[rc5] RC5 Chips?

Walter Taninatz waldo at voicenet.com
Sat Nov 1 14:57:18 EST 1997


	Actually, using an FPGA is the IDEAL way to increase a computers capacity
for doing other things. Since a particular algorithm can be implemented
directly in hardware, Many many tasks can be handled efficiently and when a
new problem comes along, that algorithm can be implemented in hardware on
the same FPGA.
	While it is true that the Bovine clients run very efficiently on Intel or
PPC chips, many other tasks will not run so efficiently,

Thanks,
Walt.


At 12:59 PM 11/1/97 -0500, you wrote:
>There is ONE substancial argument against FPGAs. According to the below
>mentioned web page, a FPGA will do approx 20Kkey/sec /chip at a cost of
>$15.00. You will still have a relatively small run of PCI  (or ISA for that
>matter) boards which will drive it's price up to $150-200 apiece (how many
>people are going to spend GOOD money on a board that can only do one thing
>(which  realisticly will never make them any money, or improve the
>performane of their computer for other tasks) So take that PCI card, drop 20
>FPGAs on it, and you have??? 400Kkey/sec for a cost of around $500. ( I just
>went to ComPUKEsa and bought 2 used Performa 6400/180s which do 500Kk/sec
>for $540 each, and they can be used for other things which WILL make me
>money) 
>So you would need HOW MANY of these FPGA cards in HOW BIG a case with WHAT
>SIZE refrigeration unit before you actually gained some real speed. My
>calculations show you'd need 330 chips to do 6600Kkeys a second. That's 16.5
>boards!  for a cost of $8250. (not counting the computer, case and afore
>mentioned refrigerator.) 
>For the PowerPC I am only suggesting as much code as needed (burned to an
>onboard ROM) to handle reading a possible combination out of a ram chip,
>doing the calculation, and returning the result to the ram chip. THE bovine
>client on the master computer would handle all netwoking using the real
>computer's functionality. 
>
>Your turn...
>The Great BrianB
>
>----------------------------------------------
>Protect your right to own and arm bears.
>----------------------------------------------
>
>----------
>From: Michael Conwell <mbcwam at anet-stl.com> 
>To: rc5 at llamas.net 
>Subject: Re: [rc5] RC5 Chips?
>Date: Sat, 1 Nov 1997 11:25:19 -0600 
>
>Using mult-purpose CPUs would be dumb.  First you'd have to Fab a board.
>Plus you'd need a OS that can multitask among 12 chips...AIX can handle
>that, the MacOS needs ASICs and ROMS and currently doesn't handle
>muliprocessors, and I don't think Linux will handle 12 processors either.
>Speaking of an OS, I guess you'd have to license it or buy a copy...another
>problem.  Writing your own would be kind of hard.  FPGAs take most of that
>out of the equation.
>
>>Aren't you wasting your time? I don't mean this in an offensive way, but el
>>cheapo PowerPC chips crank out keys checked... A 200Mhz 603e chip can do
>>550Kk/sec. These chips cost around $150? Why not build a PCI board with 12
>>empty sockets for $100-200. Then add PowerPC chips as needed. A full board
>>would do 6600Kk/sec for $2000, a starter board would do 1100Kk/sec for
>$500,
>>with an extra 550Kk/sec for every $150 spent. Plus, You could probably
>>reprogram the 603e chips to do different things later.
>>
>>
>>The Great BrianB
>>
>>----------------------------------------------
>>Protect your right to own and arm bears.
>>----------------------------------------------
>>
>>----------
>>From: Dave Zarzycki <zarzycki at ricochet.net>
>>To: <rc5 at llamas.net>
>>Subject: FPGAs and distributed computing (Was: Re: [rc5] RC5 Chips?)
>>Date: Fri, 31 Oct 97 21:12:26 -0800
>>
>>On 10/31/97 7:40 PM, Ryan Dumperth (woodie at indy.net) wrote:
>>
>>>http://www-inst.eecs.berkeley.edu/~barrel/rc5.html
>>
>>Yea!
>>
>>Okay, distributed computing is not going away as far as I can tell. Let's
>>make the best of it! :-) I'll willing to chip in a few bucks to buy a
>>FPGA card and let distributed.net have 100% access to it. By doing that,
>>v3 clients could report that a FPGA was available, get the corresponding
>>code, and have at it. In another way, it's great for distributed.net in
>>the sense that it shows commitment. It also shows distributed.net that a
>>client might be more trusted when it comes to computing a result *and
>>returning it* the servers (perhaps in a timely manor too)... Future
>>projects might heavily benefit from FPGAs too.
>>
>>davez
>>
>>----------------------------------------------------------------------
>>Dave Zarzycki                                Student
>>Intern                                       San Jose State University
>>Apple Computer, Inc.                         dzarzyck at email.sjsu.edu
>>zarzycki at apple.com                           zarzycki at ricochet.net
>>----------------------------------------------------------------------
>>PGP Fingerprints (RSA):  8AF2 1040 8A9C D025  47BE 70DD A51C C887
>>DSS/Diffie-Hellman: CB9E 2621 B4BA 3F96 3516  B312 15B4 D842 3809 EF99
>>Contact pgpkeys.mit.edu for my public keys.
>>
>>----
>>To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5'
>in
>>the body.
>>
>>
>>----
>>To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5'
>>in the body.
>
>
>
>----
>To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in
>the body.
>
>
>----
>To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5'
in the body.
>
>
  Walt Taninatz
 -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  
 mailto:waldo at voicenet.com        http://www.voicenet.com/~waldo
 -----------------------------------------------------------------
|     export-a-crypto-system-sig RSA-3-lines-PERL                 |
|     #!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj   |
|     $/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1   |
|     lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)     |
 -----------------------------------------------------------------
        Help Crack RC5! ---->  http://www.distributed.net/rc5/
----

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list