split charset in rasters

[tebe]

color yellow = charset #0

color blue = charset #1

 

http://atariarea.krap.pl/forum/viewtopic.php?pid=108675#p108675

 

Atari800Win show screen OK, but Altirra no

split_charset.zip

Edited April 11, 2010 by tebe

[Tezz]

Whilst working on MM with the cycle counting for mid scanline changes, I've noticed a slight difference in the amount of acceptable cycles between Awin+ and Alterra. I was planning to do some tests between them and the real hardware.

[phaeron]

color yellow = charset #0

color blue = charset #1

 

http://atariarea.krap.pl/forum/viewtopic.php?pid=108675#p108675

 

Atari800Win show screen OK, but Altirra no

 

Yes, it looks like you've found an emulation bug -- ANTIC needs to respond one cycle later to the CHBASE change.

 

In general, timing accuracy in Altirra is a bit better for GTIA than it is for ANTIC or POKEY. I've fixed some of the issues in 1.5, but haven't done a full release yet, so the DMACTL and STIMER fixes are only in 1.6 test releases:

 

http://www.virtualdub.org/beta/Altirra-1.6-test15.zip

[Heaven/TQA]

can someone post a screenshot how the demo should look like?

[phaeron]

can someone post a screenshot how the demo should look like?

 

Here's a picture from my NTSC 800XL. YCMV (your colors may vary).

 

Adding a one-cycle delay to CHBASE seems to fix the problem:

http://www.virtualdub.org/beta/Altirra-1.6-test17.zip

[tebe]

disabled "badlines" (GED+ mode without badlines)

 

new g2f feature, but 5-th color missed and 8 first and 4 last lines from screen

disable_badlines.zip

[Tezz]

disabled "badlines" (GED+ mode without badlines)

 

new g2f feature, but 5-th color missed and 8 first and 4 last lines from screen

interesting, so it's a software char mode via mode E? I'll go take a look at this now. Edited April 18, 2010 by Tezz

[Rybags]

So this uses some sort of VScrol trick?

 

But, what's the use of disabling badlines? That should mean you get stuck with the same 40 characters and would need to change character set every 8 lines.

[tebe]

interesting, so it's a software char mode via mode E?

 

yes

[Rybags]

But, what advantage does it offer over traditional G2F techniques, or Mode E?

 

I can visualise a situation where you might have 3 splits so you can use characters 0-119 of 8 character sets over 24 lines, but it isn't going to give any memory savings, plus you'd be stuck with whatever attributes the initial character has (bit 7).

[tebe]

But, what advantage does it offer over traditional G2F techniques, or Mode E?

 

nothing special, only "we can do it"