I am currently testing the example hashes on windows.... looks good: recovered + no freeze.
Just to make sure, do you have this patch for instance:
https://hashcat.net/wiki/doku.php?id=timeout_patch
I am currently investigating, but don't see anything suspicious so far w/ windows version.
I think that I do have the patch but I will check to make sure when I come back home. I will install catalyst 13.8beta1 instead of beta2 and test again. You are doing your test on which Windows platform and drivers? Also, do you have to reinstall the patch every time that you change drivers or you do it once and it never gets changed?
Unfortunately, I have only one win32 system w/ windows7 currently running here - gtx 580.
Therefore, neither the bitness nor the GPU vendor fit ...
Maybe I can later on reboot another system into windows7 x64 with an amd 7970.
I don't think that the bitness is a problem, but the amd beta driver on windows might be... (also if... I thought some other beta testers already tested w/ new rc7 and similar setups)
The patch is a windows registry modification, it shouldn't be modified by driver updates etc... but I'm not totally sure, you should/could simply double-check w/ regedit.
OSX 10.8 is not a very popular algorithm so even if they have the same setup and test other algos, they might not have any problems. Anyway, we'll know better once I can put the proper setup at home.
You are right, but anyway any problem should be investigated... we just need to be able to reproduce.
Good news: Rolf told me that he will also test on win7x64 with similar/same setup.... so we will hopefully know soon what the problem is
Okay, the problem is still there.
It does not crash the driver(I have wddm timeout set to 15 seconds and used --runtime 30 flag).
The problem caused total system freeze (even irc sessions got disconnected, which is disturbing), but WDDM driver restart was not invoked (important!).
Win7 x64 SP1 + GTX Titan.
What bugs me is the lags: the system lags like hell even with -n 2 -m 7100.
It shouldn't!
A curious finding: when ran with --runtime 30, if you press [p] midway, it will pause, lags will stop, and it will safely quit due to runtime limit met after some time.
Now, this may be a separate bug, but should pause behave like that?
rc8 is up! it should fix "lagging" screens and hopefully give the user a more reliable feeling at all since the commands should react faster.
do not forget, kernels like 7100 are ultra-slow. afaik they are the slowest at all. 40k round should result in ~400 H/s, not more. since you run 4k streams with default -n settings at a single 7970, it takes at least 10 seconds to give oclhashcat-plus a chance to refresh the performance meter. the changes i did will also affect other "slow" algorithms.
however, if you press "q" it should exit immedialy and not freeze the screen anymore, hopefully..
(08-23-2013, 09:35 PM)atom Wrote: [ -> ]rc8 is up! it should fix "lagging" screens and hopefully give the user a more reliable feeling at all since the commands should react faster.
do not forget, kernels like 7100 are ultra-slow. afaik they are the slowest at all. 40k round should result in ~400 H/s, not more. since you run 4k streams with default -n settings at a single 7970, it takes at least 10 seconds to give oclhashcat-plus a chance to refresh the performance meter. the changes i did will also affect other "slow" algorithms.
however, if you press "q" it should exit immedialy and not freeze the screen anymore, hopefully..
Now it runs very smoothly. I loaded 10 hashes and did dictionary attacks. There is no hangs and I have cracked the Wiki's example. I think this bug is solved.
P.S: I also downgraded the catalyst to 13.8 beta1 but I don't think it had an effect on that bug per se.