[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <c674e11c6b21917c1b02583283ace1a9@smtp.hushmail.com>
Date: Sun, 13 Jan 2013 15:05:06 +0100
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: Changes to common-opencl.c
On 13 Jan, 2013, at 1:54 , Claudio André <claudioandre.br@...il.com> wrote:
> Patch (only a few formats was tested)
> https://github.com/claudioandre/JohnTheRipper/commit/bcde5f7ddbc3686a346274d6c6ec5a05183bfdd3
I tried cherry-picking this commit to bleeding-jumbo (not committed to GitHub yet, should I?).
Here's a couple of notes from my first "user experience". First, I ran a couple of formats that are not OpenCL:
$ ../run/john -t -fo:md5
OpenCL platform 0: Apple, 2 device(s).
Device 0: Intel(R) Core(TM) i7-3615QM CPU @ 2.30GHz
Device 1: GeForce GT 650M
Benchmarking: FreeBSD MD5 [128/128 SSE2 intrinsics 12x]... DONE
Raw: 23952 c/s real, 23952 c/s virtual
$ ../run/john -t -fo:md5crypt-cuda
OpenCL platform 0: Apple, 2 device(s).
Device 0: Intel(R) Core(TM) i7-3615QM CPU @ 2.30GHz
Device 1: GeForce GT 650M
Benchmarking: md5crypt [CUDA]... DONE
Raw: 45430 c/s real, 45430 c/s virtual
While that OpenCL output is only cosmetical, it is very confusing, especially when running CUDA. But how would we know when to mute it? Should we do a strstr(format, "-opencl") or what?
Here's an OpenCL run:
$ ../run/john -t -fo:md5crypt-opencl
OpenCL platform 0: Apple, 2 device(s).
Device 0: Intel(R) Core(TM) i7-3615QM CPU @ 2.30GHz
Device 1: GeForce GT 650M
OpenCL platform 0: Apple, 2 device(s).
Using device 1: GeForce GT 650M
Local worksize (LWS) 128, Global worksize (GWS) 262144
Benchmarking: md5crypt [OpenCL]... DONE
Raw: 57111 c/s real, 6553K c/s virtual
The second "platform" line should be omitted but that should be trivial.
Here's a --test=0 run:
OpenCL platform 0: Apple, 2 device(s).
Device 0: Intel(R) Core(TM) i7-3615QM CPU @ 2.30GHz
Device 1: GeForce GT 650M
Warning: doing quick benchmarking - the performance numbers will be inaccurate
Benchmarking: Traditional DES [128/128 BS SSE2-16]... DONE
Many salts: 4172K c/s real, 4172K c/s virtual
Only one salt: 4480K c/s real, 4480K c/s virtual
Benchmarking: BSDI DES (x725) [128/128 BS SSE2-16]... DONE
(...)
Benchmarking: WinZip PBKDF2-HMAC-SHA-1 [32/64]... DONE
Raw: 800 c/s real, 800 c/s virtual
Benchmarking: dummy [N/A]... DONE
Raw: 100504K c/s real, 100504K c/s virtual
OpenCL platform 0: Apple, 2 device(s).
Using device 1: GeForce GT 650M
Local worksize (LWS) 3, Global worksize (GWS) 9
Benchmarking: Netscape LDAP salted SHA-1 [OpenCL (inefficient, development use mostly)]... DONE
Many salts: 9000 c/s real, 4500 c/s virtual
Only one salt: 18000 c/s real, 18000 c/s virtual
OpenCL platform 0: Apple, 2 device(s).
Using device 1: GeForce GT 650M
Local worksize (LWS) 3, Global worksize (GWS) 9
Benchmarking: Raw MD4 [OpenCL (inefficient, development use only)]... DONE
Raw: 5400 c/s real, 5400 c/s virtual
(...)
It's a good thing the OpenCL header only appears once and it's probably sane to have it right at the top. But then I'm not sure about the "Using device" lines. It should probably appear before the first OpenCL format just like it does, but then it should not be repeated. Or should it? I'm not sure.
magnum
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.
