[MUD-Dev] Moore's Law sucks (was: 3D graphics)

[Ling]

I wasn't gonna join coz it didn't seem particularly on topic but I'll try
and turn it round.  Ho hum.

On Sun, 15 Feb 1998, Ben Greear wrote:
> On Fri, 13 Feb 1998, Brandon J. Rickman wrote:
> > On Fri, 13 Feb 1998 16:24:56, Mike Sellers <mike at online-alchemy.com> wrote:
> >
> > Problems with Moore's Law:
> > 
> > "computational power" merely refers to a measure of how many operations
> > a chip can perform in a fixed amount of time.  The higher the MIPS
> 
> Sure, MIPS aren't everything, but Memory, HD, monitors and everthing
> else is getting faster and cheap as well...

No, MIPS aren't everything.  There are dedicated application specific
chips for digital signal processing working on the basis of one
instruction multiple data.  One instruction operates on a chunk of memory. 
Think hardware blitter.  The most recent widespread example being Pentium
MMX. The MMX thing does exactly this.  Same MIPS, more speed. 

> > Designing for an imaginary machine is a gamble.  Some people can afford
> > to make that gamble, and some of them might make a lot of money off of
> > it.  But overall, blindly accepting high-stake risks is not only
> > foolhardy, it is bad business practice.
> 
> I say those who design for today are going to be outdated by the time
> they finish their product.  Design for super high end machines today,
> and by the time yer done, it will be a common machine...

Designing for imaginary machines isn't that much of a gamble in the
computer industry I would have thought.  For computer games, the target
machine keeps moving and you *have* to design for an imaginary machine to
have a competitive product that uses all the features then.  Otherwise
it'll look pale in comparison and get relegated to the budget bins.  It's
not a blind high risk strategy coz it's expected to carry on forever.  (to
my amazment.)

Seeing as we're on muds here, I thought the main worry would be bandwidth. 
Strangely, it costs 750 million pounds to lay a 300,000 simultaneous voice
optic fibre transatlantic.  Working at full capacity with peeps paying for
a pound a minute, it breaks even after 2 days...  Okay, a more realistic
figure is 2 weeks but why aren't there more fibres being laid down?  I
find bandwidth has been quite scarce lately coz more peeps are joining and
downloading silly graphic intensive webpages than the network is
expanding...

Any predictions for this one?  Apart from we're all gonna use the mains
supply to surf the web.  The only reason I'm personally on the net is coz
I'm abusing University resources.  When I leave, I will probably lose
access bar email. (I don't care for the rest of the wiglets.) 

Maybe I should transplant myself to Singapore which is expecting to be the
first country to have optic fibres to every home.  (beating the Japanese.) 
Then again, the limiting factor on optic fibres isn't the fibre, it's the
hardware box on the ends.  It's a very new-esque field, there are no
international standards, the first decade is going to be rough with every
man and his dog inventing a new standard for their local area.  It's
expensive to have redundant outlets for 'future expansion' on fibre
networks and upgrading requires vast portions of the network to go down. 

The technology to make superfast transmitters is quite expensive.  That is
a hurdle worth more than any chip advancements for the next 5 years.

We'll see.

> > Moore's Law has been manipulated by an aggressive advertising campaign.
> > Computers now do less with more.  Productivity has not increased.
> > (Productivity was hardly even measured before computers entered the
> > workplace, so the argument is moot.)
> 
> Some productivity goes up, some goes down or stays the same.  Some do only
> slightly more with more, but I don't think many do less..  I'd like an
> example...

I do less work coz I have telnet access. :)  This is actually quite true
coz I've spent the last 12 hours doing practically nothing. (added 500
whole words to my report.)  Help!  Put me out of my misery!  Is there the
equivalent of a telnet placebo?

> > This all began with:
> > >I had a fascinating discussion with a guy from Intel recently.
> > 
> > Hardly an objective source.  I once heard that VRML was the future of
> > 3D, but I think it was Mark Pesce who said it.  

The above thing from an Intel engineer is vaguely accurate as far as I'm
concerned coz it correlates with what I know about optical chips and
copper based chips (copper based = replacing aluminium wires, not the
silicon substrate).  This has a very weak link to muds.  Most mud clients
require minimal resources to run on, there's nothing to process.  Quake is
probably a little more intensive than a hectic pkmud like Tron.  Total
Annihilation (C&C clone) actually compresses the packets before sending
them out.  Bandwidth, bandwidth, bandwidth!

VRML just looked poxy to me.  Gut instinct which said "It's poo".  I'm
usually right with these new fangled crazes. :P  It's pointless, webpages
are messie enuff as they are.

> I saw the new release where IBM had made it's first 1000mhz chip,
> prototype, but done none-the-less.  Says it should be ready by 2001 or
> so....  Intel and Alpha may beat them to it even...  Even when or if
> they slow down eventually, motherboard and CS in general will enable us to
> stack four of them side by side...or 16, 256...etc.....

Wires/tracks along PCBs stop acting like wires after the megahertz zone
and start behaving like transmission lines and waving around instead of
particling.  I'm surprised chip companies haven't hit a brick wall with
those sort of speeds where any track loops larger than 12.5 cm start
interferring with the clocks...  I have nightmares trying to design
circuits to work with a microprocessor pretending to be an antenna.

> > I'm all for progress in the world of muds, but I think the design
> > criteria, especially for the upcoming generation of graphical
> > muds/UOII/whatever, should be focused on the strengths of what is
> > already successful.
> 
> What fun would that be??  I've got a perfectly fine and brand new text
> MUD, but it's already old by 5 years.  Of course, we on this group will
> probably play them forever, I like text better than graphics, but coding
> more than both I spose.  With that in mind, I'll probably try to write
> a graphical game for Multi users..but it would be dis-respectful to call
> it a MUD...

Graphics ... Hope you can draw more than just stick men.  As for audio, I
have no idea how anyone would go about making sound fx.  Crinkle a crisp
wrapper if you want a fireplace in your mud. :)

Calling it a MUD would possibly be commercially unviable. :P Muds have a
stigma, virtual worlds sound more politically correct and bland.  Players
manipulate avatars, npcs are better known as ICE and people using
keyboards are called turtles.  Welcome to the Matrix! 

  |    Ling Lo of Remora (Top Banana)
_O_O_  Elec Eng Dept, Loughborough University, UK.     kllo at iee.org