Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 16 May 2013 09:41:17 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: Incremental mode in 1.7.9.14

On 15 May, 2013, at 22:32 , magnum <john.magnum@...hmail.com> wrote:
> But the post states I used 0x7e and 15 (and likely a min of 0x20)... is it possible we lose precision somehow with the larger figures?

Nope. Here's unstable (which I referred to in that post) at 0x20 0x7e 15:

Loaded 74784 password hashes with no different salts (Raw MD5 [128/128 AVX intrinsics 12x])
guesses: 13805  time: 0:00:00:10  c/s: 940989M  trying: ns2d97 - ns2die
guesses: 16113  time: 0:00:00:20  c/s: 984717M  trying: frerey - frereg
guesses: 17348  time: 0:00:00:30  c/s: 994677M  trying: baborbong - baborbobb
guesses: 18307  time: 0:00:00:40  c/s: 999710M  trying: meroles1025 - merolesancy
guesses: 18877  time: 0:00:00:50  c/s: 997700M  trying: jiazmh77 - jiazmh99
guesses: 19858  time: 0:00:01:00  c/s: 994820M  trying: 9m8nQA - 9m8nQI


Here's using bleeding as of Apr 26 (last CVS import "In count_sort_t, renamed the pos field to index since its...") under same params:

Loaded 74784 password hashes with no different salts (Raw MD5 [128/128 AVX intrinsics 12x])
guesses: 13486  time: 0:00:00:10  c/s: 1310G  trying: ikypim - ikypho
guesses: 16150  time: 0:00:00:20  c/s: 1323G  trying: 1cdv? - 1cdvw
guesses: 17869  time: 0:00:00:30  c/s: 1310G  trying: splidi1 - splidi3
guesses: 18685  time: 0:00:00:40  c/s: 1304G  trying: askm9752 - 1ABO
guesses: 19567  time: 0:00:00:50  c/s: 1300G  trying: gult0ra - gult0rz
guesses: 20026  time: 0:00:01:00  c/s: 1286G  trying: f1mc9 - f1mxw


In the mentioned post I said "Unstable was 1.5% faster in terms of c/s but bleeding cracked 2.5% more passwords despite that". This is not reproduced above, bleeding is much faster (but this may be the sole reason for more guesses). I have no idea how I got the results back then. Maybe I confused things :-(

Here's current bleeding under the same conditions:

Loaded 74784 password hashes with no different salts (raw-md5, Raw MD5 [128/128 AVX intrinsics 12x])
Press Ctrl-C or 'q' to abort, almost any other key for status
13491g 0:00:00:10 1234g/s 18971Kp/s 18971Kc/s 1319GC/s ief9jg..ief9cd
16300g 0:00:00:20 778.7g/s 20780Kp/s 20780Kc/s 1343GC/s 15521sl..15521az
17943g 0:00:00:30 580.1g/s 21541Kp/s 21541Kc/s 1338GC/s 0ubag3..0ubagj
18745g 0:00:00:40 457.9g/s 21971Kp/s 21971Kc/s 1332GC/s 4lpc43..4lpc4s
19640g 0:00:00:50 385.6g/s 22201Kp/s 22201Kc/s 1323GC/s u5_cj..u5_cr
20172g 0:00:01:00 331.0g/s 22280Kp/s 22280Kc/s 1310GC/s ammyard7..ammyarit


There doesn't seem to be any regression since April 26. I must have done something wrong. Same tests but status at every 200Mp for better comparison:

Unstable:
Loaded 74784 password hashes with no different salts (Raw MD5 [128/128 AVX intrinsics 12x])
guesses: 14524  time: 0:00:00:14  c/s: 900179M  trying: chubamit1 - chubamiko
guesses: 16650  time: 0:00:00:25  c/s: 976039M  trying: 0hdgs1 - 0hdgsi
guesses: 17946  time: 0:00:00:37  c/s: 969879M  trying: pryouces! - pryoucero
guesses: 18731  time: 0:00:00:48  c/s: 982748M  trying: canitty1333 - canitty2531
guesses: 19799  time: 0:00:00:59  c/s: 987663M  trying: palmy25m - palmyd2k
guesses: 20357  time: 0:00:01:11  c/s: 974949M  trying: nd12ljg - nd12lp5


Bleeding:
Loaded 74784 password hashes with no different salts (raw-md5, Raw MD5 [128/128 AVX intrinsics 12x])
Press Ctrl-C or 'q' to abort, almost any other key for status
13348g 0:00:00:10 1248g/s 18709Kp/s 18709Kc/s 1274GC/s ofbvg..ofbvs
15653g 0:00:00:19 799.8g/s 20439Kp/s 20439Kc/s 1305GC/s 4gx y3..4gx yd
17557g 0:00:00:28 618.2g/s 21126Kp/s 21126Kc/s 1299GC/s sayorts..sayort!
18379g 0:00:00:37 492.2g/s 21424Kp/s 21424Kc/s 1290GC/s djc73g0..djc73gz
19082g 0:00:00:46 411.3g/s 21556Kp/s 21556Kc/s 1281GC/s asybase..asybama
19762g 0:00:00:55 358.3g/s 21762Kp/s 21762Kc/s 1272GC/s 3iu9e0..3iu9uu


Also, some tests with contest branch (not shown in detail here) indicates it performs very similar to bleeding and is a little slower.

magnum

Powered by blists - more mailing lists

Your e-mail address:

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