Let me add that once I exchanged my Awe32 with an Awe64 Gold all my old games with Adlib problems (i.e. Secret of the Silver Blades, Hard Nova, Dynablaster, etc...) played their music flawlessly!
Remember the article didn't say the OPL chip was omitted, it was integrated. Perhaps with a couple of tweaks..?

jez: I don't understand how Qbix's explanation has anything to do with cycles...

No? With a low number of cycles, the game is able to detect Adlib, which is the preferred sound hardware and so it uses it. However, I presume that it is not able to detect it with a high number of cycles, and so falls back to rather nasty CMS sound.

I got this assumption from jal's comment, "DOSbox doesn't emulate hardware on a timing level, I think, so [not getting Adlib sound when running Prince of Persia 1 at a high speed] might be [the result of] an interaction between DOSbox and the AdLib emulation (which is taken from Mame)."

IIRC, some early games didn't detect Adlib on fast PCs.

Zorbid: I do remember Sierra games having troubles with that. I remember that it was a standard practice to kill time by playing with I/O ports a certain number of times while waiting for the Adlib to respond, which resulted in a much shorter delay on faster computers.

HunterZ wrote:

Zorbid: I do remember Sierra games having troubles with that. I remember that it was a standard practice to kill time by playing with I/O ports a certain number of times while waiting for the Adlib to respond, which resulted in a much shorter delay on faster computers.

Well, since the ISA-bus used in PC's in those days had a standard speed, reading I/O-ports resulted in a standard delay. So it is not true that faster PCs had much shorter delays. The problem was that some games (and I'm not sure Sierra did this too) detected speed in a less processor speed friendly way (e.g. empty loops), and also that some hardware used non-standard bus speeds (resulting not only in problems with detecting hardware, but the operation of it in general, causing unstable PCs).

JAL

Zorbid wrote:

IIRC, some early games didn't detect Adlib on fast PCs.

True, but that shouldn't effect (affect?) DOSbox too much, since DOSbox doesn't time its I/O to the CPU speed (or better: the cycle speed), so the timing problem with higher cycles is a result of a different technical problem, though resulting in the same hardwaer 'malfunction'.

JAL