[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 12 Aug 2012 01:52:13 +0200
From: Lukas Odzioba <lukas.odzioba@...il.com>
To: john-dev@...ts.openwall.com
Subject: Re: Current -fixes GPU formats vs TS
2012/8/12 Solar Designer <solar@...nwall.com>:
> Lukas -
>> Why do you #include "sha2.h" and how is this going to work in the -fixes
> branch?
That was unintentional, and the final patch wasn't really final.
> The CUDA and OpenCL implementations are now inconsistent in where the
> comparisons are done.
Is it a problem?
> Did you do your speed testing on your cards only or also on bull's?
> I think there may be a bit more of a difference (maybe 1% or so?) with
> the faster GPUs.
Now I did tests:
comparision on gpu:
Raw: 135929 c/s real, 134736 c/s virtual
Raw: 134736 c/s real, 135929 c/s virtual
Raw: 135929 c/s real, 134736 c/s virtual
Raw: 135929 c/s real, 134736 c/s virtual
Raw: 135929 c/s real, 134736 c/s virtual
Raw: 135929 c/s real, 134736 c/s virtual
Raw: 134736 c/s real, 135929 c/s virtual
comparision on cpu:
Raw: 134736 c/s real, 134736 c/s virtual
Raw: 133565 c/s real, 134736 c/s virtual
Raw: 133565 c/s real, 134736 c/s virtual
Raw: 133565 c/s real, 134736 c/s virtual
Raw: 134736 c/s real, 133565 c/s virtual
Raw: 134736 c/s real, 134736 c/s virtual
Raw: 134736 c/s real, 133565 c/s virtual
>>> 135929*100/133565.0
101.76992475573691
>>> 135929*100/134736.0
100.88543522147013
It seems that this way is a bit faster (or it is not slower). I guess
that we want to gpu version, so I'll make one more patch ;P
Lukas
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.
