is anyone CUDA optimized version of linux crypt(3) function?

is anyone CUDA optimized version of linux crypt(3) function?

i use this: http://openpaste.org/en/18965/ variant but it to slow

There is this version for DES?

http://s-schwarzwaldhacker.rbcmail.ru/schwarzwaldDES.html

Unfortunately it won’t help if you need it for the school, because there is not the sourcecode? :confused:

need source

using shared memory for DEC constants give only 40% speed up

mem_size = 8389760

Processing time dev : 2473.954102 (ms)

Processing time dev copy : 6.598325 (ms)

Processing time host: 5963.502441 (ms)

Total Errors = 0

Press ENTER to exit...

CUDA on GeForce 8800 640MB suck from my CLERON CORE DUO 2333 on this task

help me please to optimise http://openpaste.org/en/18965/
very very slow working on CUDA