Forums

Enri have been working on figuring out the side-effects of clearing VDP register 3 bits in mode 4:
http://smspower.org/forums/viewtopic.php?t=11582
It's in Japanese. I haven't bothered carefully reading/translating this yet, but it suggest that some tiles ends overlapping due to the effect of those bits on internal addressing.

Currently they are described as is in Charles' documentation:

Register $03 - Color Table Base Address

D7 - No effect [...] D0 - No effect

For the SMS VDP only, all bits should be set. Otherwise the VDP will fetch
pattern data and name table data incorrectly.

Register $04 - Background Pattern Generator Base Address

D7 - No effect [...] D0 - No effect

For the SMS VDP only, bits 2-0 should be set. Otherwise the VDP will fetch
pattern data and name table data incorrectly.

Bock wrote

Enri have been working on figuring out the side-effects of clearing VDP register 3 bits in mode 4:

This sounds about right. I noticed that it alters graphics in rectangular areas, as if bits of the register were masked against the tile row and column values as the VDP scans the name table.

There seemed to be a lot of garbage in the display, even when I carefully set up VRAM. So in the end it seemed unusable. But I am sure there could be applications if more details were known about how the VDP operates when this register is changed. I am glad Enri is investigating it.

(Charles do you have problems to login lately?)

Sounds similar to mode 3 colour/pattern address table masking Sean documented in tms9918a.txt, a handful of games on MSX use this to their advantage. Does the SMS have this feature in TMS mode btw?

Doesn't it just mask the tiles and tilemap addresses?
Maybe I should do a bit of testing on this...
@ hap: Do you know any these MSX games which use mode3 with address masking?

FluBBa wrote

Doesn't it just mask the tiles and tilemap addresses?
Maybe I should do a bit of testing on this...

That would be logical, but it doesn't quite seem to be as straightforward as that. It would be very cool to see the results of your tests!

I could have done something improper in my test, but even if I restricted tile data to bitplane 0 only changing this register caused data to be shown in the other bitplanes. I remember seeing background colors that shouldn't have shown up based on the tile data in VRAM.

Maybe the color table register is only used for one of several VRAM accesses, say fetching bitplanes 0,1 and not 2,3. This would allow a name table value at one location to reference bitmap data at two distinct addresses, if some of the bitplane reads were masked and others were not.

Bock wrote

(Charles do you have problems to login lately?)

I think I have it solved now. :)

FluBBa: Unfortunately, I haven't documented the affected games, here's a few from the top of my head:
- Can of Worms (not in all subgames)
- Dr. Archie
- Ice King

Author	Message
Bock Site Admin Joined: 08 Jul 2001 Posts: 8651 Location: Paris, France	VDP register 3 in mode 4 Posted: Thu Jan 15, 2009 3:48 am
	Enri have been working on figuring out the side-effects of clearing VDP register 3 bits in mode 4: http://smspower.org/forums/viewtopic.php?t=11582 It's in Japanese. I haven't bothered carefully reading/translating this yet, but it suggest that some tiles ends overlapping due to the effect of those bits on internal addressing. Currently they are described as is in Charles' documentation: Register $03 - Color Table Base Address D7 - No effect [...] D0 - No effect For the SMS VDP only, all bits should be set. Otherwise the VDP will fetch pattern data and name table data incorrectly. Register $04 - Background Pattern Generator Base Address D7 - No effect [...] D0 - No effect For the SMS VDP only, bits 2-0 should be set. Otherwise the VDP will fetch pattern data and name table data incorrectly.

Charles Guest	Posted: Fri Jan 16, 2009 1:44 am
Charles Guest	Bock wrote Enri have been working on figuring out the side-effects of clearing VDP register 3 bits in mode 4: This sounds about right. I noticed that it alters graphics in rectangular areas, as if bits of the register were masked against the tile row and column values as the VDP scans the name table. There seemed to be a lot of garbage in the display, even when I carefully set up VRAM. So in the end it seemed unusable. But I am sure there could be applications if more details were known about how the VDP operates when this register is changed. I am glad Enri is investigating it.

Bock Site Admin Joined: 08 Jul 2001 Posts: 8651 Location: Paris, France	Posted: Fri Jan 16, 2009 5:03 am
	(Charles do you have problems to login lately?)

hap Guest	Posted: Fri Jan 16, 2009 11:08 am
hap Guest	Sounds similar to mode 3 colour/pattern address table masking Sean documented in tms9918a.txt, a handful of games on MSX use this to their advantage. Does the SMS have this feature in TMS mode btw?

FluBBa Joined: 21 Jul 2005 Posts: 412 Location: GBG	Posted: Fri Jan 16, 2009 1:07 pm
FluBBa Joined: 21 Jul 2005 Posts: 412 Location: GBG	Doesn't it just mask the tiles and tilemap addresses? Maybe I should do a bit of testing on this... @ hap: Do you know any these MSX games which use mode3 with address masking?

Charles MacDonald Joined: 28 Sep 1999 Posts: 1197	Posted: Fri Jan 16, 2009 1:30 pm
Charles MacDonald Joined: 28 Sep 1999 Posts: 1197	FluBBa wrote Doesn't it just mask the tiles and tilemap addresses? Maybe I should do a bit of testing on this... That would be logical, but it doesn't quite seem to be as straightforward as that. It would be very cool to see the results of your tests! I could have done something improper in my test, but even if I restricted tile data to bitplane 0 only changing this register caused data to be shown in the other bitplanes. I remember seeing background colors that shouldn't have shown up based on the tile data in VRAM. Maybe the color table register is only used for one of several VRAM accesses, say fetching bitplanes 0,1 and not 2,3. This would allow a name table value at one location to reference bitmap data at two distinct addresses, if some of the bitplane reads were masked and others were not. Bock wrote (Charles do you have problems to login lately?) I think I have it solved now. :)

hap Guest	Posted: Fri Jan 16, 2009 4:33 pm
hap Guest	FluBBa: Unfortunately, I haven't documented the affected games, here's a few from the top of my head: - Can of Worms (not in all subgames) - Dr. Archie - Ice King

Forums

View topic - VDP register 3 in mode 4