Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Sat, 23 Jun 2012 23:11:09 +0200
From: Frank Dittrich <frank_dittrich@...mail.com>
To: john-dev@...ts.openwall.com
Subject: Some relbench results for linux-x86-sse2 (jumbo5 vs. upcoming jumbo6)

32bit linux might not be the most relevant platform, but it happens to
be the platform where I do have my local git repository, and my
benchmark-unify and relbench patches posted on john-dev are not yet
available on github.

So, I'll start with 32bit Linux:

$ ./relbench j5-sse2.txt j6-sse2.txt
More than one benchmark for MySQL:Raw in file 2
More than one benchmark for NT MD4:Raw in file 2
More than one benchmark for Raw SHA-1:Raw in file 2
Only in file 1: Oracle 11g:Only one salt
Only in file 1: MS Kerberos 5 AS-REQ Pre-Auth:Many salts
Only in file 1: Lotus5:Raw
...
Only in file 2: DragonFly BSD $3$ SHA-256 w/ bug, 64-bit:Many salts
Only in file 2: phpass MD5 ($P$9):Raw
Only in file 2: DragonFly BSD $4$ SHA-512 w/ bugs, 64-bit:Only one salt
Only in file 2: dynamic_33: md4(unicode($p)):Raw
Converting the two benchmark files using benchmark-unify might
increase the number of benchmarks which can be compared
Number of benchmarks:		71
Minimum:			0.90689 real, 0.89791 virtual
Maximum:			7.46817 real, 7.54286 virtual
Median:				0.99639 real, 0.99761 virtual
Median absolute deviation:	0.01557 real, 0.01685 virtual
Geometric mean:			1.06893 real, 1.06820 virtual
Geometric standard deviation:	1.34209 real, 1.34231 virtual

$ ./benchmark-unify < j5-sse2.txt > j5-sse2-mod.txt

$ ./relbench j5-sse2-mod.txt j6-sse2.txt
$ ./relbench j5-sse2-mod.txt j6-sse2.txt
More than one benchmark for MySQL:Raw in file 2
More than one benchmark for NT MD4:Raw in file 2
More than one benchmark for Raw SHA-1:Raw in file 2
Only in file 1: PHPS md5(md5($pass).$salt):Many salts
Only in file 1: DIGEST-MD5 C/R:Many salts
Only in file 1: MYSQL:Raw
...
Only in file 2: DragonFly BSD $3$ SHA-256 w/ bug, 64-bit:Many salts
Only in file 2: DragonFly BSD $4$ SHA-512 w/ bugs, 64-bit:Only one salt
Only in file 2: dynamic_33: md4(unicode($p)):Raw
Converting the two benchmark files using benchmark-unify might
increase the number of benchmarks which can be compared
Number of benchmarks:		149
Minimum:			0.61111 real, 0.61111 virtual
Maximum:			7.46817 real, 7.54286 virtual
Median:				0.99610 real, 0.99350 virtual
Median absolute deviation:	0.01832 real, 0.01924 virtual
Geometric mean:			1.07510 real, 1.07455 virtual
Geometric standard deviation:	1.33192 real, 1.33249 virtual

These are the top 10 improvements:

$ ./relbench -v j5-sse2-mod.txt j6-sse2.txt|grep "^Ratio:"|cut -f
2-|sort -nr|head -n 10
7.46817 real, 7.54286 virtual	Tripcode DES:Raw
3.07227 real, 3.10298 virtual	Oracle 11g SHA-1:Many salts
2.95777 real, 2.95777 virtual	Oracle 11g SHA-1:Only one salt
2.61595 real, 2.58991 virtual	Mac OS X 10.4 - 10.6 salted SHA-1:Many salts
2.56524 real, 2.59114 virtual	Invision Power Board 2.x salted MD5:Only
one salt
2.37834 real, 2.35498 virtual	Mac OS X 10.4 - 10.6 salted SHA-1:Only one
salt
2.32064 real, 2.32064 virtual	SAP CODVN B (BCODE):Only one salt
2.24781 real, 2.22533 virtual	SAP CODVN B (BCODE):Many salts
1.79418 real, 1.79418 virtual	md5(unicode($p)):Raw
1.78052 real, 1.76288 virtual	SAP CODVN F/G (PASSCODE):Only one salt

And these are the worst regressions:
$ ./relbench -v j5-sse2-mod.txt j6-sse2.txt|grep "^Ratio:"|cut -f
2-|sort -n|head -n 10
0.61111 real, 0.61111 virtual	MediaWiki md5($s.'-'.md5($p)):Many salts
0.87012 real, 0.87891 virtual	phpass MD5 ($P$9):Raw
0.88187 real, 0.87305 virtual	dynamic_17: phpass ($P$ or $H$):Raw
0.88684 real, 0.89589 virtual	MySQL:Raw
0.90689 real, 0.89791 virtual	dynamic_28: Apache MD5:Raw
0.90846 real, 0.90857 virtual	dynamic_27: FreeBSD MD5:Raw
0.91026 real, 0.91026 virtual	Generic salted MD4:Many salts
0.92756 real, 0.92756 virtual	dynamic_0: md5($p) (raw-md5):Raw
0.92910 real, 0.92910 virtual	dynamic_22: md5(sha1($p)):Raw
0.93248 real, 0.93248 virtual	Netscreen MD5:Raw

Frank

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.