Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 31 Oct 2014 08:26:58 -0800
From: Royce Williams <royce@...ho.org>
To: john-dev <john-dev@...ts.openwall.com>
Subject: Re: descrypt speed

On Fri, Oct 31, 2014 at 6:21 AM, Solar Designer <solar@...nwall.com> wrote:
> On Fri, Oct 31, 2014 at 06:58:20PM +0530, Sayantan Datta wrote:
>> We're transferring around 1.1MB to the GPU with every kernel call.
>
> How many candidate passwords does this correspond to?  Also, is this
> single kernel call applied to all loaded salts at once (already in GPU
> memory)?  Sorry I don't recall this detail.
>
> Oh, I just realized magnum and Royce never mentioned the loaded hash and
> salt counts, just the c/s rate.  So it was a meaningless comparison!
> This explains it.

My fault for not reporting single-hash results.

My original reported performance was in an x4-speed x16 slot, against:

Remaining 28 password hashes with 26 different salts

... with this performance:

No fork: 32976 Kc/s

0g 0:00:00:14  3/3 0g/s 1266Kp/s 32976Kc/s 35579KC/s GPU:39°C fan:35%
az15mv..rdmahb

With fork=2: 57699 Kc/s aggregate

1 0g 0:00:02:20  3/3 0g/s 1107Kp/s 28889Kc/s 31254KC/s GPU:62°C
fan:46% kdl2ik..hlh8VM
2 0g 0:00:02:20  3/3 0g/s 1107Kp/s 28810Kc/s 31175KC/s GPU:62°C
fan:46% 8s451t..g0o2u

With fork=4: 59281 Kc/s aggregate

2 0g 0:00:01:55  3/3 0g/s 366695 p/s 9562Kc/s 10300KC/s GPU:60°C
fan:44% lyl8.b..dyco5t
3 0g 0:00:01:55  3/3 0g/s 656097 p/s 17116Kc/s 18437KC/s GPU:60°C
fan:44% 2kuirr..dth7rj
4 0g 0:00:01:55  3/3 0g/s 602244 p/s 15657Kc/s 16861KC/s GPU:60°C
fan:44% 1qjks2..mc35sw
1 0g 0:00:01:55  3/3 0g/s 651466 p/s 16946Kc/s 18257KC/s GPU:60°C
fan:44% rmads2d..trge15m


I then moved it over to a true x16 slot.  In that new setup, here's a
single-hash set of tests, each baked at least 5 minutes, using:

./john --format=descrypt-opencl --dev=gpu descrypt.single

With no fork: 33522 Kp/s

0g 0:00:11:32 0.00% 3/3 (ETA: 2021-03-03 05:44) 0g/s 33522Kp/s
33522Kc/s 33522KC/s GPU:56°C fan:43% b1aat1ck..b1313012

With fork=2: 48019 Kps aggregate

1 0g 0:00:05:00 0.00% 3/3 (ETA: 2019-06-25 11:28) 0g/s 22848Kp/s
22848Kc/s 22848KC/s GPU:62°C fan:46% bial1nk3..bikr16jo
2 0g 0:00:05:00 0.00% 3/3 (ETA: 2019-01-19 17:51) 0g/s 25171Kp/s
25171Kc/s 25171KC/s GPU:62°C fan:46% shd28dk..st24mgm

With fork=3: 59039 Kp/s aggregate

2 0g 0:00:12:08 0.00% 3/3 (ETA: 2018-02-22 01:54) 0g/s 21381Kp/s
21381Kc/s 21381KC/s GPU:64°C fan:49% o2aoa8l..o0m&rv7
3 0g 0:00:12:08 0.00% 3/3 (ETA: 2019-05-01 07:48) 0g/s 15743Kp/s
15743Kc/s 15743KC/s GPU:64°C fan:49% dtjd6jc..d3s3io4
1 0g 0:00:12:08 0.00% 3/3 (ETA: 2018-01-23 13:55) 0g/s 21915Kp/s
21915Kc/s 21915KC/s GPU:64°C fan:49% af79joh..ahf22.8

With fork=4: 57041 Kp/s aggregate

3 0g 0:00:10:49 0.00% 3/3 (ETA: 2018-03-24 15:34) 0g/s 15641Kp/s
15641Kc/s 15641KC/s GPU:64°C fan:48% TMJWUV..TN1uk8
4 0g 0:00:10:49 0.00% 3/3 (ETA: 2019-09-30 05:48) 0g/s 10808Kp/s
10808Kc/s 10808KC/s GPU:64°C fan:48% Leqzh*..Lyifjz
2 0g 0:00:10:48 0.00% 3/3 (ETA: 2018-04-10 00:59) 0g/s 15437Kp/s
15437Kc/s 15437KC/s GPU:65°C fan:48% tnrnek6c..tndyrj3h
1 0g 0:00:10:48 0.00% 3/3 (ETA: 2018-05-03 09:18) 0g/s 15155Kp/s
15155Kc/s 15155KC/s GPU:65°C fan:48% gaah1oan..goob1nga


I can also do x1 tests with an extender if needed.

Royce

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ