So I hopped aboard the groovymame train, I love my MiSTer, but I'd also like to play a wider selection of arcade games. I got a $85 optiplex with a i5-3570, and installed a $10 radeon hd 5450.
After some initial trouble finding the right drivers, I got a basic setup going, but I was wondering what kind of values I should generally be setting for my frame delay if possible to get a close to PCB experience.
I'm kind of ignoring vsync-offset, which to my understanding only needs to be taken into account if using hlsl; I initially used a scanline effect since I'm using a 31khz pc crt, but that seemed to cause tearing with frame delay, and I didn't want to deal with vsync offset, so I just hooked up an external scanline generator.
EDIT: I actually ended up having to go back to using the scanline effect, the external scanline generator doesn't work with all resolutions, and it seems like tearing only happens when frame delay is too high for the cpu
Anyway, testing a few games at random, street fighter 2, r-type, ninja baseball batman, and dodonpachi, I generally was able to get away with a frame delay value of 6, but that seems to be the edge of what my cpu can handle, I had to dial ninja baseball batman back to 5 due to occasional frame drops, and I think the audio in dodonpachi might occasionally be stuttering, but I can't tell, and the fps counter didn't drop below 100%.
So typically what frame delay values would give PCB like results? For heavier games where I have to compromise, what lower frame delay values would at least make the experience better than stock mame?
Ideal frame delay values for Groovymame?
-
cave hermit
- Posts: 1547
- Joined: Sat Sep 07, 2013 2:46 pm
- Location: cave hermit
Re: Ideal frame delay values for Groovymame?
Its best to set it per game, but you can start with an across the board value of 6. If you get sound stutters or other tell tale signs that its too much for certain games, you can back it off per game. For most mid-90's games or earlier, you might be able to go up to 7 or even 8 possibly.