View Single Post
Old 15 December 2013, 20:32   #454
Toni Wilen
WinUAE developer
Join Date: Aug 2001
Location: Hämeenlinna/Finland
Age: 43
Posts: 20,660
Originally Posted by amilo3438 View Post
Back In 1990 (Demo) by Samplers

A500 KS1.2 ... pic.1 ... some dash remain when logo/title closes !!! (Note: Also some glitches on 1st screen logo/title doesnt happen if KS1.2 0.5+0.5 !!!)
Does not happen on real A500 but there is something strange going on because exactly identical glitch appears if I enter Action Replay 3 and then exit it..

EDIT: AR3 is just a side-effect, nothing to do with real problem, it is emulation missing feature that hasn't been triggered until now. Demo has really funny accidental bug.

Copperlist sets 3 plane lores mode, then after 3 full bitplane fetches it writes zero to BPLCON0, immediately after last "fetch block's" BPL1DAT write. This last write has some zeros in upper bits (something like $0xxx), Denise is quick enough to switch of bitplane output just before non-zero data gets shifted out.. This "Agnus fetched it all but Denise didn't output it all" condition isn't emulated.

100% accidental, there is no way coder did this on purpose

EDIT2: Above is slightly wrong, writing to BPLCON0 after Denise starts shifting won't do anything but if BPLCON0 is written immediately after BPL1DAT write, it will prevent Denise from starting to output new 16 bits of data.

Corrupted logo is confirmed, depends on memory configuration.

Mental Unbalance (Intro) by Sarvet

A500 OCS/ECS (most common) or any other model (even A1200) ... pic.2 !!!
Same problem on real A500 or A1200. It has AGA-only registers in copper list.

Megademo by Science 451

A500 OCS/ECS ... pic.3 !!!
Probably usual "lets disable sprite dma while it is active" bug. (I couldn't duplicate it by just starting emulation with disk in drive)

Party-Piece (Demo) by Scope

A500 OCS/ECS (most common) ... some dots on the object edges ... pic.4 !!!
A500 confirmed.

City Demo (Demo) by Scrapers

A500 OCS/ECS (most common) ... pic.5 !!!
A500 confirmed (also tested using A100). Same garbage.

Last edited by Toni Wilen; 16 December 2013 at 18:29.
Toni Wilen is offline  
Page generated in 0.11458 seconds with 9 queries