VOGONS


Reply 20 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t

I think it is time to test the sound card to see if it works.

I will build a DOS 6.22 + Win 3.1 box to test the card.
If the card works as well as I hope I will use this box for games that need General Midi or Roland MT-32 emulation.
In the future I am thinking of buying a real MT-32 if I find a cheap one locally or at least within the EU.
The box clearly states that this sound cards Midi/Joystick interface is "Roland MPU-401 compatible".

Im thinking of using an unlocked Pentium II "Klamath" so I can change the speed from ~133 mhz to ~375 mhz. This will make it easy to avoid the Borland Pascal compiler bug.
I have an Aopen i440BX AX6BC that I think will let me choose 2x multiplier but it dosnt support lower FSB than 66 mhz.
Another option is a QDI Legend i440BX Brilliance 1 but that board will not let me choose FSB >83mhz with a 66 mhz FSB CPU.
I also have an unnamed Slot-1 board that support lower fsb but it has jumpers...

The video card is also a hard choice.
In my other boxes I use S3, Nvidia and/or 3dfx cards.
I have a nice AGP ATI Rage Pro 8mb that I would like to find use for but all ATI cards seem to have scrolling issues in some games.

Other video cards with Windows 3.1x drivers I can choose from.
Nvidia: Riva128, TNT and TNT2
3dfx: Rush (PCI), Banshee and Voodoo3
Matrox: Millenium II (PCI), G100 and G200
S3: Trio 64 (2mb,PCI) and Virge (2mb,PCI).

Perhaps the Riva 128 would be a good choice since the TNTx seems to have bad drivers for Windows 3.1x?
Another option is the Banshee but it seems a bit overkill in a Windows 3.1 box.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 21 of 40, by gerwin

User metadata
Rank l33t
Rank
l33t
Skyscraper wrote:
Im thinking of using an unlocked Pentium II "Klamath" so I can change the speed from ~133 mhz to ~375 mhz. This will make it eas […]
Show full quote

Im thinking of using an unlocked Pentium II "Klamath" so I can change the speed from ~133 mhz to ~375 mhz. This will make it easy to avoid the Borland Pascal compiler bug.
I have an Aopen i440BX AX6BC that I think will let me choose 2x multiplier but it dosnt support lower FSB than 66 mhz.
Another option is a QDI Legend i440BX Brilliance 1 but that board will not let me choose FSB >83mhz with a 66 mhz FSB CPU.
I also have an unnamed Slot-1 board that support lower fsb but it has jumpers...

Or consider an unlocked "Deschutes". Almost the same, just a little cooler and faster.
You are right to notice that most i440BX boards have design choices which are undesirable for underclocking.
Normally my suggestion would be to select a board with the desired CPU Multiplier settings, then see if you can do the FSB settings by software, such as SoftFSB. Which overrides the mentioned FSB limitations.
However, Windows 3.1 will not support SoftFSB, And only SMB by Rayer runs in DOS, with a very small list of supported PLL chipsets.
Multiplier jumpers require one to open the case, but their settings make more sense compared to the inconsistent BIOS settings.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 22 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t
gerwin wrote:
Or consider an unlocked "Deschutes". Almost the same, just a little cooler and faster. You are right to notice that most i440BX […]
Show full quote

Or consider an unlocked "Deschutes". Almost the same, just a little cooler and faster.
You are right to notice that most i440BX boards have design choices which are undesirable for underclocking.
Normally my suggestion would be to select a board with the desired CPU Multiplier settings, then see if you can do the FSB settings by software, such as SoftFSB. Which overrides the mentioned FSB limitations.
However, Windows 3.1 will not support SoftFSB, And only SMB by Rayer runs in DOS, with a very small list of supported PLL chipsets.
Multiplier jumpers require one to open the case, but their settings make more sense compared to the inconsistent BIOS settings.

I have a Deschutes P2-400 sl2s7 DA1
It should be unlocked since it is the same stepping the Vogon user "bestemor" has.
But in one of your threads bestemor had issues with the L2 cache disappearing if multipliers below 3.0 was used.
So the speed range would be 133,136,150,166 with no L2 cache availible and 200 - ~500 mhz with or without L2 cache.
Perhaps that bug is an Abit only bug.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 23 of 40, by gerwin

User metadata
Rank l33t
Rank
l33t

It is now clear that product code is not a decisive indication for any P-II (un)locking. You need to look at the production date (9830 or earlier IIRC), and maybe country-code too (best chance with MALAY). Since recently I also obtained a SL2S7, with exactly these specs, and found it to be unlocked. The same product code, but from week 33 'Philippines', is locked.
Yeah, L2 cache won't work at 2.0X/2.5X on the Deschutes, whilst Klamath has two 2.0X multiplier settings: one with L2 and one without. Details...

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 24 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t
gerwin wrote:

It is now clear that product code is not a decisive indication for any P-II (un)locking. You need to look at the production date (9830 or earlier IIRC), and maybe country-code too (best chance with MALAY). Since recently I also obtained a SL2S7, with exactly these specs, and found it to be unlocked. The same product code, but from week 33 'Philippines', is locked.
Yeah, L2 cache won't work at 2.0X/2.5X on the Deschutes, whilst Klamath has two 2.0X multiplier settings: one with L2 and one without. Details...

9831 10371-0373 Malay so I guess my CPU is locked.

The rest of the Deschutes I checked are DBx I think. They at least have later product codes.
I have a few more, I just need to find them. With luck one of them is old enough.
I will try the Klamath until I find an unlocked Deschutes.
If Im not mistaken the unlocked P2 300 I have should do 350+ mhz.
Perhaps it is best to leave the case open...

[Edit]

I have built the new Dos/Windows 3.1 box now.

The spec.

Aopen i440BX AX6BC
Klamath P2 300
256mb SDRAM
STB Velocity 128
Audio Wave Platinum 16
2.5 gb Maxtor HDD
1.44 floppy + Random optical drive.

So it is a 1997 build with a somewhat newer motherboard and a sound card from 1994.
Windows 3.1x on this machine is not that far-fetched since many stayed with 3.1x until Windows 98 was released.
Perhaps not the same people who bought a Pentium 300 1997 😁

I have not installed the sound card yet, it will have to wait until tomorrow.

One small issue. If I use the 2x multiplier I loose the L2 cache but at 2.5x it works fine.
I have had this CPU running at 133 mhz with L2 cache on another motherboard.
This is a bit strange but I guess this motherboard sees the 2x setting as its fail safe setting...
133 - 375 mhz without L2 and 166 - 375 mhz with L2 cache gives a speed range from slow/mid Pentium to fast Pentium 2.
This Klamath @375 mhz scores 434.37 in speedsys.
@133 mhz without L2 it scores a bit better than a P133 but the memory performance suck so it will be more like a P120 in games.
I do not seem to have the option to disable the L1 cache. Perhaps a bios update will add that option.

Im thinking of switching this board with the BH6 I use in my Windows 98 Coppermine box.
But that will have to wait, first I want to test the sound card.

[/Edit]

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 25 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t

The card is installed and works both in dos and in Windows 3.11!
I have not installed any games yet but the wavetable sounds much better than Creatives AWE when it comes to playing MIDI files.

Some notes.

The documentation is not always correct.
The Windows MIDI driver I installed is called "Roland MPU-401" just as it is in the manual.
In the manual the device should be named the same in the MIDI Mapper but its called LAPC1 instead.
When I edit I can point it to the "Roland MPU-401" and I can choose or mix the "Patch Map" of my choice (two MT-32 and two other) or just use default map.
I know that the LAPC1 is a MT-32 on an ISA-card but its still a bit strange that my MIDI device got that name in the MIDI Mapper instead of MPU-401.
It works as it should though because if I change the "Patch Map" to MT-32 it sounds less good with less/other instruments compared to the default map.

One other thing

A sound card with Sound Blaster Pro compatibility + WSS compatibility and a MiDI daughterboard uses alot of resources.
It uses IRQ 7 and DMA 1 for the S.B Pro. The WSS uses IRQ 11 + DMA 3 (+DMA 0 for recording in Windows) and the wavetable uses IRQ 2 and/or? 9.
It took some negotiation to get everyone happy.

I can change the resources without re-jumpering the card with a DOS utility and in the Windows drivers. (there are no jumpers for IRQs and DMAs only for port addresses)
The WSS part of the card seem to really like IRQ 11.
If I change the WSS IRQ to 10 in the DOS utility it happily uses IRQ 10 until next reboot.
But there isnt an option to "save" and changing the WSS line in config.sys does nothing. DEVICE=C:\P16\P16.SYS /D3 /I11 <--- "/I10"
I guess this have to do with the fact that sound cards have to use IRQ 11 for WSS to be in line with the WSS standard?
Changing resources for the S.B Pro part of the card works great and there is an option to save changes.

Not related

The Windows 3.1x drivers for the AGP version of RIVA 128 and TNT in Vogons driver archive do not work with the Riva 128, they are for TNT only.
The correct RIVA 128 (and 128 ZX) AGP driver package is called RIVA128-1.27dW311.ZIP and can be found on Nvidias site.
I tried the driver pack in the archive with a Riva 128 and a Riva 128 ZX (Both cards are versions of STB Velocity 128) and it would not load Windows with those drivers.
The driver pack from Nvidias site works perfectly.

Finally a question.
In the manual for the Dream Wave wavetable board they are talking about two MIDI files used for enabling and disabling the MT-32 emulation in DOS.
I cant find those two MIDI files and I have all software disks mentioned in the checklist. I guess they forgot to include those MIDI files in the software package.
But I guessit is possible to create those MIDI files your self if you know what instruments to use... ?

Edit

I found the missing MIDI files "set_mt32.mid" and set_gmgs.mid. They were hiding on one of the software disks but diddnt get installed.

Edit

The MT-32 emulation seem to work great!

Last edited by Skyscraper on 2013-12-28, 21:02. Edited 1 time in total.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 26 of 40, by gerwin

User metadata
Rank l33t
Rank
l33t
Skyscraper wrote:

One small issue. If I use the 2x multiplier I loose the L2 cache but at 2.5x it works fine.
I have had this CPU running at 133 mhz with L2 cache on another motherboard.
This is a bit strange but I guess this motherboard sees the 2x setting as its fail safe setting...

It is because at the 2x setting the AX6BC BIOS obviously sends the X,X,X or 0,0,0 signal to the CPU at bootup, while you want the BIOS to send X,0,0. See the Table for Klamath, and a similar one for Unlocked Deschutes and Coppermine. That is the downside of jumperfree multiplier settings in any BIOS, and why I prefer jumpers. Jumpers give you full access to all 16 possible multiplier request signals through BF0+BF1+BF2+BF3. All a motherboard can do in this regard is send the Multiplier request signal to the CPU at bootup and hope for the best.

Skyscraper wrote:

I do not seem to have the option to disable the L1 cache. Perhaps a bios update will add that option.

It can be done by a few simple software tools, like throttle.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 27 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t
gerwin wrote:

It is not strange, it is because at the 2x setting the AX6BC BIOS obviously sends the X,X,X or 0,0,0 signal to the CPU at bootup, while you want the BIOS to send X,0,0. See the Table. That is the downside of jumperfree multiplier settings in any BIOS, and why I prefer jumpers. Jumpers give you full access to all 16 possible multiplier request signals through BF0+BF1+BF2+BF3. It is not how ''the motherboard sees the setting", it is about how the CPU interprets the multiplier request signal.

Obviously the board sends the wrong signal.
The strange part is why did the bios programmers think that it would be a good idea to cripple the 2x setting.
If I want to use the 2x setting without L2 cache the non strange method would be to use the bios setup option that inactivates the L2 cache 😀
But then again the board falls back to the 2x setting if I choose a multiplier that is not supported instead of falling back to the default 3.5x multiplier.
So perhaps the bios programmer thought of the 2x multiplier as a fail safe multiplier for unlocked CPUs. Still dosnt explain why he tought that it would be a good idea to cripple the L2.

I am glad I went with an unlocked CPU in any case since one of the first games I tried was Tyrian and @233 mhz which is the speed the computer defaults to the Borland Pascal compiler bug showed its ugly face.
@166 mhz Tyrian works great and the speed is good using the "slow" setting in the game.

General Midi sounds good in Tyrian using the Dream Wave board!

Last edited by Skyscraper on 2013-12-28, 21:00. Edited 2 times in total.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 28 of 40, by gerwin

User metadata
Rank l33t
Rank
l33t

For Aopen it did not matter much what they did with the multiplier interface on the AX6BC, as the retail CPU's at the time were multiplier locked anyways. Aopen may have had a Mendocino Core Engineering Sample CPU which actually used 2.0X as such. Having the L2 Disabled at some settings comes solely from Intel, for unknown reason.

The Classic 4MB Dream Wavetables are my favorite. They sound fantastic.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 29 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t

Yes I agree that it would not matter to Aopen but since the behaviour is diffrent with the 2x multiplier compared to all the rest of the multipliers I would like to think there is a reason.
Your idea about a Celeron engineering sample sounds possible.

Its not a problem in any case.
This build is not primarily for very speed sensitive games but avoiding the Borland Pascal compiler bug is always practical.
Daggerfall is a game I always wanted to try out and I think it has General MIDI support 😀. I just noticed that it is free and downloadable from the Elder Scrolls page.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 30 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t

My week 31 Malay Deschutes p2 400 is unlocked!

I tested it just to make sure it was locked but to my surprise it isnt 😀
I guess there isnt a way to be 100% sure if a CPU is unlocked or not other than trying it 😀

No L2 cache @ 2x multiplier as with the Klamath.
And just as "bestemor" experienced I now do not have any L2 cache using the 2.5x multiplier either.
@ 3x multiplier the L2 cache works fine.
4*112 mhz gives a score of 516 in Speedsys. Quake 1 timedemo1 gets 75.8 fps (software renderer)

I am starting to dislike this Aopen board. Perhaps a board with jumpers would be better.
With a 100 mhz fsb CPU I can not choose any fsb settings under 100 mhz.
I think I can get around this with a jumper that forces AGP to 1:1. (auto, 2/3 and 1/1)
If I do that I think that I have to change video card to get high fsb working. The Banshee should handle 100 mhz+ AGP.
Im also not sure if the PCI dividers gets screwed up.

Another option is to mod the CPU so the board thinks its a 66 mhz fsb CPU.
But I think that will force the AGP to always use 1:1 if the jumper is set to auto. If I set the AGP jumper to 2/3 that will probably remove the FSB settings under 100 mhz even with a 66 mhz fsb CPU...
I will start with just mounting a switch for the AGP jumper so I can change it from outside the box.
I tested Tyrian too see if the Borland bug thingy gets triggered at 2x100 mhz, it does not so ignoring the issue all together is also an option.

One thing I had to fix was that the wavetable board often failed to initialize if the CPU speed was 300 mhz or higher.
It seems to be fixed by using an 8bit ISA I/O recovery time setting of 7 instead of the default 5.

Daggerfall runs smoother with the Deschutes at 4*112 compared to the Klamath at 4.5*75 or 4.5*83.
486 DX2 66 system requirement...
Perhaps the version I am running is modified to run better in DOSbox or something.
I updated it to the latest version using the update included in the download from the Elder Scrolls page.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 31 of 40, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

The main strengths of this board is great stability (praised for having a ton of Japanese caps), fully auto config BIOS, just slot in the CPU and it detects everything and some form of hybernate / sleep function.

It was never praised as a tweaker board, but one that "just works".

An Abit of GB BX-2000 might work better for you?

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 32 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t
Mau1wurf1977 wrote:

The main strengths of this board is great stability (praised for having a ton of Japanese caps), fully auto config BIOS, just slot in the CPU and it detects everything and some form of hybernate / sleep function.

It was never praised as a tweaker board, but one that "just works".

An Abit of GB BX-2000 might work better for you?

Yes the stability seems to be great and the caps are all Rubicon or Panasonic if I remeber correctly.
The Klamath was running Daggerfall @ 4.5 * 83 without a single crash 😀
I could probably run the Deschutes @ 4*124 but 4*112 seens to be enough for Daggerfall.

I have a MSI board that I think will let me set 66 mhz fsb from the bios with a 100 mhz fsb CPU.
But that board needs a recap. The caps are not just a little bit bad, the board needs two resets just to post.
I think those late Abit BX boards are a bit too rare/expensive for my taste.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 33 of 40, by gerwin

User metadata
Rank l33t
Rank
l33t
Skyscraper wrote:

My week 31 Malay Deschutes p2 400 is unlocked!
I tested it just to make sure it was locked but to my surprise it isnt 😀
I guess there isnt a way to be 100% sure if a CPU is unlocked or not other than trying it 😀

That is why I wrote IIRC behind week 30. I haven't seen enough examples to pin down the exact week.
Maybe the lock was implemented halfway week 31, or maybe week 32?
Good news anyways. These unlocked Pentium II's with 5.0 ns cache are great to toy around with.

Skyscraper wrote:

Quake 1 timedemo1 gets 75.8 fps (software renderer)

320x200 I assume. For higher resolutions be sure to load 'MTRRLFBE LFB WC' for Write combining through the Pentium II.

Skyscraper wrote:

I am starting to dislike this Aopen board. Perhaps a board with jumpers would be better. With a 100 mhz fsb CPU I can not choose any fsb settings under 100 mhz.

I have on of these AX6BC's in storage too. At least it has the AGP divider jumper which you can set from "auto" to "2/3 manual". Which will keep the AGP frequency in check (max 89MHz at 133MHz FSB). PCI remains at 33 MHz at 66/100/133 MHz FSB, but will deviate with any other FSB speed. SoftFSB for Windows works with this board and will allow you to go down to 66.
Modifying a CPU from 100 to 66 FSB default is similar to this 133 to 100 mod, but one needs to mod pin B21 instead. Unfortunately the Pentium II SECC-1 cartridge is not mod friendly at all.

Skyscraper wrote:

One thing I had to fix was that the wavetable board often failed to initialize if the CPU speed was 300 mhz or higher. It seems to be fixed by using an 8bit ISA I/O recovery time setting of 7 instead of the default 5.

You mean that midi will not work at all, after some bootups with default BIOS settings? That is new to me. It may be caused by the host sound card which is to ground the midi DB reset pin at boot, for a second or so.
Your daughterboard has the earlier SAM 9203 chip, instead of the SAM 9233. I also noticed the 9203 is a little slow: some midi files have program changes which it cannot keep up with, resulting in a piano instrument where it should have changed to another intstrument. The descend 1 extracted midi tracks do this. I modified a few of these tracks so they play properly on a 9203.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 34 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t
gerwin wrote:

320x200 I assume. For higher resolutions be sure to load 'MTRRLFBE LFB WC' for Write combining through the Pentium II.

Yes that was with the default 320*200 setting. With write-combining 320*200 gets 83 fps and 640*480 gets 38.5 fps with sound.
I diddnt seem to have an issue running Fastvid with EMM386 loaded. My Windows 3.11 do not seem to like UMBPCI at all so I am sticking with EMM386.

gerwin wrote:

I have on of these AX6BC's in storage too. At least it has the AGP divider jumper which you can set from "auto" to "2/3 manual". Which will keep the AGP frequency in check (max 89MHz at 133MHz FSB). PCI remains at 33 MHz at 66/100/133 MHz FSB, but will deviate with any other FSB speed. SoftFSB for Windows works with this board and will allow you to go down to 66.
Modifying a CPU from 100 to 66 FSB default is similar to this 133 to 100 mod, but one needs to mod pin B21 instead. Unfortunately the Pentium II SECC-1 cartridge is not mod friendly at all.

Even removing the stock cooler mounted on the Klamath with threaded rivets was a bit of a hassle since the 1.5mm Allen key diddnt fit very well.
The Deschutes had a removable cooler already mounted so at least things were moving in the right direction.
Or perhaps not since the SECC-2 cartridge the Pentium 3 (and late P2 450) uses really sucks as it often wont fit in a Slot-1 slot without removing the "CPU supports".

gerwin wrote:

You mean that midi will not work at all, after some bootups with default BIOS settings? That is new to me. It may be caused by the host sound card which is to ground the midi DB reset pin at boot, for a second or so.
Your daughterboard has the earlier SAM 9203 chip, instead of the SAM 9233. I also noticed the 9203 is a little slow: some midi files have program changes which it cannot keep up with, resulting in a piano instrument where it should have changed to another intstrument. The descend 1 midi tracks do this. I modified a few of these tracks so they play properly on the 9203.

I got the message FM failed to initialize when the Sound Blaster Pro initialization (SPM.EXE) was loaded.
Since the FM volume controls the Wavetable volume I think its the Wavetable that fails not the basic FM synth. The error message is just a bit unclear.
The faster the CPU speed the more often it happend. With the Deschutes @ 400 mhz or more it failed every time. ´@250 mhz or below it never failed.
With the changed 8bit ISA recovery setting it has not failed to initialize a single time @448 mhz.

[edit]

Perhaps it is the OPL3 that fails if the ISA 8-bit I/O Recovery Time is not set to a high enough value with a fast CPU.
OPL3 or wavetable dosnt matter, the initialization software must be poorly written and simply talks too fast to the card.
I found some more information in this thread OPL3 output problem of the Sound Blaster Pro 2 (and clones) on 'faster' CPUs on DOS.
And even more in this thread Slowing down ISA bus: recovery time, ISA bus clock.

[/edit]

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 35 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t

The tinkering with this sound card ends here for now.
I am very pleased with it but the system I am testing it in is not an ideal DOS 6.22 computer.

I will probably use the sound card in the slow half of the Leviathan dual system build which will be a dedicated DOS 6.22 + Windows 3.11 system.
The other halft of the build will handle really demanding dos games under DOS 7 and DX3 to DX6 + Glide games under Windows 98.
That builds completion has been somewhat delayed while I have been waiting for a K6-3 CPU to arrive... By now I am pretty sure it will not arrive at all so I have sourced another one.

The Aopen AX6BC system will be used for testing old AGP video cards in Windows 98.
I have too many video cards I never even tested once to see if they are working as they should.
If I am pleased with the motherboards performance under Windows 98 it will replace my Abit BH6 in my Windows 98 build for DX7, DX8 and OpenGL games.
The 133/4 PCI divider lets the board run CPUs that uses 133 mhz FSB and that gives me more CPU options for the fast Windows 98 build.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 36 of 40, by NitroX infinity

User metadata
Rank Member
Rank
Member
Skyscraper wrote:
NitroX infinity wrote:

For the soundcard, did you try looking up the FCC ID?
http://transition.fcc.gov/oet/ea/fccid/

As for the pin that wont let go; use pliers to squeeze it then pull/puss it through the hole. Always does the trick for me.

I did try squeezing with pliers and pulling and pushing with force.
Its not that I am afraid that I will hurt the card but the pin looks unique, and therefore hard to replace and Im not sure it will go through the hole in one piece.

Nah, those pins have been used on Quantum Obsidian X-16/X-24 videocards too.

If it does damage to beyond usability, just use a tie-wrap. It's just there to make sure the daughterboard doesn't fall off if the case gets banged around.

NitroX infinity's 3D Accelerators Arena | Yamaha RPA YGV611 & RPA2 YGV612 Info

Reply 37 of 40, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++
Skyscraper wrote:

That builds completion has been somewhat delayed while I have been waiting for a K6-3 CPU to arrive... By now I am pretty sure it will not arrive at all so I have sourced another one.

Exciting!

It's such a great CPU and changing the multiplier from 2x to 6x through SOFTWARE is really cool. I put that CPU back in my main time-machine after using a Pentium for a while and it's great.

Without caches it basically gives you a 386DX-25 (with L1 and L2 disabled) and 486SX-25 (with L1 disabled but L2 on). Two very nice performance levels. And if you want to play Doom or Decent, just enable all the caches and for SVGA grunt put the multiplier to 6x.

If the FSB is set to 66 MHz you can change the clocks from 133 to 400 MHz 😀

Made a video here:

http://www.youtube.com/watch?v=Ur0WtduA2XI

For playing MS-DOS games I haven't come across a better solution yet, apart from DOSBox 😀

It's also quite easy to work with, less quirks than with 386 and 486 gear. And you can also choose between an authentic AT system or mixing new and old like I do with ATX based Super Socket 7 gear.

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 38 of 40, by Skyscraper

User metadata
Rank l33t
Rank
l33t
Mau1wurf1977 wrote:
Exciting! […]
Show full quote

Exciting!

It's such a great CPU and changing the multiplier from 2x to 6x through SOFTWARE is really cool. I put that CPU back in my main time-machine after using a Pentium for a while and it's great.

Without caches it basically gives you a 386DX-25 (with L1 and L2 disabled) and 486SX-25 (with L1 disabled but L2 on). Two very nice performance levels. And if you want to play Doom or Decent, just enable all the caches and for SVGA grunt put the multiplier to 6x.

If the FSB is set to 66 MHz you can change the clocks from 133 to 400 MHz 😀

It's also quite easy to work with, less quirks than with 386 and 486 gear. And you can also choose between an authentic AT system or mixing new and old like I do with ATX based Super Socket 7 gear.

I would like to use the K6-3 450 in the fast part of the build so I can use more than 256 mb memory on my SS7 board with 256 mb cacheable range.
The fast part will use a Geforce2/4 MX + Voodoo 2 and an AWE64 so it is not really optimal for very old games.
I do not think the normal K6-3 will allow multiplier changes on the fly anyway?

The slow part will use a Pentium slow enough to not trigger any Borland Pascal compiler bugs but a K6-3+ @ 100-300, 133 - 400 or perhaps 150 - 450 mhz is clearly a better option.
Beeing able to manipulate multipliers and caches with software sounds fun.
This system will use a S3 Trio+ video card and the soundcard I have been tinkering with in this thread.

New PC: i9 12900K @5GHz all cores @1.2v. MSI PRO Z690-A. 32GB DDR4 3600 CL14. 3070Ti.
Old PC: Dual Xeon X5690@4.6GHz, EVGA SR-2, 48GB DDR3R@2000MHz, Intel X25-M. GTX 980ti.
Older PC: K6-3+ 400@600MHz, PC-Chips M577, 256MB SDRAM, AWE64, Voodoo Banshee.

Reply 39 of 40, by vetz

User metadata
Rank l33t
Rank
l33t
Robin4 wrote:
That socket 7 586F52 is made by freetech.. http://www.tacktech.com/download.cfm?file=manuals/586f52.txt http://www.motherboards […]
Show full quote

That socket 7 586F52 is made by freetech..
http://www.tacktech.com/download.cfm?file=manuals/586f52.txt
http://www.motherboards.org/mobot/manuals/Freetech/586F52/

forum user `VETZ` as also one..

That soundcard you found is very nice, and very hard to find these days.. I think no one have one of these and are very rare..

I never got it running until today 😀 Apparently it is a bit picky with RAM.

I have an earlier revision of the board with Socket 5 with no PS/2 header 🙁. Do you have the J26 and J27 jumpers for changing multiplier and PS/2 header?

I noticed some really neat with my revision. You can underclock the CPU down to 25, 33 and 50mhz! As well as overclocking to 120mhz. Incredible!

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes