Three excellent cards there. If you're not hugely into EAX, they have similar features. So focus on what you do care about: CPU overhead.

Should be pretty simple: get a CPU-limited game with good benchmarking features (Q3A running at low resolution springs to mind, but something a few years newer and more taxing for the AXP would be fine too), install cards one by one and do a timedemo. If you only change the card, you should get a good idea of what each one does to your FPS. To be absolutely perfect, you should start each run with a clean install of your OS & drivers - and under Win9x that would really matter, but I suspect that it won't matter much under WinXP for a regular PCI card like these.

I love soundfont bank manager of Audigy2. DOSBox can use it as GM device and with a good sf2 it sounds better than Sound Canvas. Guitar Pro sounds great too (if you use it). Sure, you can use TiMIDIty or something similar but it is more CPU demanding with a big latency.
But as dionb already said, you have three great cards to pick from 🙂

Dionb, I don't have 9x on the system. That is interesting though about a timedemo. I've never thought about it more than what 3d Mark 2003 does for a sound test. I wonder if that would work?

I agree, Wolfus. It's a good problem to have. 🤓

I spent the evening swapping cards around and ran 15 benchmarks of 3dMark03 sound tests. The Audigy 2 was the clear winner with the Santa Cruz just behind. The Xonar DG needs much more CPU, which is not surprising as it was released when Core2 was the norm.

Test system:
Athlon XP Barton @ 2315mhz
Nforce 2 Ultra
2gb DDR @ 420mhz 2.5-3-2-5
Radeon x1950 Pro AGP

audigy 2 zs is where it’s at.

Auzentech x-fi was probably the best pci soundcard though

I agree, Sphere478. No doubting the performance and feature set is there. If I get my hands on an X-fi, then I would test it out too.

ptr1ck wrote on 2022-07-11, 23:37:

Dionb, I don't have 9x on the system.

I know, I was just saying it to contrast with XP. In XP you can probably get away with not reinstalling/re-imaging the system. Don't try that under Win98 though, was all I was saying 😉

That is interesting though about a timedemo. I've never thought about it more than what 3d Mark 2003 does for a sound test. I wonder if that would work?

Yes. Most likely with very similar results. It's just that I prefer benching actual applications used rather than artificial benchmarks, particularly one from an era notorious for drivers being written to identify the benchmark and modify behaviour for better results...

Q3A is a really good benchmark because it's so cleanly written and well-understood. I'm actually going to be using it at work in the near future for some network latency testing - in this case specifically because it doesn't have any modern workarounds for poor network performance. If your network hiccups, you instantly see it in Q3A, more so than in say Fortnite.

ptr1ck wrote on 2022-07-12, 02:48:

I spent the evening swapping cards around and ran 15 benchmarks of 3dMark03 sound tests. The Audigy 2 was the clear winner with […]
Show full quote

I spent the evening swapping cards around and ran 15 benchmarks of 3dMark03 sound tests. The Audigy 2 was the clear winner with the Santa Cruz just behind. The Xonar DG needs much more CPU, which is not surprising as it was released when Core2 was the norm.

Test system:
Athlon XP Barton @ 2315mhz
Nforce 2 Ultra
2gb DDR @ 420mhz 2.5-3-2-5
Radeon x1950 Pro AGP

Great results. I doubt the age of the cards is too relevant in terms of performance, rather that the first two actually do quite a bit in hardware, where the Asus has a low-end C-Media chip that basically leaves all the processing to the CPU. I'm guessing a CMI-8378 chip from the same era as the Audigy2 and Santa Cruz would have shown similar performance.

Audigy 2 has positional audio (early CMSS) and EAX4 support in hardware.
Santa Cruz has positional audio (Sensaura3D) and EAX2 in hardware.
Xonar DG is trash.

Sphere478 wrote:

Auzentech x-fi was probably the best pci soundcard though

Didn't have external I/O like X-Fi Elite, so nope. Although it could be main contender for PCI DSP with best analog signal.

dionb wrote on 2022-07-12, 08:57:

Great results. I doubt the age of the cards is too relevant in terms of performance, rather that the first two actually do quite a bit in hardware, where the Asus has a low-end C-Media chip that basically leaves all the processing to the CPU. I'm guessing a CMI-8378 chip from the same era as the Audigy2 and Santa Cruz would have shown similar performance.

That's pretty much what I was thinking except for I don't believe it matters if it's the low-end Xonar or high-end. I believe either one would count on offloading more to the CPU than either the Audigy or Santa Cruz.

The Serpent Rider- I wouldn't consider it trash for Windows 7 or newer; it's still a solid upgrade from latter model low-end onboard sound, like the Realtek ALC269 I have in an Optiplex 790. It is trash for a Windows XP build though. Seems like it comes down to Sensaura vs better EAX.

The Serpent Rider wrote on 2022-07-12, 09:21:

Audigy 2 has positional audio (early CMSS) and EAX4 support in hardware.

You can also get CMSS2 on any Audigy card (even the first one) when using the DanielK driver pack under WinXP.

ptr1ck wrote on 2022-07-12, 15:10:

Seems like it comes down to Sensaura vs better EAX.

The Audigy's EAX3 is a pretty substantial upgrade from EAX2 which Sensaura uses. Especially if you're into stealth games like Splinter Cell.

Joseph_Joestar wrote:

You can also get CMSS2 on any Audigy card (even the first one) when using the DanielK driver pack under WinXP.

Technically, it comes to two variations - Audigy CMSS or X-Fi CMSS. Practically speaking, X-Fi implementation is improved Sensaura3D in anything but name (because Creative bought them).

You can also get CMSS2 on any Audigy card (even the first one) when using the DanielK driver pack under WinXP.

Well yeah, all true Audigy cards use the same DSP, with some minor tweaks.