[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <D196B1E9-34DD-47CE-B984-E649763CE6EF@m.patpro.net>
Date: Fri, 3 Feb 2017 15:10:22 +0100
From: Patrick Proniewski <p+password@...atpro.net>
To: john-users@...ts.openwall.com
Subject: Re: to Single or not to Single
(followup)
I've compiled a fresh bleeding-jumbo, and made some benchmarks and got about the same result for every JtR-BJ I've on my disk:
JohnTheRipper-bleeding-jumbo-20160223
Benchmarking: dynamic_25 [sha1($s.$p) 128/128 AVX 4x1]... DONE
Many salts: 10105K c/s real, 10028K c/s virtual
Only one salt: 7668K c/s real, 7668K c/s virtual
JohnTheRipper-bleeding-jumbo-20160728
Benchmarking: dynamic_25 [sha1($s.$p) 128/128 AVX 4x1]... DONE
Many salts: 9867K c/s real, 9867K c/s virtual
Only one salt: 7531K c/s real, 7531K c/s virtual
JohnTheRipper-bleeding-jumbo-20161106
Benchmarking: dynamic_25 [sha1($s.$p) 128/128 AVX 4x1]... DONE
Many salts: 9960K c/s real, 9960K c/s virtual
Only one salt: 7483K c/s real, 7483K c/s virtual
JohnTheRipper-bleeding-jumbo-20170203
Benchmarking: dynamic_25 [sha1($s.$p) 128/128 AVX 4x1]... DONE
Many salts: 9849K c/s real, 9849K c/s virtual
Only one salt: 7496K c/s real, 7496K c/s virtual
I've taken a smaller sample of data (400k lines) to benchmark single mode and grabbed status line at different time:
140359g 0:00:00:51 50.00% (ETA: 14:13:44) 2738g/s 5220p/s 5220c/s 5220C/s mommysgurl
198794g 0:00:01:57 50.00% (ETA: 14:15:56) 1695g/s 3230p/s 3230c/s 3230C/s brat2728
227500g 0:00:02:56 50.00% (ETA: 14:17:54) 1290g/s 2456p/s 2456c/s 2456C/s geneen1
256269g 0:00:04:21 50.00% (ETA: 14:20:44) 980.8g/s 1869p/s 1869c/s 1869C/s trares75
273837g 0:00:05:19 DONE (2017-02-03 14:17) 857.1g/s 1634p/s 1634c/s 1634C/s kiosk1988
As you can see, there is a blatant slowdown.
If I comment "SingleRetestGuessed = N" in john.conf, I can achieve better c/s & C/s:
54717g 0:00:04:58 50.00% (ETA: 14:56:46) 183.4g/s 351.6p/s 2254Kc/s 2254KC/s lazy209..kateyez02
78115g 0:00:11:01 50.00% (ETA: 15:08:52) 118.1g/s 226.0p/s 2084Kc/s 2084KC/s cynsa4502..torres1307
95610g 0:00:17:05 50.00% (ETA: 15:21:00) 93.25g/s 178.0p/s 2018Kc/s 2018KC/s ghost5..34531218
107053g 0:00:21:49 50.00% (ETA: 15:30:28) 81.76g/s 156.3p/s 1984Kc/s 1984KC/s lovevj4..looking
...
but the real cracking perf is very bad because I test so many hashes with the wrong candidates (same file, easy to compare). And you can see that performance slowly decreases (with SingleRetestGuessed = Y it should increase because the number of unknown hashes decreases).
patpro
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.
