Re: How about a 1.1 RC for Linux? ;-) (LM, MD5(MD5()), etc)
Posted: Sun Sep 04, 2011 1:57 pmgrutz wrote:Cool. I let it run on a large LM hash with 1 network client and after 2:30 hours both client and server segfaulted. Lesser amounts had no problem. When you feel like cutting a new test version I'll run it through the same paces. It sounds like you've fixed a lot of the bugs already.
Weird.
I just ran it on 500k hashes overnight in LM, and it was fine.
I think the main issue here is that my test network is a reliable LAN, and you're on the intarblags with lots of loss. This is definitely good to test, though!
Can you please continue testing with the new RC2? It includes all the fixes listed above, and should be somewhat more reliable.
http://cryptohaze.com/RCs/CryptohazeMul ... 04.tar.bz2
Also, can you try testing with a different general server architecture? Due to the possibility of CUDA driver faults or card faults killing the server (BAD), I've added the "--serveronly" flag to the system. This runs the server with no GPUs, which should make it a bit harder to crash.
So:
Run the server with ./Cryptohaze-Multiforcer [options] --enableserver --serveronly
On the same box, to use the GPUs, run ./Cryptohaze-Multiforcer --remoteip localhost
Then on the remote system, run as normal.
See if you can reproduce the segfaults with that setup. If so, I'll ship you a debug build you can run in gdb to help sort out where things are crashing.
Thanks, I really appreciate all the testing you're doing!