Paar wrote on 2023-03-02, 09:31:

The list could be expanded with the information that all SB16s have problem with SB Pro compatibility. With normal SB16s playing sound in mono, but Vibra chips playing stereo but in reverse. Don't know if the AWE64 plays in stereo too.

So SB16 Vibra models like CT2800/2900 /4180 etc will play SB Pro games that have stereo in stereo instead of mono.
First time i've heard that/. If this has been tested what game was used that supported SB Pro but not SB16 and had stereo digital sounds.

My CT 4520 AWE 64 value plays mono in SB pro mode, but when using the Midito utility that user aitotat wrote it plays stereo in reverse.

James-F wrote on 2016-10-07, 14:38:

You don't have to assume the meaning of the term "single-cycle DMA" it is clearly documented in the official Sound Blaster Programming Guide.

This is a fascinating read, thank you for providing the programming guide.

It answers many questions.

It seems the single cycle DMA clicking is not a bug, but a feature. The application programmer is supposed to use auto-init DMA mode, not single cycle mode.

Auto-initialize DMA mode is crucial when dealing with high data rates. To give you some idea of the numbers involved, CD-quality […]
Show full quote

Auto-initialize DMA mode is crucial when dealing with high data rates. To give you
some idea of the numbers involved, CD-quality sound (16-bit stereo at 44.1 kHz)
involves transferring data at a rate of 176.4 KB/s. Under these conditions, using single-
cycle DMA mode would produce less than optimal sound quality because of the time
needed to reprogram the DMA and DSP to start transferring a new block at the end of
every block of data. The delay between the blocks (though it may be brief), is enough to
distort the sound.

Single cycle mode is only supposed to be used for a single buffer xfer, once-off and done.

We bash creative for their shitty cards, some criticisms being fair, but we somehow never seem to bash game developers for their shitty programming.

When you insist to multi-cycle in single cycle mode and don't even bother to futz around with your samples so that you at least zero cross on a cycle boundary, you are retarded. Period.

st31276a wrote on 2026-04-17, 10:13:

We bash creative for their shitty cards, some criticisms being fair, but we somehow never seem to bash game developers for their shitty programming.

*cough* A fair bit of low-hanging fruit there, if you're into throwing shade:

Joshua Jensen sound drivers that seldom play at the same pitch between different sound cards. (One Must Fall, Jazz Jackrabbit, Epic Pinball, several others)
Miles Sound System with major OPL3 FM tempo problems (Descent I comes immediately to mind - changing the music device to Adlib / OPL2 greatly reduces, but does not eliminate the tempo stumbles).
Early Epic Megagames titles with DAC sound that will just randomly decide to lock the computer up for no reason, if you're running the game on a system faster than a 486-33.
And the fact that the only really usable sound interface for QuickBASIC 4.5 is good ol' fashioned Bells, Whistles and Sound Boards. Everything else either just plain doesn't work with anything that isn't the sound card(s) the library programmer had, or is horribly temperamental with anything that isn't a 100% faithful Creative SB Pro 2 clone.

On the plus side, Creative cards are hackable enough to where you can correct some of their worst and most common problems, single cycle DMA clicking aside on SB16s.

st31276a wrote on 2026-04-17, 10:13:

When you insist to multi-cycle in single cycle mode and don't even bother to futz around with your samples so that you at least zero cross on a cycle boundary, you are retarded. Period.

To be perfectly fair, that's a rather exotic hardware fla- er, feature that is completely non-intuitive to your average programmer, unless they've read up on all the documentation. (And good luck with that if they're in a crunch cycle.) Not helped by the fact that the dominant PC sound card standard, SoundBlaster Pro, didn't have this issue at all.

Correct me if I'm wrong but didn't the auto init mode date all the way back to the original sb 1.0? Seems like single cycle was just a bad habit that never died

maxtherabbit wrote on 2026-09-01, 00:48:

Correct me if I'm wrong but didn't the auto init mode date all the way back to the original sb 1.0? Seems like single cycle was just a bad habit that never died

For games that had simple monophonic samples, it was an easy way to write the sound driver once and have it work on almost any card

It's good someone wrote up about the bugs I only wish I knew years ago before i bought one thankfully I sent it back though.
Now we know what they're REALLY worth boys and girls you should get them cheaper if you're in the "something" than "nothing" situation.
I actually thought the one card I had was faulty so I’ll be more careful if I want a Sound Blaster now.

zapbuzz wrote on 2026-09-01, 01:04:

It's good someone wrote up about the bugs I only wish I knew years ago before i bought one thankfully I sent it back though.
Now we know what they're REALLY worth boys and girls you should get them cheaper if you're in the "something" than "nothing" situation.
I actually thought the one card I had was faulty so I’ll be more careful if I want a Sound Blaster now.

IMHO the bugs/deficiencies with later Creative cards are a tad overblown. Yeah, the single cycle DMA clicking sucked, but far more damning for many was the broken MPU-401 emulation on most SB16s. This made the prospect of plugging in a Wave Blaster or MT-32 problematic at best. The other issues (well, apart from the squealing bug of the ViBRA cards) I feel are fixable by the end-user at least...

The 'self noise' issues I suspect are rather a symptom of the cards with said DACs having long, kinky trace runs, and especially floating op-amps (a common sight on *MANY* SB16 models, and even on the SoundBlaster Pro 2). I measured a *9 dB* average noise floor improvement just by addressing the floating op-amps on one card, and I wasn't being very tidy with my rework either.
I suspect the reason the assumed 'better' DAC chips are less noisy, is because Creative was simplifying the SB16 architecture around the time they were introduced, and in so doing giving their cards less literal junk for the signals to flow through (and fewer floating op-amps in the process).
I currently have a longcat OG SB AWE32, one with a 'noisy' self noise rating. But listening by ear, I don't get much noise from it at all - especially not after I've addressed the floating op-amps and such. It does have the typical single-cycle clicking though, but whatever. Genecyst x.xx is happy with it, so I'm happy with it.

As for CQM... it blows, but it's still miles better than AudioPCI's FM-to-wavetable wrapper. If it bugs you real bad, and you've got one of those intermediate cards with footprints for an OPL3 chip plus DAC (and the good fortune to have the chips needed), you can do a De-CQM mod.

Or you can learn to just live with the distorted timbres and chirpy artefacts. Given current eBay prices, it makes little sense to roll sound cards if you're on a shoestring budget anymore.

As for the single-cycle DMA clicking, I still think it's a mute circuit with a voltage error being rapidly toggled on and off when the card is in single-cycle DMA transfer. Normally the pauses in transfer shouldn't cause (very) audible clicks unless the output is momentarily railed when the mute is active. It'd instead sound 'chunky', like it's periodically skipping samples - a great example of what I'm talking about is most any Sega Mega Drive/Genesis game that uses a lot of DAC PCM samples.
Someone needs to record the output of the card playing single-cycle DMA PCM sounds, and look for telltale peaking/clipping (or just the waveform not quite settling to the zero point as it should) in the parts that are muted.