From: Eric Young <eay@mincom.oz.au>
To: cypherpunks@toad.com
Message Hash: 89f2a2bdb34f796ce3c3a0e875cf55692f39ee5a86f20c483607e9a6c4b41a04
Message ID: <Pine.SOL.3.91.950822091322.17350B-100000@orb>
Reply To: N/A
UTC Datetime: 1995-08-22 00:27:59 UTC
Raw Date: Mon, 21 Aug 95 17:27:59 PDT
From: Eric Young <eay@mincom.oz.au>
Date: Mon, 21 Aug 95 17:27:59 PDT
To: cypherpunks@toad.com
Subject: Doing the SSL challenge
Message-ID: <Pine.SOL.3.91.950822091322.17350B-100000@orb>
MIME-Version: 1.0
Content-Type: text/plain
Boy, go away for a weeks holidays and every-one else has all the fun :-).
Just a few quick ramble about my searching of the top half of the key
space in the first SSL challenge, please ignore if you are sick of this
thread.
All these estimates of the cost in CPU time are a bit silly. I started
out using a few of our bigger machines at work but when people noticed I
just stoped using them. My main workhorse machines
became desktop boxes that nobody used at night. Quite a bit of my
%50 keyspace was done by 5 dual processor sparc 20's that people around
me had on their desktops. Most of the rest was done by about 50 486's
that I rsh'ed to each night (I just tried to rsh to every Solaris/x86 box
in the company :-). Out of a company of about 300 people, about
2 (who did not know about my attempt on the challenge) asked me about this
processing and what was I actually doing, no-one else really noticed :-).
I was making no particular effort to use every machine I could but I was
still able to do about %2.5 of the key space each 12 hour night and %12.5
each weekend. If I made a real effort to harness the compute power at my
work I would have probably been able to run at twice this rate but that
would have required work on my part, alot more people would have noticed
and it would have had an impact on people doing real work. The way I was
seaching had no impact.
I will also say that screen-savers running on the above mentioned 486's
were a real pain, perhaps we will change xlock so they can only run the
blank screen :-).
I personally feel that using mas-pars is cheating a bit :-). People are
aware that something is going on when a machine that big is grunting away
all weekend :-). There is so much CPU sitting on people's desktops that
is just unused that there is very little need to use the big central boxes.
The relative speed on some of the machines I used is as follows (per CPU
is in brackets)
6 CPU SGI challenge 37 (6.1) 5 (1.6)
2 CPU sparc 20 7 (3.5) 1
1 CPU 486 DX50 1
I could never realy get more that 1/2 the SGI
but I could always get 5 sparc 20's and 50+ 486's without anyone noticing.
So I was getting the 2.5 6CPU SGI challenge's from idle machines on
people desks.
eric
--
Eric Young | Signature removed since it was generating
AARNet: eay@mincom.oz.au | more followups that the message contents :-)
Return to August 1995
Return to “Eric Young <eay@mincom.oz.au>”
1995-08-22 (Mon, 21 Aug 95 17:27:59 PDT) - Doing the SSL challenge - Eric Young <eay@mincom.oz.au>