What would you do?

[nemo]

More of an open-ended survey than a thread, I suppose... but I'm curious:

Computing power is getting cheaper by the second, allowing tasks that were previously impossible to accomplish in one lifetime to be finished before lunch. The availability and relatively easy setup of Linux clusters means you can build a monster in your basement, and if that's not enough juice there's always distributed computing (ala SETI) so that your friends can help without giving up a machine to your twisted imagination. With this in mind, I'd like to know what you'd do if you had your own supercomputer, or perhaps what you dream of doing someday when a terabit fits into RAM.

 

Here are a few thoughts I've had wandering around for some time.

• Reverse Engineering
  
  I'd like to create an ever-expanding database of source and corresponding binary code. For instance, if I were to slap together a simple "hello world" in C on a Linux box using gcc and an x86 processor, there are only so many different compiler flags one can supply that will change the appearance of the binary itself. I would not be looking to simply database what all the different incantations of "hello world" look like, I'd be looking for what all the different incantations of the individual function calls within the program look like -- essentially fingerprinting every possible use of the printf() statement, along with whatever else I tossed in to the program. My reasoning goes something like this: for every given environment, language and compiler, there are a finite number of options available for a given end-result; if you know what every possible incantation of printf() looks like in binary form, you should be able to recognize it when analyzing binary code. Build a big enough and fast enough database and you should be able to get programs from "here's the binary" to "how'd you get the source?" in a reasonable amount of time. Or perhaps I should slow down on the hallucinogens.

  
  

• Decryption
  
  Encryption is a puzzle, a very long story problem that keeps our data safe because most people either get bored with the puzzle before they solve it, or by the time they solve it the data is no longer worth anything. Math is fun, but busting a hash longhand on legal pads is right up there with picking up my socks (sorry Irish). ;) Using an encrypted password hash as an example and borrowing from my RevEng thought above, if you had a list of every possible character combination for a password with X characters (yeah, that gets big real quick), and then had every possible hash for each password using five or six of the more popular encryption algorithms and seed values 1 - Y... Now you're playing a much bigger, but simpler game of find the match (kind of like my socks) ;)

  
  

 

Anyway, what would you do?

[GAHD]

A) Try to crack some genomes

;) Create a real world emulation program: Use it to build things I don't have the money for(like an orbital launch platform, motor vehicle, armor plating, etc...), and test them in a 'real physics' environment for flaws.

C) Tribes 2 server, unlimited players.... ;)

[UncleAl]

Access to extreme computation is a soft tool. You mold it and use it as needed at the moment. It makes no sense to stockpile answers when it is faster to generate them than to search through inventory. It is more important to avoid the Microsoft paradigm: No matter how hairball large, inefficient, and buggy your pukeware... the system will overcome it.

 

HyperChem-Lite gives me molecular structures for half the purchase price of a chemistry textbook. They are remarkably accurate vs. x-ray structure determinations for 10-20 minutes of iteration in a PC. Very nice. Suppose the fast and crude mm+ calculation could be real-time replaced with a full ab initio Hartree Fock/6-31G(d) treament that would be a week in a mainframe today. What difference would it make? Stuff would skootch over may 0.01 A out of 1.54 A nominal. How good is good enough?

 

The value would be in getting answers to questions I cannot now ask. The tool and the toolmaker are inextricably interwoven. We'll know what it's good for after it arrives. A 14-year old hacker who is not yet prejudiced by common sense will tell us.

[alexander]

Hey Nemo, great to see you around here!

So you want my world domination plans huh... well, here you go, remember, you asked for them, not me:

 

1) Have the supercomputer generate a decryption algorithm that will be able to decrypt every kind of encryption known to man.

 

2) Hold the world's largest Linux install fest, use distcc to compile everything getting a little help from the supercomputer.

 

3) Run my own Portage Mirror

 

4) Create a program with numerous libraries designed to port code from any OS to Linux

 

5) Get into every software developement company in the world (major software development companies with the exception of M$, they write crappy code ;) ), steal their popular product source code, if encrypted run it through my wish #1, run it through my wish #4, release it on my gentoo mirror, finally claim that Linux really has every program as well as some altermatives of every program under any other operating system in the world and hold the worlds largest Linux install fest.

[nemo]

good to be back...

 

you know, with that kind of power, you could finally move to Fedora :)

[IrishEyes]

hey alex,

now I know why I like you so much... you are eerily similar to nemo! I had to stop and check whose post I was reading for a second, you sounded just like him... right down to the "world domination" part. He actually has a white board up on his office wall, that has 'Plan to Take Over the World' written at the top. It makes my daughter crack up every time she asks if he has any new plans.

You two could be the next Pinky and the Brain... lol :)

[alexander]

I dont know if I'd go with Fedora. Not a big fan of Fedora core, I dont like RedHat distros in general, also not a fan of RPMs precompiled software and dozens of programs that install with your system and start at the startup for no reason. I guess I like systems that are built from the ground up, I know what, where and how most things work, no surprizing packages, a nice PMS helps too, so i dont have to keep track of RPMs and packages that i have installed. I like Gentoo because it is a fully compiled, fully customizable system, with rocking init scripts and where all control belongs to me oh, and i am not confined to a GUI to operate in (hence i dont have to have Xorg and a GUI if I dont have to, or i use the gui of my choice (which at the moment is my costom breed of FVWM) like i am in Fedora Core)... But if there is no time to setup a system and to care for it, why not Fedora :) although at that point why not use Free or Open BSD?

 

You know, with that kind of power, you'd really notice the performance difference between Fedora and Stage 1 Gentoo system.

[Turtle]

With this in mind, I'd like to know what you'd do if you had your own supercomputer, ...

Anyway, what would you do?

 

___Why Katabatak exploration of course! Several weeks ago I would also have added the topics of Phat Numbers & my brute force work on the statistical distribution of integers classed by number of divisors. These topics I have rather relinguished thanks to the beautiful & succinct expressions put forward by Qwfwq, CraigD & C1ay.

___That still leaves Strange, Peculiar, Curious, etc. numbers to hunt for (the anomolous ones that is). I won't be too surprised if one of you bright people puts forward some expression or other that allows me to drop those as well.

___That leaves Katabataks (congruence/residue), which is not a topic in search of a solution, but rather an exploration of patterns for the sake of exploring. That's what I'd do with a supercomputer. ;)

[DefectiveDNA]

Well , I think more important than having a super computer is to have a net connection to each and every computer in the world at speeeds faster than the fastest Firewire device.

 

But any ways if I have such a computer then I would like to create a virtual girl friend for me based on my dream girl.

[nemo]

Well , I think more important than having a super computer is to have a net connection to each and every computer in the world at speeeds faster than the fastest Firewire device.

mmm... high speed distributed processing vs. high speed distributed processing... :eek:

But any ways if I have such a computer then I would like to create a virtual girl friend for me based on my dream girl.

[geokker]

I'm shocked no one has mentioned intelligence. I want HAL to be my mate. True intelligence must be on the evolutionary horizon for computing. Hell, I think intelligence might even be on the horizon for humans!

[blazer2000x]

If we could make a digital artificial inteligence with access to the net, we might start seeing some of those sci-fi movies about robots taking over the world come to life. :)

 

If I had a computer with that high processing power I'd probably use it to work out any kinks in my design for a computer with higher processing power. :(

My point is, no matter how fast computers get, we'll always want them faster. (unless of course they can actually get them like they always say that things "download instantly" or "get connected in a flash") How fast exactly is instantly?

[alexander]

I'm shocked no one has mentioned intelligence.

after converting the world to open-source and eniolating microsoft corporation and making bill go bankrupt, that was the next thing on my list, i figure that first we need to fix a few wrongs now, then go and think about the future, no?

[DefectiveDNA]

I'm shocked no one has mentioned intelligence. I want HAL to be my mate. True intelligence must be on the evolutionary horizon for computing. Hell, I think intelligence might even be on the horizon for humans!

 

Are U sure that will make U more happy then having a virtual dream girl.

 

Remember that perfect girl who rejected U last time., Dont U want to have her as your life partner in any dam way.

[DefectiveDNA]

...How fast exactly is instantly?

 

Do U really wana Know ?

 

The next step after being faster is being predective.

 

Think about a super intelligence conciousness that knows everything so well that its can predict concequences of each and every thing we do accurately.

[blazer2000x]

Yeah, a computer that can memorize your thought patterns, and see where your going with the current thought, then help you get there. That would be cool. But first we have to find a way for a computer to interface with a human brain. (hopefully without sharp needles stabing through your skull) And it would have to be able to read your brain as fast as you could think, so once again were back to having an AI as fast as your brain. I think everything will pretty much have to lead up to that. :)

[geokker]

i figure that first we need to fix a few wrongs now, then go and think about the future, no?

 

Anything we're going to do is in the future.

 

If indeed all matter is based on discrete fundamental particles or 'balls', then a sufficiently powerful computer may be able to simulate a human brain. Not so much artificial intelligence rather, simulated intelligence.

 

I think I would want to perform some huge finite element analyses on such research had I a monster computer and another hundred IQ points. All the most powerful computers currently work on similar problems e.g. weather prediction, quantum physics etc.