1995-08-24 - Re: Brute SSL Challenge

Header Data

From: Piete Brooks <Piete.Brooks@cl.cam.ac.uk>
To: David Neal <dneal@usis.com>
Message Hash: 35aa8b5e9a5eb647aaee6411f2c6f8f4e9dc9886dcaf8d21b642690d6fe71023
Message ID: <“swan.cl.cam.:169220:950824070545”@cl.cam.ac.uk>
Reply To: <199508240552.AAA00601@gnupln8.usis.com>
UTC Datetime: 1995-08-24 07:06:01 UTC
Raw Date: Thu, 24 Aug 95 00:06:01 PDT

Raw message

From: Piete Brooks <Piete.Brooks@cl.cam.ac.uk>
Date: Thu, 24 Aug 95 00:06:01 PDT
To: David Neal <dneal@usis.com>
Subject: Re: Brute SSL Challenge
In-Reply-To: <199508240552.AAA00601@gnupln8.usis.com>
Message-ID: <"swan.cl.cam.:169220:950824070545"@cl.cam.ac.uk>
MIME-Version: 1.0
Content-Type: text/plain


> Hopefully everyone on here is accomplished enough to know
> optimizing the brutessl code helps immensely.  If not, well I used
> 
> gcc -O6 -funroll-loops -fomit-frame-pointer -finline-functions -c search.c
> gcc -O6 -funroll-loops -fomit-frame-pointer -finline-functions -o brutessl \
>                      brutessl.c search.o
> 
> To go from 5,000 keys per second to 10,100 keys per second.

If people could send me timings for various compilers / flags I'll collate a
table of speeds [see brutessl.h 1.02 for an example -- I tried using different
sizes for the RC4 info and it appears that only ALPHAs really gain by using int]

I recommend that you do not blindly use the above flags -- I just tried it on
our fastest machines (SGIs) and it *REDUCED* the speed from 35200 to 28400.





Thread