furrtek / neogeo_mister_old

SNK NeoGeo core for the MiSTer platform

Neogeo_MiSTer

It's all in the name :) WIP, see DEV_NOTES.md and https://docs.google.com/spreadsheets/d/1CA_RlD6NTEXU8FZAATqmE4FEGLo2nq-Fe_9v-TF6sh4/edit?usp=sharing for a compatibility list.

If you want to try it out, make sure to use the modified MiSTer binary.

neogeo_mister_old's Issues

Disable all MVS elements when in console mode

Check read data from memory zones/registers dedicated to MVS components on the home console.
Open bus ? All ones ?

Replace TG68K with FX68K

See https://github.com/ijor/fx68k

Auto load appropriate system ROM file when changing system mode

Right now a game must be loaded after changing the system mode to update the system ROM, this is confusing.

Force a reload and reset if the system type was changed when exiting the OSD menu.

Timer interrupt never occurs

See Neo Turf Masters.
Maybe the cause of Viewpoint not starting up.

Check interrupt priority encoder in LSPC and test mode gates.

Add support for CHD format

Needs to be done in Main_MiSTer mostly.

Unibios detects AES when System type is set to Arcade

Apparently the jumper connected to NEO-C1 isn't the only indicator.

Check PRO-CT0 protection

Tested ok a while ago but need to see if it still works after modifications.

Protection trip effects:

• Fatal Fury 2: times out immediately or makes opponent invincible after stage 1.
• Super Sidekicks: player can't move.

The MVS self-test is sometimes skipped

Testing of memories and RTC TP should take 2~3 seconds (green garbage screen).
In some cases, the game starts right up.

Get rid of internal tri-stating

Used a lot for M68K_DATA and SDD buses.
Letting synthesis figure out how to place the muxes can be dangerous.

Sprite tile remap hack is too slow

There's not always enough time for the tile # to pass through the multiple muxes introduced by the hack before it's used by the SDRAM controller. This causes heavy glitches in some compilation runs.

Do the remap as soon as the tile # is latched from VRAM, not when it appears on the P BUS.

Try to make ROM loading faster

spi_write() is what takes the longer in the loading loop.
In spi_write(), it's the fpga i/o function calls.

Add CDZ DMA handling

Different microcode. See disasm of neocd.bin and MAME's drivers/neogeocd.cpp

Riding Hero runs slow because of absent link MCU

Implementing the whole MCU isn't worth it since it's only used for link play.
Just add a bit of logic to fake an idle reply from it to make the game happy.

Only show dipswitch options when in arcade mode

Conditionally set the core config string to hide or show the dipswitch options according to system mode.

Some sprites are cut off or not displayed at all

See League Bowling lane, Puzzle Bobble backgrounds, many games with full-screen scrolling maps...

This was happening in simulation and was fixed by adding # delays in some cells.
Can this be solved with timing constraints, or by using a 2x clock and additional registers ?

Put 68k work RAM in SDRAM

Get rid of the block RAM and merge it with the Neo CD code. Using 64kB more of SDRAM isn't a problem for cart mode and makes things simpler.

Add NeoGeo CD mode

Quite a bit of work to be done here...

No audio at all in Nazca games

They're probably all using the same sound driver so no audio in mslug and turfmast makes sense.
Check if ZMC is working properly, might be the cause of glitchy audio in many other games.

Distorted audio using Optical Audio on the MiSTer IO Board

When using Optical Audio out via the MiSTer IO Board (Playing Puzzle Bobble and Last Resort), the audio is loudly distorted similar to a bug with optical audio Sorgelig encountered and fixed.

I know the audio isn't even close to 100% on the Neo Geo core, but I don't think it would cause distortion to the point of peaking audio levels on my receiver.