[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 3 Aug 2022 16:28:31 +0200
From: Solar Designer <solar@...nwall.com>
To: dskshnd@...il.com
Cc: john-users@...ts.openwall.com
Subject: Re: Using OpenCL with Radeon RX 580
Hi,
Unfortunately, on AMD GPUs it is expected that a varying subset of JtR
formats will fail, mostly depending on software versions. We try and
include workarounds (and sometimes find and fix bugs of our own), so
almost all formats should work most of the time, but it is nevertheless
typical to see a few fail. On NVIDIA GPUs, it is common that all
formats work. Nothing wrong with AMD GPUs per se. It's the software.
But if you want reliability with JtR, use NVIDIA.
On Wed, Aug 03, 2022 at 09:43:45PM +0900, dskshnd@...il.com wrote:
> $ ./john --format=ZIP-opencl --test=0
> Device 2: Ellesmere [Radeon RX 580 Series]
> Testing: ZIP-opencl, WinZip [PBKDF2-SHA1 OpenCL]... FAILED (cmp_one(1))
You're only testing one format. Chances are that most of the rest would
have worked. I understand that is of little help to you if you need
just this one format.
Given the specific failure mode you're seeing, you can partially work
around the problem by adding these two lines:
if (crack_count_ret > 1)
return 1;
to the beginning of cmp_one() in src/opencl_zip_fmt_plug.c.
Unfortunately, this might lead to false positives. To make those less
problematic, you could also change this line:
FMT_CASE | FMT_8_BIT | FMT_DYNA_SALT | FMT_HUGE_INPUT,
to:
FMT_CASE | FMT_8_BIT | FMT_DYNA_SALT | FMT_HUGE_INPUT | FMT_NOT_EXACT,
Indeed, all of this is pretty bad. Ideally, the code would have worked
as intended. I'm sorry we don't have a better answer for you.
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.
