Memory Full

Forum / Development


Hicks * 02 Oct 2018 23:08:36 * Modified at 23:35:13

Something that I have never looked at closely... but some of you did it! I think in particular to Madram, Offset and Grim. Knowledge sharing thread!

On CRTC 1, we can do a kind of "delta-RVI": since the Memory Adress (MA) is readed when C4=C0=0 (VCC=HCC=0), then, even if we display the bloc 0 (C9=0, i.e. VLC=0) then the bloc 1, no invisible vertical splitting is necessary, we just have to put the new MA and R9=1. If we must display bloc 0 then 4, we put 3 invisible screens (1-2-3) and set MA is ok too.

But on CRTC 0, we can't do that because MA is readed only when C4=C0=C9=0, I mean, we MUST pass upon a bloc 0 (C9=0). But the rumor says that all the sequences are no longer possible on this CRTC (possible to use the bloc 7?), and that there is strange comportments...

I know that Offset analyzed this for ACE (CRTC 0 emulation is 100% accurate I think), Madram too for his incoming demo, and the bearded mole, hmm... analyze everthing :)

m_dr_m * 03 Oct 2018 19:32:02

I encounter no issue triggering all possible combinations (8x8) while refreshing the address.

The only drawback is that displaying the last "bloc" with a new address requires 7 ruptures (x2 ┬Ás). The last one of them becomes visible on sunset.

That didn't bother the authors of 'From Scratch', though.

Hicks * 05 Oct 2018 00:15:20

Quite a good news!

So maybe strange comportments occur when register 5 join the game. Offset opinion here will be useful, since he is the one who noticed the problem.

'From Scratch' only used 4 'blocs' on 4 'pages', then approx. 320 steps. But it's now 10 years old, and I hope your next demo will bury that and show us unexpected exploitation of this technic!

[Note for non-French readers: RVI mean "Rupture Verticale Invisible" = "Invisible Vertical Splitting", so we can call it IVS in english.]