clueless1 wrote:

James-F wrote:

clueless1 wrote:

Yep. that's exactly what I didn't hear until I disabled L1 cache. In that case, I think it's not a sound card issue, but rather a game issue.
It looks like we might need to add Duke2 to that page.

It looks like I've spent 20$ for no apparent reason. 😀
Still, it'll be fun to compare.

Yes, for sure. I'd like to hear sound comparisons with the YMF and ES1688F 😀

I wonder if it is possible for one sound card to be more susceptible to speed issues than another. So what if you have to disable L1 for the Audician, but not the ES1668F? Maybe you can tell us when you get your card. 😉

I think the sound card can play a factor in some games working or breaking with CPU speed. Duke Nukem II works fine with the SB16, but now we're hearing the CPU is too fast when used with the OPL3SA.

I can attest that Jill of the Jungle was running too fast with the SB16. Also it cannot auto detect the sound card during setup (there's no manual sound setup, so you're screwed if it doesn't auto detect). I had to disable L1 cache to get everything to work right, including sound detection.

HOWEVER, with the Yamaha OPL3SA card, I can play Jill of the Jungle without any speed issues AND it detects the sound card just fine without disabling any cache . So odd...

clueless1 wrote:

I'd like to hear sound comparisons with the YMF and ES1688F 😀

Will do.
Name few games that you'd like to compare.

GuyTechie wrote:

I can attest that Jill of the Jungle was running too fast with the SB16.

It might be a different issue with Jill.
I believe the SB16 is not backward compatible with the SBPro2, so older games may not detect the SB16 properly.

James-F wrote:

clueless1 wrote:

I'd like to hear sound comparisons with the YMF and ES1688F 😀

Will do.
Name few games that you'd like to compare.

Wing Commander 1-2
Ultima Underworld 1-3
Secret of Monkey Island
Doom

Pretty decently deverse range. 😀

If you don't have those, just pick anything you have that would be interesting to compare.

Thanks!

In Duke2, I confirmed both the Audician speed sensitivity bug and the bug quoted from Great Hierophant:
-on my 486-66/SB16, all sounds are there and no slowdown needed.
-on my P200MMX/Audician, I am missing lots (all?) of the SB sounds unless I reduce cpu speed to maximum very slow 486 levels.
-on P200MMX/Audician, shooting a door in Episode 1 makes no digitized sound in SB mode, even with cache disabled. If you turn on Adlib sounds, you hear the Adlib version of the door exploding. So a workaround might be to turn on both SB and Adlib sounds so the Adlib supplements where an SB sound might be missing.

If I had more time, I'd swap sound cards to see if the SB16 needs slowdown in my P200MMX, or if the Audician needs slowdown on the DX2-66. Might be a project to plan for the future.

For reference, here is the YMF bug description from GH:

Great Hierophant wrote:

Duke Nukem II uses 8-bit to 3-bit or 8-bit to 2-bit ADPCM capabilities of the Sound Blaster cards for sound effects that the YMF cards do not emulate.

Maybe it's worth building a AT-based 486 PC. Of course, those are harder and harder to find.

clueless1 wrote:

If I had more time, I'd swap sound cards to see if the SB16 needs slowdown in my P200MMX, or if the Audician needs slowdown on the DX2-66. Might be a project to plan for the future.

Not sure about the SB16 but Duke II works fine with my ES1688 based card in a PIII 1GHz.

I have this card (Audician 32 Plus) in my 486DX2/66, just tested out Duke Nukem II and I'm missing most of the SB sound effects. As mentioned switching on Adlib sound effects brought those back into the mix

This is the first game I've had any difficulty with thus far, I've got another one in a Pentium 3 I use for Duke 3D etc. which works absolutely spot on as well.

Nipedley wrote:

I have this card (Audician 32 Plus) in my 486DX2/66, just tested out Duke Nukem II and I'm missing most of the SB sound effects. As mentioned switching on Adlib sound effects brought those back into the mix

This is the first game I've had any difficulty with thus far, I've got another one in a Pentium 3 I use for Duke 3D etc. which works absolutely spot on as well.

Interesting. Can you disable caches and see if the sounds return? Either through the BIOS (L1, L2) or with Setmul (L1 only).
I'm curious if even disabling L2 only slows it down enough for sounds to play normally.

Disabling L2 cache helped a bit, and disabling L1 helped more, but it still isn't right. Looks like I would have to slow the system down even more for it to run normally, unfortunately I haven't got my turbo button hooked up yet

This game is quite the pain !

Can anyone please test this with a YMF card:
Supaplex freezes when I change the audio device in the Controls menu (in game).
Is it only on my 233MMX +YMF system?

As for DN2:

Duke Nukem II uses 8-bit to 3-bit or 8-bit to 2-bit ADPCM capabilities of the Sound Blaster cards for sound effects that the YMF cards do not emulate.

How is the CPU caches related to the above?

James-F wrote:

Can anyone please test this with a YMF card: Supaplex freezes when I change the audio device in the Controls menu (in game). Is […]
Show full quote

Can anyone please test this with a YMF card:
Supaplex freezes when I change the audio device in the Controls menu (in game).
Is it only on my 233MMX +YMF system?

As for DN2:

Duke Nukem II uses 8-bit to 3-bit or 8-bit to 2-bit ADPCM capabilities of the Sound Blaster cards for sound effects that the YMF cards do not emulate.

How is the CPU caches related to the above?

I don't have Supaplex, so can't help there.

The quoted from GH is a separate issue that appears to be inherent to all YMF cards. That is what causes the missing digital explosion when shooting doors. AFAIK, that is the only missing sound in DN2 that is due to the missing emulation that GH describes.

The cache issue seems to be an issue that the Audician (maybe all YMF cards?) has with old, speed sensitive games. Disabling L1/L2 caches slows the system down so the game runs at normal speeds, which keeps the Audician from glitching the sounds. I guess I don't notice it because I create batch files for my games that automatically disable caches to run each game at a speed appropriate for when it was released. So if a game is from 1990, I will try to match a 386DX-25. If a game is from 1993, I'll try to manipulate caches to get close to fast 486 speeds. Etc. There are exceptions, like System Shock, which is a 1994 game but still needs as much cpu as possible to run in SVGA.

edit: it seems most of the audio-related speed sensitive sound issues are from games made in 1993 or earlier:
http://www.vogonswiki.com/index.php/List_of_C … sensitive_games

edit2: from Wikipedia-

Due to hardware dependent programming, the PC version of Supaplex ran twice as fast as PCs became faster. Herman Perk disassembled the game, debugged it and re-assembled it again. The result became known as SpeedFix. Extra features have also been added without changing the game itself.

Supaplex was declared freeware a long time ago and can be downloaded from the official website: http://www.elmerproductions.com/sp/index.html
I'd really appreciate if someone would try the bug in my previous post.
Thanks! 😀

In that case, I'll see if I can duplicate your bug this evening. My filtering DNS says that site is malicious. 🤣. But I somehow think that is a false positive.

clueless1 wrote:

I guess I don't notice it because I create batch files for my games that automatically disable caches to run each game at a speed appropriate for when it was released. So if a game is from 1990, I will try to match a 386DX-25. If a game is from 1993, I'll try to manipulate caches to get close to fast 486 speeds. Etc. There are exceptions, like System Shock, which is a 1994 game but still needs as much cpu as possible.

Can you give an example of how you did this? I didn't realize there was a way to disable cache and manipulate speeds temporarily before the game is run (and then I assume you restore the speed after the game quits?).

It would be nice to do, so I can avoid rebooting to change the settings in BIOS.

Thanks for insight.

GuyTechie wrote:

Can you give an example of how you did this? I didn't realize there was a way to disable cache and manipulate speeds temporaril […]
Show full quote

clueless1 wrote:

I guess I don't notice it because I create batch files for my games that automatically disable caches to run each game at a speed appropriate for when it was released. So if a game is from 1990, I will try to match a 386DX-25. If a game is from 1993, I'll try to manipulate caches to get close to fast 486 speeds. Etc. There are exceptions, like System Shock, which is a 1994 game but still needs as much cpu as possible.

Can you give an example of how you did this? I didn't realize there was a way to disable cache and manipulate speeds temporarily before the game is run (and then I assume you restore the speed after the game quits?).

It would be nice to do, so I can avoid rebooting to change the settings in BIOS.

Thanks for insight.

I use Setmul, written by forum member gerwin. Its usage depends on your CPU. For sure it will disable L1 on any 486 and higher. Other CPUs (some VIA, AMD and Cyrix) let you change multipliers or disable L2. Lastly, Setmul supports disabling Test Registers, which are in Pentium P54Cs. The test registers give you many more options for slowing down to different levels. My Pentium Overdrive 200MMX happens to support ALL the test registers, but my Pentium 120 only support some of them. Only way to know for sure is to test.

So here is how I use it in my batch files. Let's say you want to disable L1 for Wing Commander. Make a batch file in your WC directory with the following:
setmul l1d
wc.exe
setmul l1e

On my POD, if I want to make it as slow as possible for Knights of the Sky, I disable L1 as well as the Test Registers Branch Prediction and V-Pipeline:
setmul l1d bpd vpd
knights.com
setmul l1e pfe

Check the setmul page I linked to above for all the command parameters. It's an incredible tool.

This 8bit to 2bit/ 3bit ADPCM sound also affects PCI YMF cards in DN2. IIRC, also Crystal based ISA cards are affected, and probably Analogue Devices, too.

James-F wrote:

Supaplex was declared freeware a long time ago and can be downloaded from the official website: http://www.elmerproductions.com/sp/index.html
I'd really appreciate if someone would try the bug in my previous post.
Thanks! 😀

Hey James! I was able to duplicate your bug on my P200MMX/Audician system. But, I think I found the cause--the mouse. I can make it crash every time when my mouse driver is loaded and I'm navigating with it, clicking between the sound options. Interestingly, the crash ALWAYS occurs when I select Sound Blaster and click FX to play the digital explosion. However, when I boot to a configuration without my mouse driver loaded, and navigate with the arrow keys, I cannot get the crash to occur. So it seems to be a conflict between my mouse driver and the game? I use ctmouse.exe. Try booting up without your mouse and see if it stops crashing. I'm curious!

edit: I'm going to try on my 486/Sound Blaster rig and see if I can duplicate there.

edit2: cannot duplicate on the 486/Sound Blaster. Weird.

edit3: cannot get the arrow keys to move the cursor on my 486. Weird squared. Have to disable the mouse and use the arrow keys on my Pentium, have to enable mouse and use it to move the cursor on my 486. 🤣.

What is your mouse/driver?

gdjacobs wrote:

What is your mouse/driver?

On the Pentium/Audician it is a Logitech PS/2 optical with ctmouse.exe
On the 486/SB16 it is a BSR BX400 serial (plugged into serial port--no serial-to-PS/2 adapter). Again, ctmouse.exe

Same with the Logitech or MS driver?