[Discuss] Add support to Rad Racer II (4screen mirroring, scanline IRQ)

[grimoire]

Hi, I am working on support Rad Racer II. Which include 4 screen and scanline IRQ.

before:

After:

There are still some glitch, I guess it has something to do with the way I add scanline IRQ in PPU. But I have no idea how to fix it.

I have add extra RAM in mapper to support nametable 2 and 3. It would be much easer to just resize the picturebus. But I think we should emulate what the NES do, right? Would it be better to add another mirroring type MapperControled so we can support more complex mirroring for some Mapper?

[amhndu]

Thank you for the pull request!

I have add extra RAM in mapper to support nametable 2 and 3. It would be much easer to just resize the picturebus. But I think we should emulate what the NES do, right? Would it be better to add another mirroring type MapperControled so we can support more complex mirroring for some Mapper?

I think this approach is good, since it's mapper specific, the extra memory should go there. I like the idea of MapperControlled!

[amhndu]

Hey, sorry for the delay. I'm currently cleaning up some mappers, including MMC3, unfortunately I have little time, so I'm not done yet. I'll try to clean that as soon as possible. I also just found out that nametable isn't correctly read from an iNES file. I'll need to fix that before this will work well.

MapperControlled

After reading more about this, I realize that FourScreen and MapperControlled basically mean the same thing. I don't think we need a new enum for that anymore.

[grimoire]

MMC5 has a complex mirroring mode which is configured by a mapping register, that is why I add this new enum (maybe renaming the FourScreen would be better?).

I am refactoring PPU steps following PPU rendering because I thought it has something to do with the glitch (trigger the IRQ by A13). But the glitch still exists after I finish(almost) the refactor. Do you have any idea what should I do to fix it?

By the way, do you need github hooks? Such as format and spellcheck. I have some experiments on that.

[amhndu]

GitHub hooks

That sounds great! Some contributed files have bad indentation, this would really help

[grimoire]

Sure. I will add the hooks after you cleaning up the mappers.

[amhndu]

I've pushed up some changes to master, including a fix for ines header reading. (Although I've yet to find a Rad Racer dump that has the correct 4 screen header). And getting back to NameTableMirroring, I suggest removing MapperControlled, and using FourScreen instead. FourScreen should mean mapper controlled. That's the only it's possible and that's the nesdev wiki seems to be calling it.

[amhndu]

But the glitch still exists after I finish(almost) the refactor. Do you have any idea what should I do to fix it?

I also see glitches in SMB2 and Kirby's Adventure, it's possible that MMC3 itself has a bug. Another functionality missing is the APU, it's possible that some ROMs are using that for timing. Working on the APU is also on my queue.

[SuzieQQ]

Sure. I will add the hooks after you cleaning up the mappers.

Hi I saw that you fixed IRQ did you try rad racer 2 which uses mapper 4 you can try to see if it goes in alien 3 game should go now

[grimoire]

aliens 3 works on 7f4aac5 but failed on 1617ff7. guess I have made some mistake.

[grimoire]

@SuzieQQ Alien 3 works now.

[SuzieQQ]

@SuzieQQ Alien 3 works now. as soon as i can i try it, now i don't remember i had a small list with IRQ bugs with other games maybe super mario 2 should go in game now! this is a particular game they say

[SuzieQQ]

i have tested all the games i had tried in the past these are the improvements, the rest either behaves the same, or some no longer work. I noticed that in particular all the ninja turtle games never worked

[amhndu]

@SuzieQQ Does SMB3 work now?

[SuzieQQ]

@SuzieQQ Does SMB3 work now?

no, before it went into the menu now graphical glitch and does not enter the menu

[grimoire]

@amhndu @SuzieQQ Hi, I think I find out why SMB3 not working. I download the rom from here and the head[6] & 0x2 is 0, that means no extended RAM. But after IRQ, CPU will try to read $0x7964, which leads to 0 because there is no such RAM. After I force the RAM enable, the game works.

I am not sure if it is caused by the ROM(maybe this is a hacked one). Any advice? May I remove else if (addr < 0x8000) branch in MainBus.cpp and leave every read to m_mapper->readPRG(addr)?

[amhndu]

Great find! I think a simpler fix could be to simply always have extended ram available. There's no particular reason to "disable" it I think. I've made some more changes so I'll do the change. This is good work! I can merge this MR if you feel this is in a good state.