DIWStart & DIWStop Help!

[DanScott]

I have an overscan (352 pixel wide) screen set up, for scrolling with an extra 16 pixels on the left... so...

DDFSTART = $28
DDFSTOP = $d8


but it does not seem to be working with these
DIWSTART = $2471
DIWSTOP = $34d1

It seems to be displaying (in WinUAE at least.. I don't have an A500) the 16 pixels to the left, that should be "hidden" by the DIWSTART & DIWSTOP settings.

Can anyone see where I am going wrong?

with
DIWSTART = $2481
DIWSTOP = $34c1
it is displaying the centred 320 pixels ok...

[Leffmann]

It's a bug in the hardware that occurs when the right edge of the display window extends past $C7. I don't know if there's any way to work around it other than moving the display to the left.

[DanScott]

so I guess there were no full overscan scrolling games?

I *could* cover up the left edge with a hardware sprite... it's a bit hacky though...

[Galahad/FLT]

Originally Posted by DanScott

so I guess there were no full overscan scrolling games?

I *could* cover up the left edge with a hardware sprite... it's a bit hacky though...

Double Dragon 2 was left and right scrolling and full overscan

[DanScott]

Originally Posted by Galahad/FLT

Double Dragon 2 was left and right scrolling and full overscan

what's the copper-list saying there? or did it have bitplane fetch glitching on the very left (off the edge of the visible TV screen) ?

Knowing Rich, he probably found a way around this bug

[DanScott]

Originally Posted by Galahad/FLT

Double Dragon 2 was left and right scrolling and full overscan

320x268 overscan in the Y only

[DanScott]

Pacmania is overscan.. it has horizontal DIWStop at $c5

[Galahad/FLT]

Originally Posted by DanScott

what's the copper-list saying there? or did it have bitplane fetch glitching on the very left (off the edge of the visible TV screen) ?

Knowing Rich, he probably found a way around this bug

DIWSTRT=$2081
DIWSTOP=$2cc1
DDFSTRT=$0030
DDFSTOP=$00d0

[DanScott]

I've set my DIWStop at $c7, it's the best i can do, and looks...well..looks ok

[Toni Wilen]

It happens because $c8+ is larger than largest Denise horizontal counter value. It is position where counter is reset (actual hpos=0 position) but it is visible in right border because horizontal blanking starts slightly later, same reason which makes copper COLOR0 changes visible in right border if they are not delayed by few cycles.

$c7 = $1c7 = 455 dec, Denise internal horizontal counter counts twice the speed of Agnus horizontal counter (2 to 455, more information in undocumented hardware thread), in lores pixel rate. AGA introduced hires/shres rates = sprite and scrolling can be hires/shres positioned.

$c8 = $1c8 = 456 dec which never happens. This causes horizontal window to stay open all the time (open position matches, closing position never matches) = full horizontal overscan.

btw, larger overscan (in right border) has also another problem if you need horizontal scrolling: BPLCON1 values "glitch" when horizontal counter is reset. (Check Seven Seas / Andromeda horizontal scrolling max overscan image copper list for more details.. It has a workaround..)

[DanScott]

Thanks for the info Toni.

Good to have a proper description of what is happening internally.

[Photon]

Originally Posted by Galahad/FLT

Double Dragon 2 was left and right scrolling and full overscan

This doesn't mean so much if it wasn't using the hardware scroll registers. From memory the game performance was an order of magnitude worse than the song, which they botched by using the wrong player.

From memory, $30..$d8 never gave me any problems but you could only get 4 pixels past that at left edge with scrolling? I did do a maximum overscan test years ago and wrote a thread here.

Oh well, it's a very simple test to perform.

I never use DIW values that aren't evenly divisible by 8.