Open soxrok2212 opened 3 years ago
Can you please provide the benchmark results for these two tasks and what chunk lengths were assigned in these cases?
On a single GTX 1080, it benchmarked as the following (from debug output):
'type': 'speed', 'result': '370103:137159056900.88'
The chunk length is 600s
I also noticed this issue, currently running an attack hybrid attack with the command tasks, its really frustrating and makes me need to make dictionaries from all these masks to make it go full speed. It would be really helpful if this could be resolved.. Also i noticed the computers are generating a lot less heat now its going slower, which is expected i guess. Chunksize is 600
Attack Commands: "-a6 #HL# wordlist.txt ?d?d" (Speed : ~250.000 H/s) "-a6 #HL# wordlist.txt ?d?s" (Speed : ~250.000 H/s) "-a6 #HL# wordlist.txt ?d?d?d" (Speed : ~50.000 H/s) "-a6 #HL# wordlist.txt ?d?d?s" (Speed : ~50.000 H/s) "-a6 #HL# wordlist.txt ?d?d?d?d" (Speed : ~4000 H/s) "-a6 #HL# wordlist.txt ?d?d?d?s" (Speed : ~4000 H/s)
As a workaround, you should be able to add the -S flag to have the candidates generated on the host, which should avoid this problem. Unfortunately, this is not supported by hashcat, so one workaround would be to add hashcat itself as a preprocessor and use it instead of PRINCE. This restores full speed in a hybrid attack, but you no longer get reliable readings of progress.
Hashtopolis: 0.12.0 commit 2402201, client version 0.6.0.20
I have a wordlist with about 400,000 words in it. Task is running against WPA mode 22000. When running hybrid mode -a 6 with ?a or ?a?a, all is fine. When running ?a?a?a or more, speed goes to roughly 1/10 of expected.
Tested on hashcat locally and could not reproduce the issue, it only shows when running through hashtopolis. Both are using hashcat v6.1.1.
Server is on Ubuntu 20.04 arm64 build, both windows and linux agents are affected.
GPUs are GTX 970, 980 and 1080 all using CUDA, but that shouldn't matter.
Example:
#HL# -a 6 english.txt ?a
is fine, but.#HL# -a 6 english.txt ?a?a?a
runs poorly.I've tried adjusting the chunk size between 600 and 1600s, doesn't seem to have an effect.