At 01:41 1/12/98 -0600, John Flanagan wrote:
>At 12:53 AM 11/30/98 -0600, Greg Teiber wrote:
>>Forget the '040......... the '060 is out :) And I don't see why it would
>be so difficult to just put stock traces and stock chips on a Large Hunk of
>Sillicon. Sorta like the 512k onboard wiht the PPro. I think the hardest
>part woudl be lowereing the power consumption of the '040-'060 series.
>They do requie Heatsinks, even at the 33mhz level. I guess it's time to
>start petitioning Motorolla :)
Actually, I have a feeling the '060 uses less power than the '040. It
certainly is much faster at similar clock speeds and has FPU and MMU on chip.
But the PPC chips are also fairly low power and can emulate 68k series of
chips as well as easily outperforming all the Pentium breeds.
But I would rather do assembly programming for a CISC processor like the
'060 than a RISC one like the PPC. Mind you, many of the good optimising
compilers achieve efficiency approaching a good hand-coder these days, and
with the cost of RAM these days giant compiled projects are not a problem.
>Lotsa problems besides power consumption.
>
>1) A very important consideration for full-fledged CPUs is the number of
>pins available. Every feature that you tack onto the chip is more pins
>that have to come out of the package. This causes more problems than you'd
>think.
Not really. All the pins stay inside the package, except where brought out
for I/O, and if you keep that to serial busses (Comm, IR, USB, Display)
then it would likely have less pins than a conventional chip.
>2) The reason that the PPro never really took off was because it was
>*expensive*. The bigger your die size, the lower your dies-per-wafer, and
>the lower your yield (because larger dies have more room for mistakes).
>Doubling die size will cut dies-per-wafer by *more* than half (due to
>edge-of-wafer lossage), and will also roughly double your yield loss (which
>if you have high yield isn't so bad, but if you have low yield to begin
>with, you're hosed).
Yes, this is a biggie. And you can only miniaturise just so far...
>3) Concentration of heat. All that stuff you throw onto that single chip
>will be producing heat in one place, rather than being scattered around.
>This will aggravate heat problems.
If you use Intel technology then yes it is a problem but if you go with
Motorola or ARM technology then there is little heat to dissipate to start
with. And Transputers or FORTH chips... well they don't even need heat
sinks in the first place. And the old 8/16 bit chips are still very useful
for projects not requiring graphics intensive stuff. Have you ever noticed
that Word98 runs the same speed on the new PentiumII's as it does on old 8
bit 64k byte machines that ran at a humble 1 MHz?
What a great list this is!
My best wishes to all you pioneers,
- Miriam
-----------------------------------------------------------------
My page is at http://werple.net.au/~miriam/
Virtual Reality Association (VRA) Electra City
Melbourne, Australia Multi-user virtual world
http://www.vr.org.au/ http://ariadne.iz.net/~miriam/
--
Subcription/unsubscription/info requests: send e-mail with subject of
"subscribe", "unsubscribe", or "info" to 
Wear-Hard Mailing List Archive (searchable): http://wearables.ml.org
From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty
Archive created with babymail