Hashtopus - distributed solution

[undeath]

what license does this sw have? BSD? LGPL? Would be nice to see this on github.

[curlyboi]

I will need to figure this out, this is the first piece of code I have given the world

[curlyboi]

Demo was taken down since Hashtopus is now public and ppl were only constantly trying to break it

[blazer]

hi

no luck here.

http://imgur.com/a/I2GDu#0

Just getting errors, do you need a log file?

[curlyboi]

the ERROR: ......... is a hashcat output. It seems like hashcat is not reading your hashes. Check file hashlists/2 if it is correct.

[blazer]

ok you were correct. It's probably because I uploaded the hashlist as a 7z file.

Is it normal for the benchmark to run for a while?
Seems to loop again and again, not sure if this is because the benchmark option was set to 0. I changed it to 5 and it starting cracking but didn't see any load on the GPUs.

[curlyboi]

oh yeah, the benchmark cannot be 0, i would say that 10s is the optimal value.

if its cracking now, you should see the speed of agents in the task details. if you think it is too low, increase agent benchmark value manually in the task details in the assigned agents section. that will make them cut bigger chunks which can utilize gpu better. too small chunks are the most common problem with performance here.

[curlyboi]

Important! Please redownload installation package and replace server.php, there was a stupid SQL bug there. Thanks bwall for discovery.

[blazer]

hi

So I've been playing around and have some questions and comments.

a/ Trying to find more on the chunk size (seconds) param. What is an optimal setting, I assume this is dependent on keyspace?
b/ When the benchmark value is adjusted it appears to change the chunk size (above).
b-1/ When the benchmark valie is set too low it doesn't achieve optimal speeds (way underperforms)
b-2/ When it is set too high eg 12000 it seems to increase the chunk size
c/ I'm not sure if the chunk size is too big or something but when I unassign a client during a run, the client doesn't appear to get the message to stop and continues running.

It's possible that it's because i'm testing on a fast algo (MD5) so am not seeing great speed results when the chunk size is small?

Also just a suggestion, you may want to include the EULA in the agent deployment area since it is already accepted.

Btw: keep up the great work. Looking forward to running more tests.

[curlyboi]

The benchmark value at each agent says, how big chunks will be assigned to it. bigger chunks = take longer time but utilize GPU better so sometime larger chunk can actually take less time than a small one.