[comp.sys.amiga.hardware] Cost of color

daveh@cbmvax.commodore.com (Dave Haynie) (04/10/90)

In article <5975@tekig5.PEN.TEK.COM> wayneck@tekig5.PEN.TEK.COM (Wayne C Knapp) writes:
>In article <10581@cbmvax.commodore.com>, mks@cbmvax.commodore.com (Michael Sinz - CATS) writes:

>It wouldn't be for everyone, that's for sure.  However, I think that the
>fact that it isn't easy to do on the Amiga does hurt the Amiga's image.
>Having a card like the Hercules card that ran with the Amiga wouldn't 
>hurt Amiga sales.  Plus, if it didn't require a bridge board and supported
>intuition it would be economical and very attractive to Amiga developers. 

The problem isn't the hardware; such a card would at least as easy to design
for the Amiga bus.  The problem is the software, and the question of where the
software display intelligence resides.  In most of the MS-DOS world today (of
course I know this is changing, however slowly, with things like Windows and
TIGA) the intelligence for driving a particular graphics card is contained in 
each application program.  Each program must implement it's own graphics, 
window, etc. drivers.  That's kind of evil, but it does make the choice of 
support for graphics cards up to the program designer.	It's also why most of 
the cool graphics hardware contains hardware or firmware emulations of some
of the lesser graphics systems on the market.  

The Amiga, on the other hand, implements a graphics library which contains 
most of the display intelligence for any application program.  Ideally, all
the device dependent graphics operations would go through graphics, and I
suppose at some point they will.  There are a considerable number of calls
in the graphics library, far more than any simple MS-DOS based application
would use it its program-resident graphics driver.  Which currently makes 
it a real piece of work for any company to build a graphics board that can
run the average Amiga program.  Nothing's stoping anyone from taking the
MS-DOS approach (board-specific program knowledge) or the Mac approach
(a new graphics kernel replacement for add-on boards).  It's just alot of
work, and apparently things like Intuition aren't completely pure; on
occasion they do things on their own, instead of going through graphics
library.

>                                         Wayne Knapp


-- 
Dave Haynie Commodore-Amiga (Systems Engineering) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
                    Too much of everything is just enough

nicholso@hpcuha.HP.COM (Ron Nicholson) (04/11/90)

In <13580@spudge.UUCP> johnm@spudge.UUCP (John Munsch) writes:

>It's small consolation, but when Jay Miner came to talk to the Mid Cities 
>Commodore Club a while back he said that if he had it all to do over again he
>would have made the whole thing pixel oriented rather than bitplane oriented.
>Sigh...

Alas. Designing in retrospect is so much easier...

But lets look at some of the constraints that went into the design.  The
people who funded the venture were interested in a video game, not a
personal computer.  So they wanted a competitive price in the consumer
marketplace.  The original date of introduction was to be Christmas of
1984.  Only 64k DRAMS were commodities at that time.  This meant the base
system was limited to 128k RAM.  (and no keyboard, no disk!!!)

A 640x400 16 color image would have taken all of memory.  So being able
to use 3 bits per pixels was seen as important.  6 bitplanes saturated
the memory bus.  8 wasn't possible.  Would you really want to design a
video memory map with packed 3, 5 and 6 bit pixels?  And with only 128k
could you afford to waste memory by using unpacked pixels?  The processor
memory timing was so tight that an MMU had to be left out.  (Also a
fourth custom chip would've taken the design over budget.)  The lack of
any memory mapping precluded using a sparse memory map or multiple ways
of addressing the same pixels.  Chunky pixels?  No way.

But if I could design it over again today...

---
Ronald H. Nicholson, Jr.		Hewlett Packard
uucp: nicholso@hpda.HP.COM 		Cupertino, CA
(408) 447-6603				#include <canonical disclaimer>

Oh goodie, a new one: 4,874,164 issued Oct. 1989
---

sparks@corpane.UUCP (John Sparks) (04/12/90)

daveh@cbmvax.commodore.com (Dave Haynie) writes:

>In article <5975@tekig5.PEN.TEK.COM> wayneck@tekig5.PEN.TEK.COM (Wayne C Knapp) writes:
>>In article <10581@cbmvax.commodore.com>, mks@cbmvax.commodore.com (Michael Sinz - CATS) writes:

>>Having a card like the Hercules card that ran with the Amiga wouldn't 
>>hurt Amiga sales.  Plus, if it didn't require a bridge board and supported
>>intuition it would be economical and very attractive to Amiga developers. 

>The problem isn't the hardware; such a card would at least as easy to design
>for the Amiga bus.  The problem is the software, and the question of where the
>software display intelligence resides.  In most of the MS-DOS world today (of
>course I know this is changing, however slowly, with things like Windows and
>TIGA) the intelligence for driving a particular graphics card is contained in 
>each application program.  Each program must implement it's own graphics, 

>Nothing's stoping anyone from taking the
>MS-DOS approach (board-specific program knowledge) or the Mac approach

ARRRGH!! This is one of my biggest beefs against MSDOS! (I use it everyday at
work), every damn board has to have a specific driver in each software package
that you want to use it with. A real pain in the behind. We just got an
Orchid Pro VGA board (does up to 1024 x 768). But can we use it transparently
with all software? NOOOOOOOooo! I have to have a special driver for AutoCAD,
another for Windows, and another for Ventura (which doesn't work!). And 
under Xwindows under unix, it says it supports the Orchid board but the
driver doesn't work here either. And Deluxe Paint II says it supports the
Orchid, but the 800 X 600 X 256 color mode doesn't work. The board does
have a standard VGA mode (320X200) (640X480) that does work with most programs.
so I have to go with the least common denomenator in most cases. So why bother
paying oodles for a board I can't use most of the time? If it didn't work
with AutoCAD I would have shipped it back.

But the point is, I have no desire to see such kludges appear on the Amiga.
I love the transparency of the hardware to the software. Graphics, sound,
peripherals (printers and modems and such). Don't even get me started on
printers under MSDOS! even worse than the graphics cards. 


-- 
John Sparks  | D.I.S.K. 24hrs 2400bps. Accessable via Starlink (Louisville KY)
sparks@corpane.UUCP |                                     | PH: (502) 968-DISK 
The future isn't what it used to be.