r206 and r208 attack_cowpatty decrease performaces versus Pyrit in BackTrak4

[GoogleCodeExporter]

What steps will reproduce the problem?
1. boot backtrack4 live dvd, mount hard disk partition (sda2) with inside
my cowpatty exported file, do command "time pyrit -e ESSID -r HANDSHAKE.cap
-i FILE.cow attack_cowpatty": filed scanned in 7min 5 sec (it means ~575000
PSKs/s)

2. boot my hard disk (sda1), mount hard disk partition (sda2) with inside
my cowpatty exported file, do command "time pyrit -e ESSID -r HANDSHAKE.cap
-i FILE.cow attack_cowpatty": filed scanned in 26min 0 sec (it means
~156500 PSKs/s)

3.
Note that password was not present in cowpatty file, so the file was
scanned from beginning to the end in both tests.

What is the expected output? What do you see instead?
Aspected output is that pyrit r206 and r208 do the same or better PSKs/s
than pyrit version in BT4, instead is the opposite.

What version of the product are you using? On what operating system?
on backtrack4 live DVD I don't remember the pyrit version, on my hard disk
I am running r206, CUDA and NVIDIA 9800GTX+, OS is Debian "lenny" 5.0.3, I
compiled pyrit and cpyrit_cuda with gcc-4.1 (using command export CC='gcc-4.1')

Please provide any additional information below.
using backtrack4, I run attack_cowpatty in console #1 and in console #2 I
run 'top' that report my 4 cores are at 99.9% of load: the same test with
r206 on my hard disk teport that only one core is at 92-95%, 2 core about
at 17-19% and one core at 0% of load.

Pyrit r206 was got form svn some days ago, so I suppose some problem in
r206 and I repeated the test also with r208 I downloaded 20 minutes ago,
but results are the same.

Original issue reported on code.google.com by pyrit.lo...@gmail.com on 25 Jan 2010 at 8:17

[GoogleCodeExporter]

A clarification: r208 does the job in 25min 3sec (~162400 PSKs/s), so results 
are not
"the same" but "little better than" .

Original comment by pyrit.lo...@gmail.com on 25 Jan 2010 at 8:45

[GoogleCodeExporter]

which version is on backtrack4-final (the one that does 575.000 keys/s) ?

Original comment by lukas.l...@gmail.com on 29 Jan 2010 at 11:07

[GoogleCodeExporter]

hi lukas.
I duno what version of pyrit is inside BT4 because when i run it, i see "pyrit
0.2.5-svn" but i dont see the right version. I contacted people of bactrack4 in 
their
irc channel but no info they was able to give me: more, they say it does not 
exist
the list of all software inside BT4 and exact version of those.
I can tell you that pyrit in BT4 accept '-i' and '-o' option, maybe this can 
give you
a suggestion related to how recent that pyrit is.
If there is some way to get the right version, (i.e. read the hex in binary 
mode of
pyrit) please tell me and I will do.
Last option is you get one ISO/DVD of BT4 and investigate directly. 

Original comment by pyrit.lo...@gmail.com on 30 Jan 2010 at 12:10

[GoogleCodeExporter]

I've identified the problem and will update svn later this weekend.

Original comment by lukas.l...@gmail.com on 30 Jan 2010 at 3:23

• Changed state: Started

[GoogleCodeExporter]

Is #103 related to #65?
I mean, after you fix both, have I to aspect to have back 575.000 key/s on my 
system
or shall I aspect better as planned in #65? (575.000 x 1,7 ~ 977.000).

Original comment by pyrit.lo...@gmail.com on 30 Jan 2010 at 6:09

[GoogleCodeExporter]

No it's not related.

The SSE2-path is about 3.3 times faster on my system. Please remember that you 
may
hit a performance-wall regarding your hard-drive: 1.000.000 PMKs per second = 
(32
bytes for the PMK + ~6 bytes for the password)*1.000.000 = 36mb of data to be 
read
from the disk and being processed by python.

Original comment by lukas.l...@gmail.com on 30 Jan 2010 at 6:16

[GoogleCodeExporter]

fixed in r209

Original comment by lukas.l...@gmail.com on 30 Jan 2010 at 7:05

• Changed state: Fixed