hashcat Forum

Full Version: Possible issue with charset and multi gpu system
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Hi,

I'm getting different behaviour when running on two different computers with the following command:

Code:
cudaHashcat-lite64.exe -o out.txt -s 0 -l 65610000 --pw-min 4 --pw-max 4 -m 1400 F6F2EA8F45D8A057C9566A33F99474DA2E5C6A6604D736121650E2730C6FB0A3 -1 "1234567890qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM~@#$%^&*()_+{}:|<>,./;'[]\-=" ?1?1?1?1

Computer 1: Win7 64b with GT 430 driver ver nvidia 285.62
Works fine

Computer 2: Win7 64b with 8x GTX580 driver ver nvidia 285.27
It will write the plaintext to out.txt but will keep running

Below is the console output:

Code:
C:\ProgramData\BOINC\slots\0 - Copy\oclHashcat-lite-0.09>cudaHashcat-lite32.exe
-o out.txt --pw-min 4 --pw-max 4 -m 1400 F6F2EA8F45D8A057C9566A33F99474DA2E5C6A6
604D736121650E2730C6FB0A3 -1 "1234567890qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDF
GHJKLZXCVBNM~@#$%^&*()_+{}:|<>,./;'[]\-=" ?1?1?1?1
cudaHashcat-lite v0.9 by atom starting...

GPU-Loops: 64
GPU-Accel: 80
Password lengths range: 4 - 4
Platform: NVidia compatible platform found
Watchdog: Temperature limit set to 90c
Device #1: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #2: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #3: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #4: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #5: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #6: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #7: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
Device #8: GeForce GTX 580, 1536MB, 1544Mhz, 16MCU
[s]tatus [p]ause [r]esume [q]uit => s
Status.......: Cracked
Hash.Target..: f6f2ea8f45d8a057c9566a33f99474da2e5c6a6604d736121650e2730c6fb0a3
Hash.Type....: SHA256
Time.Running.: 3 secs
Time.Left....: 0 secs
Plain.Mask...: ?1?1?1?1
Plain.Text...: *a90
Plain.Length.: 4
Progress.....: 27599120/65610000 (42.07%)
Speed.GPU.#1.:   285.1M/s
Speed.GPU.#2.:        0/s
Speed.GPU.#3.:   264.0M/s
Speed.GPU.#4.:        0/s
Speed.GPU.#5.:        0/s
Speed.GPU.#6.:        0/s
Speed.GPU.#7.:        0/s
Speed.GPU.#8.:        0/s
Speed.GPU.#*.:  2237.6M/s
HWMon.GPU.#1.:  0% GPU, 45c Temp
HWMon.GPU.#2.:  0% GPU, 46c Temp
HWMon.GPU.#3.:  0% GPU, 40c Temp
HWMon.GPU.#4.:  0% GPU, 43c Temp
HWMon.GPU.#5.:  0% GPU, 40c Temp
HWMon.GPU.#6.:  0% GPU, 42c Temp
HWMon.GPU.#7.:  0% GPU, 43c Temp
HWMon.GPU.#8.:  0% GPU, 43c Temp
^C

It gets stuck at the last line or if I don't press s it will keep running for ever.

After playing with the parameters I discovered that removing -= from charset made it work on both computers but it doesn't make sense to me.

Code:
cudaHashcat-lite64.exe -o out.txt -s 0 -l 65610000 --pw-min 4 --pw-max 4 -m 1400 F6F2EA8F45D8A057C9566A33F99474DA2E5C6A6604D736121650E2730C6FB0A3 -1 "1234567890qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM~@#$%^&*()_+{}:|<>,./;'[]\" ?1?1?1?1

I don't have these issues with version 0.08

BTW if you wonder I'm playing with distributed hashing and trying to make it cracker independent that's why the parameters are like that.

leau
This is because 65610000 > 59969536 which is 88 (nr of chars in your charset) ^ 4 (pw-min/pw-max).
(02-27-2012, 06:24 PM)atom Wrote: [ -> ]This is because 65610000 > 59969536 which is 88 (nr of chars in your charset) ^ 4 (pw-min/pw-max).

Oh my mistake I pasted the wrong version of the command line. I remove the -s -l during testing because I didn't want to have to recalculate it. BTW I always get an error when the -l value is too great.

That would be the correct version
Code:
cudaHashcat-lite64.exe -o out.txt --pw-min 4 --pw-max 4 -m 1400 F6F2EA8F45D8A057C9566A33F99474DA2E5C6A6604D736121650E2730C6FB0A3 -1 "1234567890qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM~@#$%^&*()_+{}:|<>,./;'[]\-=" ?1?1?1?1

and if I remove the --pw-min and --pw-max I get the same behavior

Code:
cudaHashcat-lite64.exe -o out.txt -m 1400 F6F2EA8F45D8A057C9566A33F99474DA2E5C6A6604D736121650E2730C6FB0A3 -1 "1234567890qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM~@#$%^&*()_+{}:|<>,./;'[]\-=" ?1?1?1?1
I guess there is also something wrong with the escaping in your cmdline. For example, you must escape the ^ char with ^ - but you dont do it?

I recommend to convert the charset to hex and use the --hex-charset parameter.

(02-27-2012, 07:00 PM)atom Wrote: [ -> ]I guess there is also something wrong with the escaping in your cmdline. For example, you must escape the ^ char with ^ - but you dont do it?

I recommend to convert the charset to hex and use the --hex-charset parameter.

Tried using --hex-charset and it works on the computer with 1 GPU but not 8.
Also I tried running with one card (-d1) and it works without changing the charset (to hex or removing characters) for both computers.

If I don't specify a charset it works on both computers also.
what exactly is the problem? if you are doing --pw-min 4 --pw-max 4 then it can not use all gpus but one. you need a keyspace that is big enough to feed all gpus with work. try again with --pw-min 8 and --pw-max 8
(02-27-2012, 08:52 PM)atom Wrote: [ -> ]what exactly is the problem? if you are doing --pw-min 4 --pw-max 4 then it can not use all gpus but one. you need a keyspace that is big enough to feed all gpus with work. try again with --pw-min 8 and --pw-max 8

It looks like it might be hardware or OS related. It runs on 6 of the 8 GPU. I used -d1 to -d8 for testing and I can also run it using -d1,2,3,... excluding the two faulty GPU. I tried it on other computers and it seems to be running fine.

FYI It didn't matter what I used (--pw-max 4, --pw-max 8, no --pw-max) as soon as I used that charset oclhashcat-lite would start but never terminate, typing s would render status info and freeze and typing q would also freeze. I had to CRTL-C to get back to the prompt.
faulty hardware can generate irritating errors.

do i understand this correct: if you exclude the faulty hardware everything works fine?
(02-28-2012, 11:09 AM)atom Wrote: [ -> ]faulty hardware can generate irritating errors.

do i understand this correct: if you exclude the faulty hardware everything works fine?

Yes that's correct.

Thanks for the help and the great software