Modules playing slower on WinUAE than on DeliPlayer Win32?

[andreas]

(apologies - i should possibly file this as a bug (?) on deliplayer forum)

[Toni Wilen]

This is nothing special and very possible.

There are two view points. Emulated program sees things synced and "perfect" but for outside viewer it may be something different.

For example background program can steal all cpu time for 1 second, emulation appears to be paused + sound keeps repeating but emulated program does not see anything special and is still perfectly happy.

[andreas]

Phh, you were slightly too quick (as always )
In a quick thought, I planned to actually get this thread deleted again, as ...

WinUAE + my REAL THING (A1000) *are* in sync!! (at least in "human terms")

But DeliPlayer is a bit too quick, *significantly* faster, about 1/3 BPM!
With a sequencer in your reach, this difference is possible to measure, as it will create "divergence" with the one, while it will not with the other!

[thomas]

Doesn't DeliPlayer have an option to switch between 50Hz and 60Hz (PAL and NTSC repectively) when playing Amiga mods ? "Too fast" might be a European opinion. Americans might find it "too slow". Mod players on the Amiga always used the vertical blank interrupt for playing mods and didn't care about PAL or NTSC. So this doesn't need to be a bug but rather a feature.

[andreas]

It might, but where?
Moreover, I think that the 50/60 Hz difference is MUCH more than just 1/3 bpm in playing speed. I can upload you an NTSC game, and you can try yourself to set it in "wrong" 60 Hz mode and then hear the huge difference in sound. :-) In BPM terms, this may be roughly 5 whole ones apart, not just a fraction < 0.5.

Oh, last minute I read this from you: vertical blank interrupt!
Yes, I did have seen this option (Players->Tracker), and it was set to:

[x] CIA timing (else VBlank Timing)

CIA timing is preset, I haven't yet changed this.

[Toni Wilen]

I am surprised, I was expecting some small difference

[andreas]

I reckon that not. A tune that is played in "wrong" mode (50/60, ntsc/pal, vsync on etc.), i. e. a NTSC game's in-game tune in PAL mode or a PAL game's in-game tune in NTSC mode rather sounds as if you have the tune pressed on a 12'' and crank up the PITCH slider on your turntable, as DJs sometimes do it...

1/3 (read: one third) bpm difference is *far* too small to perceive unless you measure it with a good software or either hardware "BPM meter" or sequencer.
If *I* can hear it without measuring in PAL/NTSC stuff, it must be a whole 2 or more.

Originally Posted by andreas

Phh, you were slightly too quick (as always )
In a quick thought, I planned to actually get this thread deleted again, as ...

WinUAE + my REAL THING (A1000) *are* in sync!! (at least in "human terms")

But DeliPlayer is a bit too quick, *significantly* faster, about 1/3 BPM!
With a sequencer in your reach, this difference is possible to measure, as it will create "divergence" with the one, while it will not with the other!

If I would have to take a stab at guessing the cause for this, it could be rounding error when calculating "ticks of paula clock between output samples".

The PAL paula clock is 3546895 Hz if I recall correctly, and division by either 44100 or 48000 isn't going to go even. If you ignore the residual, you produce values such as 80 ticks of Paula per sample instead of 80.3 which is more correct. (In this case you should actually play slower than the real Amiga, so I guess deliplayer guys have made some other kind of error. They might round up, or use wrong clock value, or something else altogether.)

Slight speed errors are common, because most mod player writers ignore this kind of problems because you can't really tell it without comparing playback closely like you have done. In UADE, this calculation uses floating point and corrects for any accumulated error, and the constants should be authentic Amiga values, so the playback speed should be exact (in audible comparisons, further errors may arise from inaccuracy in your sound hardware: maybe you don't get exact 44.1 kHz or 48 kHz playback. Lots of hardware actually produces something pretty close, but not always quite exactly that output sample rate).

[andreas]

Thanks alankila!

That's a fantastic explanation for the matter! I wager Toni read this too with interest! If I had spotted before that somebody replied to my 3-year-old thread, I'd have replied you much earlier too - sorry!

[cosmicfrog]

and I though I was forgetful hehehhee

[Toni Wilen]

Thanks for reminding, it is finally done

(not that it has nothing to do with playing speed but it made sound sync more stable)