[plug] john the ripper

[W.Kenworthy]

Obviously -test isnt quite the same as working on a real file.
Interesting how much more efficient the athlon is on most cyphers -
keeping in mind that john on the P4 was compiled with gcc 3.3.5, while
the athlon used the faster (by reputation) gcc-3.4.4.  Yet the P4 wins
on standard DES.  The P3 667 at the end is included for novelty
value ...

athlon-xp 2500+ (1900Mhz)
rattus ~ # john -test
Benchmarking: Standard DES [48/64 4K]... DONE
Many salts:     127411 c/s real, 134117 c/s virtual
Only one salt:  122675 c/s real, 130505 c/s virtual

Benchmarking: BSDI DES (x725) [48/64 4K]... DONE
Many salts:     4464 c/s real, 4679 c/s virtual
Only one salt:  4358 c/s real, 4626 c/s virtual

Benchmarking: FreeBSD MD5 [32/32]... DONE
Raw:    4739 c/s real, 5020 c/s virtual

Benchmarking: OpenBSD Blowfish (x32) [32/32]... DONE
Raw:    285 c/s real, 300 c/s virtual

Benchmarking: Kerberos AFS DES [48/64 4K]... DONE
Short:  120012 c/s real, 127945 c/s virtual
Long:   449149 c/s real, 471748 c/s virtual

Benchmarking: NT LM DES [48/64 4K]... DONE
Raw:    1549401 c/s real, 1644800 c/s virtual


P4M 2.2Ghz (No HT, laptop)
bunyip ~ # john -test
Benchmarking: Standard DES [48/64 4K]... DONE
Many salts:     144716 c/s real, 147972 c/s virtual
Only one salt:  136883 c/s real, 139676 c/s virtual

Benchmarking: BSDI DES (x725) [48/64 4K]... DONE
Many salts:     4761 c/s real, 4908 c/s virtual
Only one salt:  4854 c/s real, 4963 c/s virtual

Benchmarking: FreeBSD MD5 [32/32]... DONE
Raw:    3633 c/s real, 3738 c/s virtual

Benchmarking: OpenBSD Blowfish (x32) [32/32]... DONE
Raw:    276 c/s real, 282 c/s virtual

Benchmarking: Kerberos AFS DES [48/64 4K]... DONE
Short:  130350 c/s real, 134650 c/s virtual
Long:   324403 c/s real, 330349 c/s virtual

Benchmarking: NT LM DES [48/64 4K]... DONE
Raw:    1208998 c/s real, 1283437 c/s virtual


P3 coppermine 667Mhz
moriah bind # john -test
Benchmarking: Standard DES [48/64 4K]... DONE
Many salts:     92236 c/s real, 94504 c/s virtual
Only one salt:  83251 c/s real, 86181 c/s virtual

Benchmarking: BSDI DES (x725) [48/64 4K]... DONE
Many salts:     2425 c/s real, 2516 c/s virtual
Only one salt:  2163 c/s real, 2221 c/s virtual

Benchmarking: FreeBSD MD5 [32/32]... DONE
Raw:    1564 c/s real, 1623 c/s virtual

Benchmarking: OpenBSD Blowfish (x32) [32/32]... DONE
Raw:    78.4 c/s real, 93.1 c/s virtual

Benchmarking: Kerberos AFS DES [48/64 4K]... DONE
Short:  56832 c/s real, 86370 c/s virtual
Long:   185344 c/s real, 241963 c/s virtual

Benchmarking: NT LM DES [48/64 4K]... DONE
Raw:    648409 c/s real, 665718 c/s virtual


On Wed, 2005-07-06 at 22:50 +0800, Cameron Patrick wrote:
> Matt Kemner wrote:
> 
> > > running without nice gives a solid 2916 c/s each time
> > 
> > Interesting that you only get 50% better speed than this old PIII
> 
> This got me curious so I ran john -test on a few machines at home.
> 
>                                  MD5 c/s     DES c/s
>  Wizard (233MHz Geode)               246        8737
>  Pentium M 1500MHz                  3645      169062
>  Athlon XP 2600 (Barton, 1900MHz)   5427      146969
>  Athlon XP 2600 (T'bred, 2100MHz)   5966      160998
>  Athlon 64 3500 (Venice, 2200MHz)   4374      477703
>  Athlon 64 3500, 32-bit john        6478      171289
> 
> And for giggles, the PLUG servers:
>  Spark (Sparcstation 20 clone)       110        8312
>  Shameless (1GHz Via C3 Nehemiah)   1402       42956
> 
> (The 32-bit john contains MMX assembler optimisations whereas the
> 64-bit one is straight C code generated by gcc.  Interesting that DES
> really loves the 64-bit code, though.  I thought it might be some
> screwed up outlier so ran it again and got an even higher value.
> Odd.)
> 
> Those of you with variable-width fonts will find this table more
> entertaining to interpret :-)
> 
> Cameron.
> 
> _______________________________________________
> PLUG discussion list: plug at plug.org.au
> http://www.plug.org.au/mailman/listinfo/plug
> Committee e-mail: committee at plug.linux.org.au
>