Starting point/hash

Discussion and support for the CUDA Multiforcers (Windows and Linux)
  • Ads

Re: Starting point/hash

Postby kevinkk5 » Sat Feb 19, 2011 11:32 am

can i have this php script? or the list?
kevinkk5
 
Posts: 39
Joined: Thu Feb 17, 2011 12:13 pm

Re: Starting point/hash

Postby kevinkk5 » Sat Feb 19, 2011 1:32 pm

before i told you that i would have the best performance with 768 threads and 7 blocks under win7.
i tested the new alpha under linux 64 bit and there i got the best results with 300 threads and 900 blocks. but it doesn vary too much from the normal speed without choosing manually threads and blocks. of course lookup is active. i can get about 20-40 M/s more than in windows but the screen is slower too. the only problem is that under windows 7 i could do it with 768 threads and if i do that under linux i get that ressource error. but great work this beta!
kevinkk5
 
Posts: 39
Joined: Thu Feb 17, 2011 12:13 pm

Re: Starting point/hash

Postby Bitweasil » Sat Feb 19, 2011 2:40 pm

The defaults for the 0.7 series are pretty bad for Fermis. I'm working on improving that.

I'm surprised you're finding best results with a non-power-of-32. Usually threads should be a multiple of 32 (the warp size), and I've found block sizes of multiples of 60 work nicely.

Linux is a better environment for this. The Windows driver model really sucks. The fact that you're getting significantly more speed points to this fact :) The tools are really designed for a headless Linux environment, they just happen to work tolerably with GUIs under Linux, and work with GUIs under Windows.

As for the PHP script... not difficult. :-P Just generate random character strings, run it through the built in md5 function, and output to a file. :-P I don't keep simple stuff like that in version control.
Bitweasil
Site Admin
 
Posts: 912
Joined: Tue Jan 20, 2009 4:26 pm

Re: Starting point/hash

Postby kevinkk5 » Sat Feb 19, 2011 3:06 pm

oh you're right! I somehow forgot to test 320 threads! with that change i got 30 M/s more! quite impressive!
but if i change the charset to numericwith length 14 this configuration is quite slow. if i change it to 256 threads and 300 blocks i get hardly twice M/S! For charset numeric that was the best i could find.
kevinkk5
 
Posts: 39
Joined: Thu Feb 17, 2011 12:13 pm

Previous

Return to CUDA Multiforcers

Who is online

Users browsing this forum: No registered users and 1 guest

cron