[rc5] RC5 Chips?

Darxus darxus at Op.Net
Sun Nov 2 11:08:13 EST 1997

On Sun, 2 Nov 1997, Joseph Fisk wrote:

> On Sat, 1 Nov 1997, Darxus wrote:
> > On Sat, 1 Nov 1997, Brian B. Burton wrote:
> > 
> > > I'm all for experimentation, and reusing free stuff. (I have 5 20Mhz 68040's
> > > cranking out 20Kkeys/sec myself!) BUT, I am just the kind of computer nut
> > > who WILL spend $1000-1500-maybe $2000 to get a board that I could stuff in a
> > > PCI slot that gives me 10,000Kk/sec. (even if it can't do anything else.) 
> > 
> > I'm not interested in a card that will *only* do RC5 keys, but I *am*
> > willing to put $$$ into a PCI card full of chips dedicated to
> > distributed.net's V3 protocol, when it happens.  I don't care if it's a
> > card full of FPGAs, or some kind of commonly mass-produced CPUs with
> > specially written low-level code.  whichever can do the most crunching for
> > the $$$.
> > 
> > And I'm guessing me & Brian aren't the only ones. 
> Well, I'm certainly interested in the idea but I definitely would not pay
> hundreds of dollars for a dedicated card.  It provides no tangible benefit
> to _me_, so why should I spend _my_ money?  I'm more than happy to go out
> of my way to recruit as many CPUs as possible, and I love the competition.

I'm not talking about a card that is dedicated only to The Network, I'm
talking about a card that is capable of being dedicated to a specific
process in general.

RC5, Mersine Primes, Chess, oh, 3D rendering (Quake2...)... hmm, just
about CPU intensive process that's worth the time to write low-level code

so when you're not playing Quake, or something else you'd use this card
for, you crunch at much higher speeds...  (and while you're playing Quake,
you get effects similar to using a 3DFX Monster... ?)

> Anyway, doesn't making a dedicated multi-CPU card take us above the level
> of a "loosely organized group of hackers using their idle CPU time"?

Yup.  You consider that a bad thing ?

V3 itself will be beyond that level, it will be a protocol that can be
installed on every computer in existance, including those owned/run by the
clueless, and after being installed, will never need to be touched again
(unless you want to upgrade, and I'm assuming there will be full backwards
compatibility, so that wouldn't be absolutely manditory).

A netork comprised of *all* idle CPU cycles is in the process of being
developed.  developing an expansion card wich is relatively inexpensive
that a few hundred people would be interested in buying would increase teh
capacity of such a network.

actually, it'd be more fun to buy a blank circuit board & a couple chips,
download a diagram, and start soldering... :)

> BTW the card doesn't need to be PCI.. the I/O requirement is minimal.
> Just send the board a start address (the block number) and tell it to do
> 2^28 iterations.

I dunno, if I were to buy(/make) a card designed to do intensive
processing, I think I'd want to leave open the option of using it for
something that requires intensive I/O...

remember that Gates once said that 640k would be enough for anyone....
***PGP fingerprint = D5 EB F8 E7 64 55 CF 91  C2 4F E0 4D 18 B6 7C 27***
               darxus at op.net / http://www.op.net/~darxus 
         "You shall know the truth, and it shall make you odd."

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

More information about the rc5 mailing list