[BUG]: Faced an benchmark failure when using cuda toolkit "12.6.3"

[amzarahamed]

Version Information

latest

Hashcat

6.2.6

Description

I upgraded my Cuda toolkit version from '12.4.1' to '12.6.3' so that my hashtopolis agent (i.e agent-python) utilises the latest Cuda toolkit to use the GPU efficiently. When i created a task to crack an bitcoin wallet it failed with the below error,

I thought that maybe this version of cuda toolkit wasn't supported by hashtopolis. So just to double check i went inside the cracker folder and executed the hashcat benchmark command and it didn't throw any error. I also tried to execute the hashcat cracking command for bitcoin wallet "hashcat -m11300 -a3 hash.txt ?a?a?a?a?a?a?a" and it didn't throw any error. Can someone please help me on this?

[amzarahamed]

Hi @s3inlc / @hops / @zyronix,

I am new to hashtopolis. Apologies if i pulled you guys in without reaching out to you first, I honestly don't know how to proceed.
Will it be possible for someone to take a look at the above?

Thanks

[amzarahamed]

Hi Team,

The same is happening when i use cuda toolkit 12.8 as well. I tried directly executing the hashcat command within the cracker folder and it's working but when i try to execute via hashtopolis it's giving the benchmark failure error.

[phillmorgan28]

Also getting this bug with the agent. Hashcat binary runs fine on it's own, but no benchmark (or output)

[amzarahamed]

Hi @phillmorgan28,

I found out the issue. This issue seems to be happening when we try to crack an wallet hash that will contain several characters. For example I got the above error when i was trying to crack an wallet that would contain 7 character, according to my assumption it runs the benchmark to see if the instance has the required GPU resource to crack the 7 character hash and if it doesn't it fails. I tried executing the same command through hashtopolis on an instance containing 2x GPU and the benchmark succeeded.

[jessevz]

Can you run the agent with the debug flag and show the output?