Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <55E9691A.60402@mailbox.org>
Date: Fri, 4 Sep 2015 11:49:14 +0200
From: Frank Dittrich <frank.dittrich@...lbox.org>
To: john-dev@...ts.openwall.com
Subject: Re: 23% performance regression for brypt (Intel i5-4570
 CPU)

On 09/03/2015 07:59 PM, magnum wrote:
> Some more tests on various gear:
> 
> i7-4790 (core i7 Haswell, AVX2) gcc 4.8.2:
> Huge boost for X3 except when running HT (slight loss)


BTW, on well (i7-4770K CPU @ 3.50GHz), the best combination for combined
c/s rate is to run 6 processes with BF_X2 = 1 and 2 processes with BF_X2
= 3.
This resulted in a combined c/s rate of 6912, while 8* BF_X2 = 1 just
got 6764.
(I ran --test=120 with 8 separate processes.)

I guess for more processes with BF_X2 = 3, the limited cache could be
the problem.

Of course, combining processes with different max. keys per crypt makes
defining an optimal attack harder.

Frank

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.