[rescue] Re: rescue digest, Vol 1 #1783 - 14 msgs

[joshua d boyd]

On Fri, Aug 03, 2001 at 11:55:10AM -0400, Mike Dombrowski wrote:
> Um, why exactly are two cpus needed? I can do all of this fine on one 
> cpu with room to spare. I'm not debating that dual cpus rock, this is a 
> dual xeon, but for home use two CPUS are silly from many different 
> views. And with most windows games not being multithreaded they become 
> even less attractive.

You assume that most PC owners game.  Most PC owners I know (my family for
the most part, pretty much all my friends) don't game.  So say that games
don't support something as a reason why we should or shouldn't have it is
kinda silly.

I don't have the cold hard numbers here, but what I can say is that I find
that my dual 350 is more responsive than the gigahertz machines I've sat
at, even if it isn't always "faster".  Further, it tasks (like MP3
playing, DVD playing, etc) don't get interupted as much as the seem to on
"faster" machines.
 
> Um, I don't think Be did 4 cpu machines. I'm decently active in the Be 
> community and know a couple engineers there. I've also been following 
> Be since they announced the BeBox, I've never seen anyone mention a 
> 4cpu Be Inc machine. 

I thought they had prototypes that they were showing off.  Could be I'm
wrong though.

> They gave up because at the time there was no 
> cheap MP box, when they stopped x86 MP was cheap enough for a common 
> person to buy. Also, like you mentioned previously the hardware was 
> pretty damn slow and it would have drained Be's resources far too much 
> to design a new system.
 
I don't know.  I'm not an expert, but I still think it was a bad
move.  People were willing to buy the BeBoxes, but they seem less willing
to buy special commodity hardware for it.  But who knows.  Despite claims
otherwise, hindsight is rarely 20/20 either.

> I simply pointed out that the company that the orignal poster held up 
> as an example of how to design hardware did the same things as what he 
> admonished in the next sentence. I know why Sun, NeXT and others did 
> it, it makes perfect sense. The cost savings of the windmodem might 
> easily allow an increase in CPU speed which more than makes up for the 
> little processing power needed by the winmodem. The end result is a 
> faster computer for the same cost, is this bad?

In the case of WinModems (and WinPrinters, etc) it seems that the machines
are slowed down far out of proportion to how much CPU power the devices
actually need.  The faster the machine is, the more CPU power it needs to
drive the exact same printer it seems.  I'm told this has something to do
with polling and improper interupt useage, but I don't really know.

-- 
Joshua D. Boyd