C NES programming: tutorials, guides, anything's welcome

[tokumaru]

I'm writing a small library for cc65 to ease the NES dev, who knows, maybe you can write a game in c with decent fps.

It really depends on the type of the game. If you mean a platformer with fast scrolling and many active entities, you probably can't do it in C. A simpler game with a static camera and a controlled number of entities can probably be done in C if a proper library in ASM is available.

where can I get Tetramino?

About Tetramino, check tepples' profile for a link to his page.

[Petruza]

Hi, I wrote this test program, which will be a metronome.

It makes a beep every 3rd of a second.
I'm relying on vertical retrace being 60hz because it's NTSC, as I don't know any other way to measure time in the NES. ( do you? )
It works fine in emulators, but it hangs my PMP after some beeps.
Am I doing something wrong or is just my PMP that is buggy?
It runs most of commercial games very well, although others crash.

PS: I figuerd out how to make sounds somewhat randomly, but it works.

Code: Select all

#include <nes.h>
#define addr(_addr)		(*(unsigned char*) (_addr))
void main()
{
	unsigned char a = 0;
	addr(0x4015) = 1; // init sound
	addr(0x4001) = 0;
	while(1)
	{
		waitvblank();
		++a;
		if( a == 20 )
		{
			a = 0;
			addr(0x4003) = 2; // make a beep
		}
	}
}

[Celius]

I think you should initialize registers $4000 and $4002. $4002 holds the 8 least significant bits of the pitch value, and $4000 determines whether or not it sweeps up, the volume, etc. I don't actually know much about hardware volume decay, because I do volume decay manually to have more control over it. I would just advise against assuming $4000 or $4002 have any value, so just initialize them.

Also, it looks like your code is waiting for Vblank in an infinite loop. It would be best to put the code that's executed once every frame in the NMI routine, which is fired up every Vblank.

[Memblers]

We'd have to see the ROM or assembly. But if your PMP crashes with some games, that doesn't sound very promising.

[Petruza]

Celius: I don't know how to use the NMI routine, can it be accessed in c?
Or is it like a callback or something only feasible in asm?

Memblers: It's strange, but I added a clrscr() and cprintf() calls, and now it doesn't crash.

[kyuusaku]

NMI should be a function in the NES library.

[Celius]

The NMI is an interrupt fired every Vblank, and the starting address of the NMI is located at $FFFA (I think) in the fixed bank. If you're doing just an NROM 16 or 32k PRG game, the address for the NMI will be located just at $FFFA. So if bit 7 of $2000 is set, the NMI will be fired every Vblank. Otherwise if bit 7 of $2000 is 0 then it won't fire.

I don't know how you set that up in C though. Just set it up so $FFFA/$FFFB of the fixed bank holds the address of the start of the NMI routine. Then make sure you end the NMI routine with RTI. Though I don't know how you say "RTI" in C, unless you can just type "RTI".

I think you should be able to access it in C. I don't see why you wouldn't be able to. If you can't, then you're missing one of the most (if not the most) essential things in programming the NES.

[tepples]

Then make sure you end the NMI routine with RTI. Though I don't know how you say "RTI" in C, unless you can just type "RTI".

In some dialects of C, you can add a keyword before the definition of a function to get it to generate different calling conventions. For example, a compiler might generate RTI instead of RTS at the end of a function declared this way:

Code: Select all

__interrupt__ void isr() {
  /* blah */
}

I don't know whether CC65 implements anything like this, but it does have inline assembly as one of its extensions.

Still, I'd recommend writing your ISRs in assembly language. NMI in particular only has to increment a variable and then return.