Progress March - II : Fix rendering works !
Added 2016-03-11 00:55:42 +0000 UTC...from a tilemap snapshot ;)
In the last post I was aiming at making ao68000 run at least up to the MVS checks. I actually got fed up of trying to debug a poorly documented core, so I switched to specific NeoGeo stuff instead for this month.
Since I believe this can also be seen as a milestone, I'm making this a paid update. Thank you everyone :)
As you can see, the output of the design is able to render the fix layer correctly. The complete screencap is from MAME, and the fix-only one is the output from the design, obtained by feeding it a virtual Joy Joy Kid cartridge and a VRAM dump from MAME.
This means that the data path and sequencing from the cartridge's ROM and tilemap in VRAM to the video output is good enough to be glitch-free.
This doesn't mean that all the timings are good yet. All memories are simulated by having the worst possible delays and most signals match those of a real NeoGeo, but some are passed between chips which should not, meaning that there are some simplifications to be done.
The next step will be to do more measurements on my AES to see what doesn't match, and correct my guesswork. Once done, ground will be clean enough to move on to sprites.
