12-05-2013, 08:19 PM
Hi. I know these has been reported in the trac (#203 and #90) already but they are still present in the beta62. I am willing now to take the time to help in troubleshooting them in a reasonable time. I did a couple of tests with the following specs: Windows7 64 bit, catalyst 13.8beta1, oclhashcat 1.0b62, 2 x HD6970 + 1x HD6950. My router at home died and I will only have internet in more or less 48 hours (I am writing from work) but I can summarize my findings. Once I have internet back, I will also send the .restore files created.
Test1: Simple wordlist attack with 1 x 6970 on about 450000 salted hashes (-m 1420).
Result1: Upon stopping after about 2 minutes of running using CTRL-C, restoring the session works but the progress jumps from 2.5 to 22%. The remainder time is also greatly underestimated and go up over time.
Test2: Simple wordlist attack with 2 x 6970 on about 450000 salted hashes (-m 1420).
Result2: Upon stopping after about 2 minutes of running using CTRL-C, the session refuses to restore giving the following error message:
Additional info that might help or not: The Speed.GPU.#x is accurate and different for each card but the HWMon.GPU.#x is the same for both card (so inaccurate for one of the card). Also, if I open the .restore file with Notepad, the only GPU reported is 1.
Test3: Simple bruteforce with 1 x 6970 on about 450000 salted hashes (-m 1420).
Result3: Upon stopping after about 2 minutes of running using CTRL-C, restoring will show some workload on the card but there is no progress showing/happening. Additional info that might help or not: the Elapsed time picks up exactly at the point where it was stopped.
Test4 and result4 are the same as for test2 but for bruteforce.
I think that it's enough to work with for now. I will send the files as soon as I can.
Test1: Simple wordlist attack with 1 x 6970 on about 450000 salted hashes (-m 1420).
Result1: Upon stopping after about 2 minutes of running using CTRL-C, restoring the session works but the progress jumps from 2.5 to 22%. The remainder time is also greatly underestimated and go up over time.
Test2: Simple wordlist attack with 2 x 6970 on about 450000 salted hashes (-m 1420).
Result2: Upon stopping after about 2 minutes of running using CTRL-C, the session refuses to restore giving the following error message:
Code:
ERROR: Loaded number of devices is async to data in restore fileTest3: Simple bruteforce with 1 x 6970 on about 450000 salted hashes (-m 1420).
Result3: Upon stopping after about 2 minutes of running using CTRL-C, restoring will show some workload on the card but there is no progress showing/happening. Additional info that might help or not: the Elapsed time picks up exactly at the point where it was stopped.
Test4 and result4 are the same as for test2 but for bruteforce.
I think that it's enough to work with for now. I will send the files as soon as I can.