Re: ISA Vector card already done...

On Wed, 25 Mar 1998 13:26:33 -0800, Clay Cowgill <ClayC@diamondmm.com> wrote:

>> I didn't worry about the max number of vectors, refresh rate, etc. I
>> looked at
>> it from the point of view of the monitor. I can't remember the slew
>> rate off
>> the top of my head, but I took the slew rate in in/sec, used
>> displayable screen
>> size in the horizontal direction (15 inches I think), divided by 1024,
>> etc. and
>> ended up with 6.xx mhz. I then looked at the clock going into the
>> DACs on an
>> asteroids and saw a 6 mhz clock.
>>
>I follow you, but is that the number we really want to be figuring out?
>The worst-case vector slew rates would be caused by an instantaneous
>re-positioning of the beam cause by changing the DAC outputs in large
>steps (like from 0V to +fullscale output).

Worst case for the monitor, no difference for the generating hardware (who
care's what value we place on the DAC)

>We're only moving a few bits
>at a time and changing the output slowly in comparison.

The speed of the updates inside a vector is constant whether we're drawing a
vector or jumping to a new one. Jumping to a new one saves us time since we
must now wait for the monitor to catch up.
>
>I looked at Asteroids on the scope once and it looked like it was using
>a 16ms "frame" speed. The "draw" time was a fraction of that amount. I
>want to say something like 6-8ms when things were actually getting
>drawn.
>
>Vector display signals "seem" pretty slow. (You can hear the neck
>chatter, look at them on a slow scope w/out problems, etc.)

You obviously can't here the 6mhz sample rate! And as for a scope, if the trace
was small enough, and the screen big enough you should be able to see the same
6mhz staircase pattern you see on a 19" display. My guess is very few scopes
allow you to see this. Take a careful look at a finely focused 19" asteroid
game. It's more obvious a certain angle, but you can definitely see the digital
artifacts of the 6mhz sample rate of the DACs. Not nice and smooth like the
Cinematronics.
>
>Something just feels kinda "fishy" about needing a super-fast system to
>make it work. (Well, we know that Atari's last vector design was DSP
>based, so something must be possible...)

You could do it with a 50mhz (one clock per instruction) PIC.
>
>> This is the same problem I had with using software. I calculated that
>> I would
>> be able to run around 20 instruction per update in the yet to be
>> released 50mhz
>> PIC clone and it looked like I could write a line draw routine that
>> would fit...
>>
>Still should work on the DSP then 'cause loops are free and I've got a
>33MHz clock rate internally.

At one instruction per clock it will be very close...

This is why I wanted to place a design in a GAL or something, it was hard
finding a processor fast enough. The thing is you don't need complex math
instruction, you need very simple instructions a *raw* speed!

>I think we're figuring something wrong
>though in the necessary speed. I can't explain where the error is, but
>it doesn't "feel" quite right.

The Asteroids schematic worked for me, that and it matched my independent
calculations on how fast I would have to update. I was really shooting for a
2048x2048 resolution, but that would of course double the update speed! :^(

The best conformation would be to place a scope directly on the DAC clock of an
Asteroids game (something I didn't do) just to make sure I'm reading the
schematic properly. If the DACs are updating at a 6mhz rate, you can bet we're
going to have to do the same.

-Zonn

<><><><><><><><><><><><><><><><><><><><><><><><><><><><

 ------ ___ Member of A.A.C.S.:
 |---- | ( ) Association for Artistically
    / / ( () ) Challenged Signatures
   / / //\\ // (__)
  / ---/ // \\ //\\ // zonn @ zonn . com
 -------| // \\/
Received on Wed Mar 25 14:07:54 1998