[phpBB Debug] PHP Warning: in file [ROOT]/phpbb/session.php on line 580: sizeof(): Parameter must be an array or an object that implements Countable
[phpBB Debug] PHP Warning: in file [ROOT]/phpbb/session.php on line 636: sizeof(): Parameter must be an array or an object that implements Countable
3.14.by forum • One more GPU hash cracker - Page 2
Page 2 of 3

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 10:26 am
by panda
I see, I see. Have any solution for my friend in my previous post then? :)
Oh and I have to thank you for your kind answers! Cudos!

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 10:52 am
by IvanG
panda wrote:my friend (who is to laaazy to register here) has a radeon 4850 with latest drivers and an amd phenom quad core. the problem is that when he starts your program his whole computer freezes and, according to task manager, utilizes 25% cpu (i bet one core) however his whole computer totally freezes like its using 100% cpu.
do you have a solution for this?
It's probably video subsystem that freezes not the system itself. Though with 4850 it shouldn't happens as it fast enough to process big blocks of passwords. Anyway, you can try -blocksize switch, by default it's 23 (which means blocks are 2^23 passwords in size), lowering this value will increase video responsibility (and decrease cracking speed as well).

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 11:03 am
by BarsMonster
panda wrote:my friend (who is to laaazy to register here) has a radeon 4850 with latest drivers and an amd phenom quad core. the problem is that when he starts your program his whole computer freezes and, according to task manager, utilizes 25% cpu (i bet one core) however his whole computer totally freezes like its using 100% cpu.
do you have a solution for this?
This might be a GPU driver crash issue, caused by extreme overheat.
Bench your PC with FurMark with renamed executable, and let us know if that freezes PC too :-)

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 12:31 pm
by panda
thank you both for your answers, ill relay them to my friend when he wakes up and get back to you if it doesnt help

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 3:36 pm
by panda
he tried 3dmark 03 and it didnt freeze.
-blocksize:16 worked fine
23 still lags A LOT like you cannot do anything
22 a bit better
21 even more better but still everything lags like you have a pc from 1998
20 a lot better even though the "windows" lag he says

He is running XP Pro SP3 32Bit if that is info you need.
http://service.futuremark.com/home.acti ... 7AC6B9ABF2
Thats the info from 3Dmark

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 7:27 pm
by IvanG
Well, it's still weird that 4850 freezes. It processing one 2^23 block in ~8ms, so there must be enough time to update screen for GPU. I don't have such problem with my 4770 & 4850 though I'm using Intel Q6600@3Ghz & Vista64.

Re: One more GPU hash cracker

Posted: Fri Sep 18, 2009 9:10 pm
by BarsMonster
panda wrote:he tried 3dmark 03 and it didnt freeze.
-blocksize:16 worked fine
23 still lags A LOT like you cannot do anything
22 a bit better
21 even more better but still everything lags like you have a pc from 1998
20 a lot better even though the "windows" lag he says

He is running XP Pro SP3 32Bit if that is info you need.
http://service.futuremark.com/home.acti ... 7AC6B9ABF2
Thats the info from 3Dmark
Well, I said FurMark with renamed exe, not 3DMark, and that was for a reason :-). 3DMark does not put THAT much stress on videocard.
And before you would try it again, I need to say that you really have to rename executable of FurMark - AMD drivers throttles GPU when they notice FurMark :crazy: :crazy: :crazy:

Re: One more GPU hash cracker

Posted: Mon Sep 21, 2009 9:41 am
by panda
Apparently it was overheating and hes going to buy new fans :) thank you both. Ivan I see that you've released a new version with colors and somehow figured out how to reduce the cpu usage. Great work!

Edit:

Bug: colors. When ctr+c to stop the cracking the prompt gets the same color as the eta and speed. Think you forgot to close the color after avrspd.

Feature: Should also have another info, perhaps after curpwd that tells you which char length you are at

Code: Select all

CURPWD: gx1r0Jr  CHRLEN: 7 DONE: 00.51% HASHES: 0/1741 ETA: 1h 10m AVRSPD: 823.1M

Re: One more GPU hash cracker

Posted: Mon Sep 21, 2009 1:26 pm
by panda
I meant passlen ofc :)

Re: One more GPU hash cracker

Posted: Mon Sep 21, 2009 1:57 pm
by LordMike
panda wrote:

Code: Select all

CURPWD: gx1r0Jr  CHRLEN: 7 DONE: 00.51% HASHES: 0/1741 ETA: 1h 10m AVRSPD: 823.1M
Purely aesthetic.. Shouldn't it be "ETC"?

Re: One more GPU hash cracker

Posted: Tue Sep 22, 2009 2:37 pm
by IvanG
panda wrote:and somehow figured out how to reduce the cpu usage.
For cuda it was simply one constant change when creating GPU context. For ATI (as always!) it takes a lot of programming and it's still possible that reduced CPU usage will results in reduced GPU performance...
panda wrote:Bug: colors. When ctr+c to stop the cracking the prompt gets the same color as the eta and speed. Think you forgot to close the color after avrspd.
Yeah, it stays in highlighted mode. However you can exit in "normal way" by pressing Escape not Ctrl-C ;).
panda wrote:Feature: Should also have another info, perhaps after curpwd that tells you which char length you are at
Problem is that string already long enough to hit standard 80 symbols limit (for dual GPU configs it "AVRSPD: 1920.0M=940.0M+980.0M"), so I don't wanna add not that important information to keep it as short as possible. And anyway I have GUI in plans, so don't wanna waste too much time designing console output :).


LordMike wrote:Purely aesthetic.. Shouldn't it be "ETC"?
Depends for what "C" stands. I more like ETA, in additional to usual "arrival" it's also can be read as "attack" ;).

Re: One more GPU hash cracker

Posted: Tue Sep 22, 2009 7:41 pm
by LordMike
IvanG wrote:
LordMike wrote:Purely aesthetic.. Shouldn't it be "ETC"?
Depends for what "C" stands. I more like ETA, in additional to usual "arrival" it's also can be read as "attack" ;).
ETA is Estimated Time of Arrival
Whilst
ETC is Estimated Time of Completion

Re: One more GPU hash cracker

Posted: Tue Sep 22, 2009 8:40 pm
by panda
IvanG wrote:For cuda it was simply one constant change when creating GPU context. For ATI (as always!) it takes a lot of programming and it's still possible that reduced CPU usage will results in reduced GPU performance...

Yeah, it stays in highlighted mode. However you can exit in "normal way" by pressing Escape not Ctrl-C ;).

Problem is that string already long enough to hit standard 80 symbols limit (for dual GPU configs it "AVRSPD: 1920.0M=940.0M+980.0M"), so I don't wanna add not that important information to keep it as short as possible. And anyway I have GUI in plans, so don't wanna waste too much time designing console output :).
I havent noticed any reduced gpu performance though. Maye a little little bit but thats better than having a core max'ed :) Youve done a great job"

Hehe didnt know the escape function :)

Re: One more GPU hash cracker

Posted: Wed Sep 23, 2009 2:57 pm
by D3ad0ne
IvanG have a feature request. Would it be possiable to make the IBP hashes batch friendly? For example if you had a text file with 10 IBP hashes (hash:salt), then you ran "ighashgpu.exe /t:ipb /c:s /max:6 ibp.txt" it would go through each item in ibp.txt look for passwords in lowercase up to the length of 6. Then it goes to the next item until it reaches end of file. Then it could just output the results in ighashgpu_results.txt just like the multi-hash does.

I've tried to create a batch file do this for me but the output file ends up with thousands of "CURPWD: aaa DONE: 69.23% AVRSPD: 11.9M" etc etc for each update there is on the screen. I suppose i could pipe it through grep but might just be easier for you to update your exe.

Re: One more GPU hash cracker

Posted: Wed Sep 23, 2009 6:26 pm
by meinname123
I have a request too:
it would be great if resuming would be better. like saving actual position every 5 minutes or so to a file and resume from this position on start if this file was found or if -r is set..

and i noticed, that the password generation is a little bit strange.
other programs count up the first postition, then 2nd, then 3rd and so on till the last position which is counted up as last. - but on ighashgpu the 5th position is counted up last.
This makes it harder to change from other programs like egb to ighashgpu (i have a huge attack running which is in the middle of ETA 2 years..)

Re: One more GPU hash cracker

Posted: Thu Sep 24, 2009 5:19 am
by IvanG
Queuing hashes, advanced statistics/options, auto-save/resuming, etc -- everything planned for GUI version. I guess it'll commercial though as I really bored to fight with ATI SDK/crappy written software and PR their GPUs for free :lol:.

About password generation -- it done that way to get optimal speed. When I'll implement, for example, mysql hashes there'll be another password generation scheme for them as it'll faster.

Re: One more GPU hash cracker

Posted: Thu Sep 24, 2009 10:19 pm
by D3ad0ne
I don't know if any one uses hashlist for salted attacks but I don't have time to sit and copy & paste each one from a list. So I wrote a quick VB program to do it. All that should be needed is my vb program, ighashgpu.exe, and a hashlist. Must be delimted hash(32/40):salt no whitespace. Also you must use the ighashgpu.exe that is in the attachment as it is slightly modified to work with my program. Once started you will see a blank cmd prompt pop up in the task bar, and your machine may slow down. Output goes into the result_output.txt in the run directory Msg box will pop up notifing when entire hashlist is complete. I mostly made this for my use and until IvanG comes out with his GUI version. Any questions just let me know.

Re: One more GPU hash cracker

Posted: Mon Sep 28, 2009 6:10 am
by d0PING
@IvanG:obviously, you had written your stuff in .ptx and maybe compiled with ptxas, but where's 'module' aka .cubin in raw???
example:
architecture {sm_10}
abiversion {1}
modname {cubin}
code {
name = my_kernel
lmem = 0
smem = 24
reg = 3
bar = 0
const {
segname = const
all cuda programs must have it!!!(i guess :crazy: )

Re: One more GPU hash cracker

Posted: Mon Sep 28, 2009 8:00 pm
by K.9
D3ad0ne wrote:I don't know if any one uses hashlist for salted attacks but I don't have time to sit and copy & paste each one from a list. So I wrote a quick VB program to do it.
It's nice but 1 strong password would just need to long. And the others wont get cracked :|

Re: One more GPU hash cracker

Posted: Mon Sep 28, 2009 8:47 pm
by D3ad0ne
It's nice but 1 strong password would just need to long. And the others wont get cracked
Well that kinda depends, the way I look at it is like this. Lets say I have 200 salted IPB hashes and I run my batch program on that list. Lets say I have the program set to look for min:4 max:8 c:sd (small + digits) that means there are 2821108227840 combinatinons, at a rate of 984000000 attempts a second my machine can compute every combination in about 48 seconds. 48sec X 200 = 2.7 hours to complete all the hashes even if it didn't find a single password. But that is at the very least 2.7 hours I didn't have to sit in front of my computer copying each hash into a command prompt waiting 45 seconds to see if it cracks it and then moving on to the next one.

Your right if you consider doing min:4 max:7 c:csd = 60 mins per hash X 200 = 200 hours comes out to 8.3 days. Thats only if I can't find a single password, but thats still 8 days I'm not sitting at my computer copying hashes from a text file. Not to mention that at the same time I could be running that same hash list through dictionary list like passwords pro. Because passwordspro only uses CPU and doesn't affect GPU speed. Least not on my core i7. :D

Re: One more GPU hash cracker

Posted: Tue Sep 29, 2009 7:01 pm
by IvanG
d0PING wrote:@IvanG:obviously, you had written your stuff in .ptx and maybe compiled with ptxas, but where's 'module' aka .cubin in raw???
No point to use .ptx, compiler is good enough to use plain C. Cubins compressed and stored inside .exe. Actually if you can't find them by you own then you don't need them at all :P.

Re: One more GPU hash cracker

Posted: Tue Sep 29, 2009 7:27 pm
by K.9
D3ad0ne wrote:Not to mention that at the same time I could be running that same hash list through dictionary list like passwords pro. Because passwordspro only uses CPU and doesn't affect GPU speed. Least not on my core i7. :D
That's the point. If you try to crack more than 10,000 salted hashes it would take too much time.

Re: One more GPU hash cracker

Posted: Tue Sep 29, 2009 8:59 pm
by D3ad0ne
That's the point. If you try to crack more than 10,000 salted hashes it would take too much time.
So is your answer to run the 10,000 hashes through a dictionary attack and call it a day? "Oh to many hashes to try and brute-force, guess that's that."
Even with 10,000 passwords running a batch of min:4 max:10 c:d takes 9 sec each that's only 24 hrs or min:4 max:8 c:s takes 13 sec, that comes out to 36 hrs.

Just last night I was given a list of 1574 salted hashes. I ran the list through simple dictionary attacks using passwords pro and found 186 or 11.8% passwords. This took all night, during that time I ran the same list through preset 1 and 2 using my batch program. I found 5 that the dictionary attack didn't get, and 12 more using preset 2. Sure that's only 1% more but that's 17 more I wouldn't have had otherwise.

K.9 So I'm curious do you get salted hash list of more than 10,000? And if you do what do you do with it?

Re: One more GPU hash cracker

Posted: Wed Sep 30, 2009 2:12 pm
by K.9
K.9 So I'm curious do you get salted hash list of more than 10,000? And if you do what do you do with it?
19537x vbulletin :crazy:

http://hashkiller.com/index.php?topic=1810.0

Re: One more GPU hash cracker

Posted: Wed Sep 30, 2009 3:22 pm
by D3ad0ne
K.9 Holy crap thats a lot of hashes. Granted 27.4% cracked by dictionary alone? That is pretty good.

But I ran your list of 14,165 unfound hashes for 29 minutes on my really slow work machine (9400gt). I only checked 1000 hashes with a min:4 max:4 c:s. From those 1000 hashes I was able to get 8 passwords.

All I'm saying is that it may be worth it to run the list through even a low level brute force to make sure you get those couple that are easy to crack but just not in your dictionary files.

***Edit ***
When I got home I ran the whole 14,165 through a min:4 max:5 c:s and found 171 more in 38 minutes. And another 1,533 using a min:7 max:9 c:d So thats another 12% just by simple bruteforcing K.9 I PM'ed you the list if you want it.