[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20241117173215.GA4474@openwall.com>
Date: Sun, 17 Nov 2024 18:32:15 +0100
From: Solar Designer <solar@...nwall.com>
To: john-users@...ts.openwall.com
Subject: Re: Performance problem running on gpu (windows 10)
On Fri, Nov 15, 2024 at 07:26:53AM -0800, jeff wrote:
> john.exe --format=NT-opencl
> --wordlist=\pw-crack\dictionaries\found-ntlm_811m.dic --rules=all
> --dupe=0 \pw-crack\pwn_ntlm_88m.rawest
We do not have an on-device wordlist rules engine, so a command like
this means that you're applying rules on host, but computing and
comparing hashes on device. So for a fast hash such as NT, the host and
data transfer become the bottleneck, leaving the device mostly idle.
We do instead have an on-device mask engine, so you should use that.
> Device 1: NVIDIA GeForce RTX 3060 Ti
> Using default input encoding: UTF-8
> Loaded 88409988 password hashes with no different salts (NT-opencl [MD4
> OpenCL])
> Note: Passwords longer than 27 rejected
> Note: This format may be a lot faster with --mask acceleration (see
> doc/MASK).
As this tells you, it "may be a lot faster with --mask acceleration (see
doc/MASK)." In most cases, this message means that it's pointless
(performance-wise) to use this format without a mask. Maybe we need to
use stronger wording there.
So please do see the examples given in doc/MASK. You'll generally want
to add masks on top of other modes, but you may also run some masks on
their own.
Alexander
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
