🚀 go-pugleaf

> for drawing on 13h, u will draw in memory first (in an array) and then use
>  repz movsw to move it out to screen. that will be the fastest routine to 
> do it!

That's normally how it is done.

> if u do the same for modeY or X, then it will be very slow! it will be ten
> times slower if u draw to array first then flip to screen!

I draw directly to the screen for these modes.

> i have been using modex, modey for quite some time and found that this is the
> faster way. but it will still be slower than 13h (by around two times)

Ah, somebody is agreeing that it is slower!!!

> anyway, why waste this flicker-free mode and use 13h, which jerks when doing
> large scrollings?

The reason to waste Mode X and Y is because they aren't fast enough.  I've
not seen one super-fast arcade game pulled of in either of these modes (full-
screen, not like Jill).

Josh

 a

i have found that Mode X is easily fast enough if you use the latches
and copy from screen memory (page 3 or 4) to screen memory, and then
flip the pages.  

daniel reznick
reznick@acf3.nyu.edu

> i have found that Mode X is easily fast enough if you use the latches
> and copy from screen memory (page 3 or 4) to screen memory, and then
> flip the pages.  

That is, if you have the memory to laoad everything into screen memory.

Question:  As far as I can tell, Wolf3d does not store all of its images
is screen memory (they wouldn't fit...lessee 64x64=4096 bytes, dividing
256k by it...).  OK, the perspective mapping routine is probably written
to map directly to screen memory.  It is an easy conversion on mine...you
just have to change planes.  Is this true?

Josh

"you just have to change planes"

I'm not sure what this is asking.  In Mode X the pixels are stored in
memory in a weird way, yes.  pixel one on the screen is in memory
location 0, plane 0.  pixel two on the screen is in memory location 0,
plane 1.  pixel three on the screen is in memory location 0, plane 2.
pixel four on the screen is in memory location 0, plane 3.

sorry.. ii I should have numbered pixels from 0 also.

Since the latches are set for all four planes from one read, you can
just set the VGA to take all of its data from the latches and none from
the CPU, and then it just takes one read and one write and viola.. four
pixels.

anyway, it IS confusing.

a point of terminology.. are the bitplanes 0-3 or 1-4?  or does it
matter?

daiel reznick
reznick@acf3.nyu.edu

In article <1992Aug6.155255.57923@cc.usu.edu> sl8nl@cc.usu.edu writes:
>> i have found that Mode X is easily fast enough if you use the latches
>> and copy from screen memory (page 3 or 4) to screen memory, and then
>> flip the pages.  
>
>That is, if you have the memory to laoad everything into screen memory.
>
>Question:  As far as I can tell, Wolf3d does not store all of its images
>is screen memory (they wouldn't fit...lessee 64x64=4096 bytes, dividing
>256k by it...).  OK, the perspective mapping routine is probably written
>to map directly to screen memory.  It is an easy conversion on mine...you
>just have to change planes.  Is this true?
>
>Josh
>

A LOT of games store images in video memory.  Everything from cockpit
overlays to sprites.  You don't need to store every image in the game
in vid. mem., though, just the ones you need "at the moment".

An easy way to tell whether games do page-flipping or store images in
video memory is to run it under Windows, switch away, then switch back
to it.  If the screen is messed up, then the game is using video
memory as more than just a single page frame buffer.

Phat.

In article <62430008@acf3.NYU.EDU> reznick@acf3.NYU.EDU (Daniel Reznick) writes:
>
>a point of terminology.. are the bitplanes 0-3 or 1-4?  or does it
>matter?
>

It would probably be prudent to say that they are numbered from 0-3 because 
there seem to be a fair number of people who are starting to program in graphics
and it would probably confused them.  You need to put the bits in the right 
place to get everything to work right.

Just a thought.


Matt Smith
msmith@peruvian.utah.edu
Student - University of Utah Department of Computer Science

> a point of terminology.. are the bitplanes 0-3 or 1-4?  or does it
> matter?

Michael Abrash numbers them 0-3.  Maybe we ought to stay with an "authority"
on the subject.

Josh

Does anyone know if Wolf3D is in ModeX, Y, or 13h?

is it possible to tell?

and also, if you are going to bother programming an interrupt for the
vertical retrace, then why bother wasting precious display memory for
switching pages?  Just make your copy routine check to make sure it is
OK to write to the screen before it does, and then more images can be
fit into display memory.

you, of course, would have to dosome pretty fancy footwork to make sure
you never accessed the displayed page except when vertical blanking was
in effect, but I imagine it could be done.

dan reznick
reznick@acf3.nyu.edu

You would not need >512k, because in the EGA one byte holds 8 pixels..
The "bit planes" actually function as expected.  one bit across four
planes = 0000 - 1111 (binary) = 2^4 = 16 colors.  Get it?

daniel reznck
reznick@acf3.nyu.edu

In article <1992Aug7.094432.57938@cc.usu.edu> sl8nl@cc.usu.edu writes:
>> a point of terminology.. are the bitplanes 0-3 or 1-4?  or does it
>> matter?
>
>Michael Abrash numbers them 0-3.  Maybe we ought to stay with an "authority"
>on the subject.

The VGA hardware numbers them 1, 2, 4, and 8.(*) I'd say that was the
ultimate authority. ;-)

(*) These are the numbers you put into the low order nibble of the Map
Register in order to select each bitplane.

> You would not need >512k, because in the EGA one byte holds 8 pixels..
> The "bit planes" actually function as expected.  one bit across four
> planes = 0000 - 1111 (binary) = 2^4 = 16 colors.  Get it?

Whoops!  I wasn't the one who said >512k, but I did realize I said that EGA
fit 2 pixels per byte.  I knew better than that!  I was thinking 4 bits per
pixel and just said 2 pixels per byte off the top of my head.

Josh 

>>Michael Abrash numbers them 0-3.  Maybe we ought to stay with an "authority"
>>on the subject.
> 
> The VGA hardware numbers them 1, 2, 4, and 8.(*) I'd say that was the
> ultimate authority. ;-)
> 
> (*) These are the numbers you put into the low order nibble of the Map
> Register in order to select each bitplane.

Ah, but that is the Map mask register.  What about the Read Map register which
numbers them 0, 1, 2, and 3???

Josh

In article <1992Aug10.085720.57998@cc.usu.edu> sl8nl@cc.usu.edu writes:
>>>Michael Abrash numbers them 0-3.  Maybe we ought to stay with an "authority"
>>>on the subject.
>> 
>> The VGA hardware numbers them 1, 2, 4, and 8.(*) I'd say that was the
>> ultimate authority. ;-)
>> 
>> (*) These are the numbers you put into the low order nibble of the Map
>> Register in order to select each bitplane.
>
>Ah, but that is the Map mask register.  What about the Read Map register which
>numbers them 0, 1, 2, and 3???

;-)

I guess this really is a sort of moot discussion; I'll always refer to
them as 0-based and sequentially numbered. Usually it becomes clear in
context, even though there are several mutually exclusive ways to number
the bitplanes.