5 Mar
1999
5 Mar
'99
2:58 p.m.
Eric Smith wrote:
Again, I was thinking of the reconfigurability of the 34010.
Ah. Well, there goes my idea of havng an "uber-graphics board" that does
everything... As for emulating the older cards, I have a _Micro Cornucopia_
issue which describes a board that promises exactly that. There's some
external logic, but I doubt there are any ASICs.
Derek wrote:
I should have used a different phrase. Isn't it possible to have the 34010
interpret the data in (data) memory in various ways, based on the code you
put in (code) memory? That's what I meant by "flexible".
Exactly why isn't it VGA-compatible?
The 34010 isn't even *similar* to a VGA. The exact extent of the
compatability is that both hook up to a host processor, and both produce
a video output.
I thought that certain things are true of the
34010:
- flexible memory architecture
If a large flat address space is a "flexible memory architecture",
then the 34010 certainly has it. However, that isn't what VGA uses. VGA
has all sorts of crufty bank switching and other disgusting hacks. - many
operations done in software (not hardware)
Actually most of the interesting operations of the 34010 are done in
microcode. But those operations aren't similar to VGA. It's
certainly possible to emulate the Hercules, CGA, and MDA. Is there
some lack of configuration registers, or some internal timing constraint,
that makes emuilating the VGA impossible?
The 34010 doesn't have any built-in support for emulating any of those things.
Sure, you could do it by adding a big pile of external logic (which you could
put into an ASIC, or some PLDs). Due to communication overhead, it would
be a fair bit slower to emulate a VGA with a 34010. It's cheaper and easier to just put a VGA
controller on the card, since that
does exactly what you need in a single inexpensive chip. Leave the 34010
to doing what it does well. For example, acting as a display list processor
for AutoCAD. The VGA interface is too low-level and baroque to be
efficiently emulated in software.
Hmph. I should have guessed. You might have already figured out that I'm
always thinking of the design of my Perfect Computer (tm). The problem is:
The IBM PC has software, the Perfect Computer doesn't. I'd either have to
write all the software myself, port it from the IBM -- but the point was
to *avoid* all the cruftiness of the IBM -- or write a PC emulator. Ick.
Of course I could get the software from somewhere else.
Also, I fear that any design I could come up with will never be as fast as
any current PowerPC or Intel systems. Those hacks do serve a purpose!
Unfortunately, PC software is successful *in spite of* the nature of the PC,
not *because* of it. :(
What are the chances of successfully emulating EGA?
-- Derek