>I'll leave it as it is now, most games using the DMC IRQ works now.
So, you're saying that the new value for type $D works better than the old?
>the DMC DMA steals 2 cycles every time it fetches a byte, can anyone confirm this?
No. However, I honestly don't believe that Nintendo would design such a sloppy DMA design. 2 bus clocks for 1 bus read? That would be pretty sad (and not to sound arrogant, but even I could design a 1 clock cycle DMA read setup between the 6502 & the DMC).
>This would be good to have in the DMC.txt file, whatever the value is ;) just for the sake of completenes. And maybe some notes about the frequency is only for NTSC. Aaaaaand a smal note about $4017 bit 6 & 7?
I haven't really been maintaining the sound docs in past years, mainly due to lack of interest. For adding information on things like the relationship between writes to $4017, and the sound channel timing, I think that this would be more suited to the person who figured it out (Ki), since it seems he did extensive work on figuring out minor quirks about the sound hardware, which I didn't even notice were there. I have no idea what other sound discoveries may have been made (other than the "using $4011 for volume" trick discussed here recently), as I've been off the mailing list since october 2000.