Jo22 wrote on 2023-04-22, 23:02:

So if some of you have a 286 or higher PC with an AdLib card installed, please have a look it. 😀
It also works with Windows 3.0 MME and can coexist with the AdLib MIDI driver..

Thanks! I can confirm this drivers works (and sound is very clear, really) in Windows 95 on Hand386

n0p wrote on 2023-08-06, 10:59:

Jo22 wrote on 2023-04-22, 23:02:

So if some of you have a 286 or higher PC with an AdLib card installed, please have a look it. 😀
It also works with Windows 3.0 MME and can coexist with the AdLib MIDI driver..

Thanks! I can confirm this drivers works (and sound is very clear, really) in Windows 95 on Hand386

Glad to hear! ^^ Thank you, too for testing! 😎 👍

The quality is so good. Adlib is a no DMA sound blaster. Why was it not used more

Cyberdyne wrote on 2023-08-06, 21:46:

The quality is so good. Adlib is a no DMA sound blaster. Why was it not used more

Adlib as DAC is 6-bit, right?
So it's worse than non-DMA SB, or Covox.

6bit logarithmic per operator, but you won't be using several channels or ops to get more since the access is far too slow to make it work well enough to increase resolution. That 6bit log volume control is comparable to something between 4...5 bits linear DAC

Isn't logarithmic supposed to provide better dynamic range than linear?
A-law and μ-law are based on that idea...

Here you one side of the waveform (because you play with the volume control and not the source sample itself in the YM chip) and not both sides around the zero point, so it really really hurts whan can be achieved. One half of the waveform you try to play has very low precision and other half has increasing precision toward the extremes where it doesn't need it, while stuff like A and µLaw have the precision near the zero point where it is most needed and begin to lose in as amplitude increases but where it no longer is necessary to have. I don't right now have access to my test data from older times when I was playing around with it, I could show exactly the problem and its effects.

Tiido wrote on 2023-08-07, 17:48:

6bit logarithmic per operator, but you won't be using several channels or ops to get more since the access is far too slow to make it work well enough to increase resolution. That 6bit log volume control is comparable to something between 4...5 bits linear DAC

Yes, I vaguely recall that I once worked out that even a single channel can only play at about 30 kHz on an OPL2 because you couldn't send the commands faster.

In optimal case you can only reach the chips' own sample rate (which requires synchronity with the chip for best results) but to do any of the advanced stuff you need more than that and it is impossible even on the chips with much shorter access times like OPL3. (I have not tested OPL2 per se but I have worked with its relative OPLL, OPN, OPM chips and all of those can take a new byte every sample period when you don't set up index value)

Yea, for AdLib the OPL2 runs at 49,7 kHz, but you're unlikely to reach that.