you squeeze any lipstick back out of the pig with markov chains. I don't think so.

>>16762474
https://www.tomshardware.com/pc-components/cpus/its-been-30-years-since-intels-infamous-pentium-fdiv-bug-reared-its-ugly-head-a-math-bug-caused-intels-first-cpu-recall

>>16762480
Thank you this flipped my shit.

>>16762460
Do you understand p values?

>>16761831
Spbp.
There's usually a way to trick the universe into doing your math for you. Just have to be crafty about it.

>>16762489
I understand them enough to understand how we could hack them, as a team.

>>16762484
No, but you can use point process calculus to historically model the number and intensity of previous events of interest. Once you've established that model, you can then try to use said model to gain more insight into what factors impact the events of interest. What factors in the historical data correlated with said events occuring more frequently? What factors in the data correlated with increases or decreases in intensity?

>>16762492
We could collectively pull the fire alarm repeatedly. That will screw up all of the probabilities of false alarm! p-values are generally something only people who don't know anything about statistics give a shit about. If your results aren't Gaussian, then their squares aren't necessarily chi-squared.

>>16762506
You can also use that to model where the stream of your piss will go the morning after. It's a joke, you're either amplifying a joke or laughing along with it. No actual math is involved lol.

>>16762511
> You can also use that to model where the stream of your piss will go the morning after.

That one's really easy. No math required. Your mom's mouth.

>>16762510
If we did that, black scholes wouldn't be a priority now would it

>>16762513
No math involved there either.

>>16762515
Black scholes is a priority because there's a lot of people making a living off of selling expensive blue suits and brown shoes to undergrads. If you learned probability in the context of engineering, or computer science, or physics, or even non-financial applied mathematics, you'd probably not even know black scholes exists.

>>16762517
Hey, I had to do some complicated trig in my head to aim.

>>16762522
>complicated trig in my head to aim
This. We all know it's a joke that works