[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

View post   
File: 453 KB, 658x491, chinese-supercomputer.png [View same] [iqdb] [saucenao] [google]
3984439 No.3984439 [Reply] [Original]

supercomputers... what do you actually DO with them...?

>> No.3984443

finally run Crysis on maxed settings

>> No.3984450

its funny how computers went from being the size of a building to desktop size back to the size of a building.

>> No.3984458

Ask it to teach you how to love.

>> No.3984476

Run simulations of crazy stuff that takes ridiculous power to run.

Simulations of the evolution of the universe, for example.

>> No.3984480

Keep track of everything you do

>> No.3984483

Breaking encrypted messages from countries that aren't your own.

>> No.3984494

>>3984450

They never really stopped building supercomputers. Governments have always wanted the upper hand in computation power for decryption, or other things.

I'd imagine a lot of number crunching goes on in there as well. Governments keep a shit-ton of records, and people couldn't really go through all that on their own these days.

>> No.3984501

Playing Jeopardy

>> No.3984515

fluid dynamics

>> No.3984519

>>3984494
i wonder if the whole bit coin thing is just a superconducting scam....

>> No.3984523

Video games. True story.

>> No.3984526

Q: why not just do everything with distributed computing?

>> No.3984536

>>3984526
Folding@Home was great.

Back when I gave a shit about vidya. ah notalgia.

>> No.3984543

Supercomputers do number crunching. Things like password bruteforcing, encryption/decryption, and predicting weather forecast data.

>>3984526
What do you mean by 'do everything'? An average computer user doesn't need to do any huge amounts of data processing.

>> No.3984572
File: 196 KB, 661x768, moot servers wat.jpg [View same] [iqdb] [saucenao] [google]
3984572

They make a handy hiding spot.

>> No.3984576

storage, lots and lots of storage.

>> No.3984595

>>3984576

This man has no idea what he is talking about.

>> No.3984596

>>3984526
what do you think supercomputers are?
they're basically lots of computers wired together so they act as synchronously as possible.

>> No.3984603

>>3984576
you can hollow out a fancy desk and use it to hold all the data a supercomputer will ever need to save

>> No.3984607

to mine gold on WoW.

>> No.3984612

Data simulations. HUGE data simulations.

>> No.3984623

mine bitcoins

>> No.3984630

You stick your penis into the special slot and gain the knowledge of the universe.

>> No.3984636

>>3984439
Figure out if we should spam you with p0rn, penis size videos or barbie dolls.

>> No.3984653

They're good for performing computation, but not as good at other things.

>> No.3984662

predict earthquakes

>> No.3984668

porns. lots and lots of porns.

>> No.3984685
File: 112 KB, 487x321, Supercomputing.jpg [View same] [iqdb] [saucenao] [google]
3984685

Sadly most of them are used for simulating various outcomes of stocks and short term simulations of f how a market could turn out over the course of a few days or a week using gigantic amounts of plotted data.

There are also so called "nanosecond-trading supercomputers" who trade stocks in literally nanosecond intervals to little profits who of course accumulate after millions of trades.

Very questionable stuff. I don't know what that has to do with what stock were intended to be and do.

>> No.3984690

>>3984685
excuse my typos and herpderp grammar, it's late here.

>> No.3984698

>>3984690

you also must be a nigger. I can tell from the grammar and I've seen a lot of niggers in my day.

>> No.3984713

>>3984698
No, I'm German.

>> No.3984726
File: 31 KB, 320x286, babby genes LOL.jpg [View same] [iqdb] [saucenao] [google]
3984726

>>3984698
You must be an ignorant piece of white trash.

>>3984685
You know, I think a lot of people actually forget the original intent for the stock market system. It doesn't accomplish its original task and is simply a medium for the latent benefits which it originally presented.

>> No.3984745

>>3984713
Oh, sorry. Is this better?
Sie müssen ein Nigger sein.
My apologies for the automatic translation.

>> No.3984760

>>3984439
Make gigantic fractals, what else?

>> No.3984780

My friend in environmental engineering uses them to do complex modeling of molecular dynamics / quantum mechanics that would otherwise take him months on his (awesome) laptop.

The problem is that you don't get access to supercomputers unless you can prove that you need its computing power. Oftentimes, students and professors will purposely make things more complicated than they should be just so that they'll be allowed to use them.

>> No.3984781

>>3984636
sadly, pretty much this.

>> No.3984797

>>3984439
i personally use supercomputers to solve computational fluid dynamics problems. i realize that actually having degrees makes me an outlier on sci.

>> No.3984815

>>3984797
I've also seen people use it for that in civil / environmental engineering. How long would you estimate a problem like those to take to solve on a powerful laptop?

>> No.3984834

>>3984815
There's a massive variation in complexity of solving CFD problems (i'm in aerospace, but similar methods used in civil/environmental). it ranges from using the fastest supercomputers on the planet for weeks at a time, to stuff you can do in seconds on a laptop. it really depends on the resolution in your simulation you need to get the required accuracy in your results. high quality turbulent flow simulations like direct numerical simulation of any appreciable size require millions of cpu-hours. a baseline calculation of say, steady flow on a streamlined body like an aircraft at low angle of attack, i can get you accurate lift and drag information on 64 processors in 4 hours. that's with an extremely efficient code.

i can also get you less accurate results on a laptop in 4 hours.

>> No.3984860

>>3984834

Do you mind if I ask who you work for? I'm only asking because I have a job interview coming up for CFD and have never meet/seen anyone else who works in the field

>> No.3984866

>>3984797

Fuck fluid dynamics.

Man, I hated that shit. That's all I had to say.

>> No.3984875

>>3984860
i have a BS/MS in aero, and am currently a phd student/ full time researcher in CFD. i have worked for a small CFD company you've never heard of.

what sort of cfd interview do you have, having never even met anyone in the field? i had a decent amount of cfd exposure even in undergrad.

>> No.3984884
File: 144 KB, 648x597, cscads01.jpg [View same] [iqdb] [saucenao] [google]
3984884

>>3984866
>Fuck fluid dynamics.
>Man, I hated that shit. That's all I had to say.
why, because it was difficult? it's one of the most beautiful unsolved problems in the natural world.

>> No.3984897

>>3984875

I have 2 interviews for L3. One is an optical engineering position and the other is some kind of CFD related job. My uncle is a PhD areoeng with them and apparently he tests (the simply version) if it's safe to add external "pieces" to existing airframes (ie radar dishes on AWACS).

My degree is in physics and I'm currently in a PhD astrophysics program but due to finances and desire to do something more applied I started looking for a job.

>> No.3984914

>>3984897
right on, good luck. my phd work is super applicable so my job prospects and unprovoked offers have been pretty excellent. i understand the allure of physics/astrophysics, and the bummer it must be that aere's get better jobs =)

best of luck to you sir

>> No.3986559

op here . thanks for all the interesting answers.

>> No.3986578

I would guess making super complex calculations, and running state of the art simulations.

>> No.3986617
File: 43 KB, 450x300, stallion.jpg [View same] [iqdb] [saucenao] [google]
3986617

>>3984439
Simulating stuff to figure out what the fuck is happening with lab experiments. Visualizing shit(pic related it can display a gigapixel image in near close to full resolution). Simulating more shit to figure out how to run experiments. Doing simulations of stuff you can't do in a labratory like orbital dynamics. Running code that assumes memory and computational power are infinite(code made by people like me).
Also evolving virtual creatures, not necessarily done with supercomputers, but when you have a bunch of idle computational power and administrative privileges, you tend to get ideas...

>> No.3986816

>>3986617
interesting fellow.