Bitcoin Forum

Other => Off-topic => Topic started by: Ad on October 22, 2011, 03:34:43 AM



Title: other work for GPU's
Post by: Ad on October 22, 2011, 03:34:43 AM
 Hello everyone, just wondering about other work that GPU's can do while their value is less than the electricty to mine them.
I know there are several places where one can donate ones GPU processing power, nice to do but it doesent pay the bills.
 Does anyone of any company(s) offering payment for GPU processing? Just think, with all the crunch power all of us miners
have, a company could save the cost of a super computer just by paying for our crunch power!
 Thanks, Ad


Title: Re: other work for GPU's
Post by: Jack of Diamonds on October 24, 2011, 02:44:45 AM
There are plenty of forums which will pay for your GPU time
for cracking high-end mobile phone security codes through brute force OpenCL calculation (iPhone, Nokia, Samsung etc).

ATI/AMD cards excel in that area too, so you can use your existing setup.
Depending on the rules here, that might be legal or illegal so you should use Google if interested.


Title: Re: other work for GPU's
Post by: Transisto on October 25, 2011, 03:09:33 AM
Been searching on this for the last 45min, I can't believe there isn't a service that pay BTC for hashing power,

If someone want something cracked, he could pay let's say 2000 btc for having access to massive computing power.

Think EC2 paid in Bitcoin for cracking password.

Depending on the rules here, that might be legal or illegal so you should use Google if interested.
There is no such rules as cracking password is not in itself malicious.

Many times people loose their passwords and get told it's impossible to recover in a reasonable about of time,

now , If they were able to bid high on brute-forcing I think lots of people might be interested.

Example : Someone who has heritages documents held onto a true crypt volume of a deceased person ...


Title: Re: other work for GPU's
Post by: RandyFolds on October 25, 2011, 03:46:09 AM
Isn't it unrealistic to hope to crack SHA256?


Title: Re: other work for GPU's
Post by: Ad on October 25, 2011, 04:04:15 PM
Yes, with all the collective crunch power that we have, we could offer a great service to those in need of recovering
 a lost password/code. Once we start thinking out of the "mining box" we can do even greater things.


Title: Re: other work for GPU's
Post by: ryannathans on October 25, 2011, 11:44:56 PM
Isn't it unrealistic to hope to crack SHA256?

No.

It all comes down to what's hashed. If it's a whole document, well good luck. If it's a word or two it shouldn't be too difficult if it's just a single hash.

When someone encrypts something its different to a hash though ;)


Title: Re: other work for GPU's
Post by: RandyFolds on October 26, 2011, 12:40:47 AM
Isn't it unrealistic to hope to crack SHA256?

No.

It all comes down to what's hashed. If it's a whole document, well good luck. If it's a word or two it shouldn't be too difficult if it's just a single hash.

When someone encrypts something its different to a hash though ;)

OP mentions truecrypt. If something is encrypted with AES+twofish+serpent or whatever it is(the default setup for truecrypt), isn't that into the millions of years to crack?


Title: Re: other work for GPU's
Post by: Transisto on October 26, 2011, 03:31:59 AM
OP mentions truecrypt. If something is encrypted with AES+twofish+serpent or whatever it is(the default setup for truecrypt), isn't that into the millions of years to crack?
The default for truecrypt is AES,

It only depend on the password length,  as you may have noticed truecrypt ask for confirmation when using weak password, 14 character is considered weak I think. (very conservative)

When cracking truecrypt one does not process the whole volume to see if password match, only a part at the beginning.


Title: Re: other work for GPU's
Post by: Transisto on October 26, 2011, 03:37:30 AM
Isn't it unrealistic to hope to crack SHA256?
If you mean finding the hash generated from a random array of 40 or more bytes.
No, it is impossible,

If you mean finding the hash generated from a 12 or less character word,
Yes it may be possible.


Title: Re: other work for GPU's
Post by: RandyFolds on October 26, 2011, 03:38:16 AM
OP mentions truecrypt. If something is encrypted with AES+twofish+serpent or whatever it is(the default setup for truecrypt), isn't that into the millions of years to crack?
The default for truecrypt is AES,

It only depend on the password length,  as you may have noticed truecrypt ask for confirmation when using weak password, 14 character is considered weak I think. (very conservative)

When cracking truecrypt one does not process the whole volume to see if password match, only a part at the beginning.

I guess I never really though about the fact that volume size/contents are irrelevant. Teach us more about brute forcing stuff.


Title: Re: other work for GPU's
Post by: Transisto on October 26, 2011, 03:44:10 AM
I guess I never really though about the fact that volume size/contents are irrelevant. Teach us more about brute forcing stuff.

Truecrypt is very well documented,

http://www.truecrypt.org/docs/encryption-scheme


Title: Re: other work for GPU's
Post by: ryannathans on October 26, 2011, 07:25:04 AM
Ofcourse, we would need to know HOW it's encrypted before attempting to crack it, no use trying twofish when it's serpent ;)

Ofc, this shouldn't matter when cracking truecrypt volumes, unless we are using some extremely fast custom written software.


Title: Re: other work for GPU's
Post by: Ad on October 26, 2011, 07:53:20 PM
So what do you guys think? I have no programing skills to start a cracking operation but,
 I do have 19 GPUs that could be put to work on such an operation.
Anyone else, programing and startup skills? GPUs that could process/crunch the cracking?


Title: Re: other work for GPU's
Post by: Dan The Man on October 26, 2011, 08:06:45 PM
I think there should be a way to automate this into the bitcoin network. A client should be able to be able to broadcast a brute force problem with a money transfer (reward) embedded that requires proof of the solution to receive the money.


Title: Re: other work for GPU's
Post by: RandyFolds on October 26, 2011, 10:31:17 PM
I think there should be a way to automate this into the bitcoin network. A client should be able to be able to broadcast a brute force problem with a money transfer (reward) embedded that requires proof of the solution to receive the money.

Yeah, I wouldn't mine if that was done. You would become unwillingly complicit in all sorts of shady shit.


Title: Re: other work for GPU's
Post by: MelMan2002 on October 26, 2011, 10:47:55 PM
Yeah I've been looking as well and every time I google all I come up with are articles about bitcoin mining.


Title: Re: other work for GPU's
Post by: ryannathans on October 26, 2011, 11:46:05 PM
I have a few gpus also.


Sadly I'm not fluent enough in C or openCL D:

I know of gpu cracking programs but I don't know of anything distributed. (besides distributed.net)


Title: Re: other work for GPU's
Post by: Ad on October 27, 2011, 04:56:36 PM
So back to the orgional question. Can we provide our collective GPU power to a company or company(s)
 in need of crunching alot of data?


Title: Re: other work for GPU's
Post by: organofcorti on October 30, 2011, 11:10:28 AM
Boinc does it but not well apparently. Need to have someone with the time to develop a pool and a bespoke miner, something modular so other calculations than SHA256(SHA256)) can be performed.

Then you'd have not just corporations and crims but also researchers who can't get time on their local supercomputer to model climate change (ha! gpus modeling and causing climate change at the same time).


Title: Re: other work for GPU's
Post by: ryannathans on October 30, 2011, 12:15:21 PM
Are you implying we base it on a block system like bitcoin?


Title: Re: other work for GPU's
Post by: organofcorti on October 30, 2011, 12:20:06 PM
No, there's no need for a blockchain if you're not doing crypto. Just work received and work submitted. Is that what you meant?


Title: Re: other work for GPU's
Post by: ryannathans on October 30, 2011, 12:38:47 PM
I was thinking a worker/server system, or perhaps a p2p system?

workers download a file that contains information about the job at hand, after multiple jobs of the same kind (eg, 10 md5 hashes) the work would begin.


Title: Re: other work for GPU's
Post by: organofcorti on October 30, 2011, 12:45:25 PM
Unless you were planning on only doing the same work every time (eg rainbow table generation) then you'd need a new client or at least a new module to produce the work to send back. Like doing work for seti@home or folding at home.

Why the 10 MD5 hashes at the start?


Title: Re: other work for GPU's
Post by: ryannathans on October 30, 2011, 01:30:47 PM
If you were to bruteforce a single MD5 hash, it would take x time to go through all combinations up to 8 characters.
If you were to bruteforce two MD5 hashes, it would still take x time to go through all combinations up to 8 characters.


If two+ hashes/whatever? needs to be cracked, do them at once and it will take half as long.

I guess you can get where I am going with this. Pay more if you want it done instantly, or less if you want to wait for another of the same type


Title: Re: other work for GPU's
Post by: PatrickHarnett on October 31, 2011, 12:06:30 AM
Unless you were planning on only doing the same work every time (eg rainbow table generation) then you'd need a new client or at least a new module to produce the work to send back. Like doing work for seti@home or folding at home.

Why the 10 MD5 hashes at the start?

There is a GPU rainbow table project in boinc  -  http://boinc.freerainbowtables.com/distrrtgen/ and http://www.freerainbowtables.com/