There sure is.

I've got a crazy idea that maybe we're supposed to be patching the Japanese rom.

Edit: This is the part where I remember what "(JU)" means.

Edit II: Here's the solution: it only works in VirtuaNES

Really? With the proper music and everything? Now that's weird.

Now, here's the big question - is this because the other emulators are flawed, or is this because VirtuaNES is flawed? And what will happen when this runs on the real hardware?

I ran it through readnes and it reports that it's MMC3, and I don't have an MMC3 test cart made up yet - I've been having a hard time locating donor games. I have like three or four carts that would be perfect - except the plating on the edge connectors are so worn down that they barely work as it is. What is it with MMC3 games that leads them to be improperly cleaned with abraisives?!?

Thinking about the problem, it's almost as if the data in memory (pointer for the location of the music data) wasn't getting properly initialized before it's read by the program. Once a section of code forces it to update (mario has died, etc,) it'll work. I'll have to try it again, but ISTR that if you play until a Game Over, then the next game starts with the correct music.

The plot thickens.

-Ian

Ehhhh...
I'd put NEStopia and Nintendulator well above FCEUX in terms of accuracy.

I finally got around to actually trying the patch provided by KillaB earlier in the thread.  Not on a flash cart, though, but I don't think I need that for confirmation.  I'm pretty sure I know what's going on here (very common problem with old SMB hacks).

This is the same old, same old -1 world caused by the hack being made with a bad ROM as a base.  The reason you start on world  0-1 instead of 1-1 is because the game uses unprepped RAM for the level marker and assume it's zero'd, when in fact it isn't (it's actually usually $FF, hence why you start on level "-1" instead of level 0).

Nintendulator does seem to play it properly -- but it may zero RAM on startup for compatibility reasons?

In any event -- this is a problem with the ROM/patch.  NEStopia and FCEUX are doing it right.

Awesome! This is great. And I'm glad you were able to figure out what causes these problems. I'm at work right now, but when I get home, I'll apply the patches and burn it into EPROMS and try it in my test cart.

But it makes perfect sense what's going on. Is this the same sort of thing that's affecting the beginning music of Extra Mario Bros. as well? Uninitialized memory being read as a variable?

Thanks for your help Killa B! I can't wait to try this out tonight!

-Ian

On the Derok site it plays fine for me. Was this the game you were talking about?
http://www.derok.net/nes-roms/super-mario-bros/strange

I had a discussion with vx and he's confirmed that Extra Mario Bros. plays the glitched death sound at the beginning, the way it does in Nestopia and FCEUX. So Nintendulator is emulating it incorrectly (perhaps intentionally, for compatability reasons, like Disch said?).

I'll probably see what I can find about whether it's a known bug (feature?) tomorrow, and file a report if necessary. As long as I remember to do it.

Edit: Remembered just now to check this. ...I can't find a bug tracker anywhere. Guess I'm not filing a report.