Great builds with really nice components!

I got to try to get my own socket 8 motherboard to work again, it is a bit unhappy at the moment 😒

H3nrik V! wrote on 2023-12-06, 07:49:

That capacitor's just crazy stuff. Looks like it never got soldered and cut flush. But maybe at that time, manaual hand mounting was still used to some extend?

It did have some solder in it, just one of the holes was mostly empty. It does seem like it came from the factory that way. I'm a bit mystified as I don't think I've seen a board with either partially soldered components and uncut leads.

OSkar000 wrote on 2023-12-08, 09:46:

Great builds with really nice components!

I got to try to get my own socket 8 motherboard to work again, it is a bit unhappy at the moment 😒

Thank you! Good luck with your own build and getting it working. These old boards can be temperamental!

A bit more progress on the 1996 builds...

First, I confirmed the booting instability on the Pentium 200 motherboard is related to the Bank 0 RAM slots. With RAM installed in Bank 0 it will intermittently fail to POST. I tried various RAM sticks and that made no difference.

It works perfectly with Bank 1, so for now I'm just leaving Bank 0 unpopulated. If I feel ambitious I might try to further diagnose what the issue is, but for now I'm leaving it.

Second, I set up DOS and started to play around with sound card configurations. This is where I ran into a lot of issues with the AWE64. I've used the AWE64 and GUS Extreme (and Roland SCC-1) combo in other builds and have managed to get it fully working in those setups. Not so in this setup.

In prior setups, I installed the AWE64 first to its default settings and then configured everything else based on free resources. For these builds I wanted to configure the GUS Extreme's ESS1688 chip to a default Sound Blaster config (i.e. A220 I5 D1/5) and set the AWE64 around that.

I discovered that the Creative Labs PnP drivers will reconfigure the card on their own if they detect something they don't like. This includes overwriting the BLASTER environmental variable to those default settings. But the GUS Extreme also needs a BLASTER setting and the Creative Labs drivers overwrites both variables. In doing so the AWE64 would initialize, but the GUS Extreme would fail.

I also ran into issues with some games either not working properly or the mixer producing really quiet output. In testing Warcraft 2, digital audio would default to a minimum volume while AWE32 music wouldn't play at all.

I am also unable to get CD Audio playback through the AWE64 at all with the Creative drivers. The mixer application seems bugged as it doesn't seem to save settings. And there is the aforementioned issue where volume settings change on their own depending on what game I am testing.

If I use UNISOUND for enabling the AWE64 everything works perfectly. I can set any hardware and mixer settings, and everything works perfectly including CD audio. The AWE64 also works fine under Windows 95.

Because I'm trying to make these more "period correct" builds, I didn't want to rely on UNISOUND. But those Creative Labs DOS drivers are a hot mess.

My options are now:

1) Start from scratch and leave the AWE64 at the default settings. Configure the GUS Extreme and Roland SCC-1 based on available resources. Pray that everything will work.
2) Skip the AWE64 for DOS and just use the GUS Extreme / Roland combo.
3) Forego period correctness and use UNISOUND instead. It just works.

Shponglefan wrote on 2023-12-10, 18:33:

Because I'm trying to make these more "period correct" builds, I didn't want to rely on UNISOUND. But those Creative Labs DOS drivers are a hot mess.

While I don't have a GUS, I never encountered any issues when pairing my AWE64 together with another sound card, usually a Yamaha YMF719. Both cards were using their stock drivers and everything worked fine, as long as they were configured to utilize different resources.

I do remember needing two separate .BAT files for selecting which BLASTER variable I wanted to set, but otherwise, it was smooth sailing.

Joseph_Joestar wrote on 2023-12-10, 18:43:

While I don't have a GUS, I never encountered any issues when pairing my AWE64 together with another sound card, usually a Yamaha YMF719. Both cards were using their stock drivers and everything worked fine, as long as they were configured to utilize different resources.

I had success in a previous build with this combo and the native AWE64 drivers. That's why I baffled why I'm encountering so many issues this time around.

Maybe I just need to nuke the whole thing and start over.

I do remember needing two separate .BAT files for selecting which BLASTER variable I wanted to set, but otherwise, it was smooth sailing.

I do the same thing to toggle the BLASTER variable for game setup programs.

I reinstalled the DOS sound drivers. In the process I downloaded a fresh set of CTCM and SB Basic drivers from Vogons just to be on the safe side.

This time I installed the AWE64 at the default config 0 settings, and tested everything thoroughly. Everything seemed to work including Warcraft 2 digital audio and music (at proper volumes) and CD audio. I then changed the MIDI port to P300 and re-installed the Roland SCC-1 at default P330. Installed the GUS afterwards and configured it based on free settings.

Installed things on the Pentium 200 first, then copied over the install from the P200 to Pentium Pro 200. Testing on the Pentium Pro, everything seemed to work as well.

AUTOEXEC config is as follows (Roland SCC-1 is set to IRQ 2 and P330):

1REM -------- AWE64 init ----------
2SET SOUND=C:\SB16
3SET BLASTER=A220 I5 D1 H5 P300 E620 T6
4SET MIDI=SYNTH:1 MAP:E MODE:0
5SET CTCM=C:\CTCM
6C:\SB16\DIAGNOSE /S
7C:\SB16\AWEUTIL /S
8C:\SB16\MIXERSET /P /Q
9C:\CTCM\CTCU /S
10REM -------- AWE64 ends ----------
11
12@REM ===== Gravis initialization (4.11) =====
13@SET ULTRASND=260,7,7,3,3
14@SET BLASTER=A240 I7 D3 T4
15@SET ULTRADIR=C:\GRAVIS\ULTRASND
16@C:\GRAVIS\ULTRASND\ULTRINIT.EXE -dj
17@REM ===== Gravis initialization ends =====

After getting everything working in DOS, started on installing Windows 95 drivers. For these builds, I'm using Windows 95 OSR2.

Pentium 200 system worked fine. Got the Matrox Mystique and Diamond Monster 3D installed without issue.

Then I moved over to the Pentium Pro 200 and immediately ran into problems with the Matrox Mystique. I installed the latest Win 9x drivers from Matrox (v4.12.013). Upon rebooting the entire system locked up at a blank desktop and frozen mouse icon.

I did some Googling and found this thread about this issue, but related to PCem. In this case, it describes issues with the Matrox drivers, Windows 95 and the Intel 430VX chipset. Another thread noting similar issues was found regarding 86Box.

Interestingly I couldn't find anything on this issue on real hardware.

I tested two different Matrox Mystiques with BIOS versions 1.2 and 1.3 respectively. Both hang. I also tried an older Matrox Millennium and had the same issue.

The board I'm using (DataExpert EXP8P61) has BIOS version 1.0. There's a newer 1.1 version available from RetroWeb, so I may update the BIOS and see if it does anything. I might also try updating one of the Matrox cards to a later BIOS and see if that fixes anything.

If anyone knows an immediate solution to this problem it would be welcome. Otherwise, I have a few things to try out, but it's going to take some time.

Shponglefan wrote on 2023-12-17, 23:59:

If anyone knows an immediate solution to this problem it would be welcome. Otherwise, I have a few things to try out, but it's going to take some time.

Might be a similar issue I had 3 years ago:
Re: ASUS Media Bus - PCI-AS2940UW with SCSI & Vibra16

vetz wrote on 2023-12-18, 09:45:

Shponglefan wrote on 2023-12-17, 23:59:

If anyone knows an immediate solution to this problem it would be welcome. Otherwise, I have a few things to try out, but it's going to take some time.

Might be a similar issue I had 3 years ago:
Re: ASUS Media Bus - PCI-AS2940UW with SCSI & Vibra16

Thank you! This put me on the right track.

The problem was the Windows 95 installing a default SB16/AWE32 driver upon setup. I hadn't bothered to install the AWE64 driver before installing the Matrox drivers. Turns out the SB 16/Matrox driver combination did not work.

I re-installed everything starting with the AWE64 driver followed by the Matrox driver and now it's working just fine. 😁

All Windows 95 drivers are installed including the chipset driver, Matrox Mystique, Monster 3D, AWE64 and Roland SCC-1 (MPU-401 Compatible).

Basic testing has been completed and everything seems to be working. Next up will be installing some games!

Hit another wall with the Pentium Pro setup and Windows 95.

Now it's the the Diamond Monster 3D that doesn't want to work. And I'm getting audio artifacts in games on top of things.

Tested both Quake 2 and Half-Life and neither game worked. Interestedly, Half-Life did load and run the opening level for about 5 seconds before freezing. Quake 2 just locked up at the main menu. Even the Diamond Monster 3D test graphic would lock up.

Tested various driver and DirectX combinations to no avail. I even tried three different Diamond Monster 3D cards (all confirmed working). Still nada.

This DataExpert EXP8P61 motherboard is really not happy under Windows 95 with the hardware I'm running.

I then plugged everything into a Supermicro P6SNE motherboard and re-ran the whole set up. Everything now worked. Got Quake 2 installed and it ran just fine. The only issue is for some reason it won't let me change the audio quality to "high". But otherwise it ran just fine with the Diamond Monster 3D.

Also Tomb Raider with the 3DFX patch and that worked as well.

Unfortunately the Supermicro board is 2mm too wide to the fit in the current AT case I'm using. The power switch barely blocks it.

I think I have a case that may work, although it's an older full tower case and not exactly quite as period correct as I want. It should work until I can find something more suitable.

Nice project. Love how period correct the build is and the synergy between components. I'm doing something similar for a bit of a large span of time. Creating high-end (game) builds from 1993 to 2011. Currently complete or nearly with 1993, 1997, 1998, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2009,2011. Just have to get a few more period correct PC cases.

PleaseHelpMeAdmin wrote on 2023-12-20, 09:46:

Nice project. Love how period correct the build is and the synergy between components. I'm doing something similar for a bit of a large span of time. Creating high-end (game) builds from 1993 to 2011. Currently complete or nearly with 1993, 1997, 1998, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2009,2011. Just have to get a few more period correct PC cases.

Thank you!

That's an impressively ambitious list of builds you've got going! Would love to see some pics/specs when if you get a chance to post them.

Got the Pentium Pro 200 installed in a different case. It's not the case I wanted to use, but it's only AT case I have on hand that will fit this particular motherboard. Being an early 90s AT case, it unfortunately lacks spots for either intake or exhaust fans.

I also moved the Pentium 200 build into the case previously used by the Pro 200. Since the 80mm intake fan is more in-line with the VRM and CPU heatsinks, this should provide better cooling for that set up.

The 1996 builds are almost done. Only thing I'm waiting on is some EDO RAM for the Pentium 200. Currently it has FP RAM. EDO should give it a marginal performance boost before I do any benchmarking.

While waiting for that RAM to arrive, I've turned my attention to my planned 1993 build: a 486 DX2-66.

Most 486 builds I've seen tend to focus on later hardware circa 1994 or even 1995. I think trying to do a more period correct 1993 build will be an interesting exercise.

For this build I have an early 90s case I plan to use. Unfortunately this one lacks the MHz display, but otherwise fits that early 90s aesthetic.

For the motherboard, I'm planning to use this ECS UM486V board. Most of the Socket 3 boards I have are all circa 1994, but this one was built early 1993.

For video cards, I currently have a Cirrus Logic CL-GD5428 and a Trident TGUI9400CXi card. I'll test both and see which one yields better performance. I've read these particular Trident cards are decent performers and tend to be underrated.

For sound cards, I'm thinking of going with a Pro Audio Spectrum 16, Gravis Ultrasound, and Roland SCC-1 combo. It's something a little different than just sticking in a typical Sound Blaster.

The GUS is going to need a bit of work. It lacks a working GF1 chip, though I have another card I can borrow one from. The line out socket is also quite loose, so I'll need to repair that as well. I might de-solder one of the other sockets and swap them around.

Did some testing with that ECS UM486V motherboard and it seemed unusually slow. Norton Sysinfo's CPU Benchmark gave a score of 99, 3D Bench a score of 38, and Doom performed at about 22.5 FPS.

This is noticeably worse performance than I would expect from a typical 486 DX2-66. I did some tests just to make sure both cache's were enabled and working, things like ROM shadowing as enabled, etc. Still not entirely sure why it's so slow.

Digging through my parts bins and I found this Genoa Systems Turboexpress motherboard also manufactured in 1993.

It yields performance more in line with what I expected. With default BIOS settings Norton Sysinfo posted a CPU benchmark of 131.9. With more aggressive RAM settings, it scored 143.8. 3D Bench topped out at 45.4 and Doom's FPS was over 26.

Set up the bulk of the hardware including the three sound cards (GUS, PAS16 and SCC-1) and have been doing some testing.

I've run into a few issues so far:

1) After deleting the partition on a Sandisk 512MB card, FDISK wouldn't work with it. The BIOS would detect it just fine, but FDISK complained that it couldn't read the drive. A Wintec 512MB card worked fine, though.

2) Speedsys won't work and locks up at the Y2K test. Some Googling reveals this is a problem with some setups and there doesn't appear to be a workaround.

3) Doom v1.2 would randomly crash. Typically it would crash an invalid sprite number error. A couple times it just locked up and required a hard reboot. It seems more prevalent with the PAS16, although it did happen a couple times with the GUS. I haven't figured out exactly what is causing this yet.

4) PAS16 works but suffers from reversed stereo channels. Fortunately the GUS has proper stereo.

I also think the motherboard may have had the wrong BIOS installed. I flashed a new chip using a BIOS downloaded from TheRetroWeb. After doing this, FDISK was able to recognize the Sandisk 512MB card.

I've initially been testing everything with a DOS 6.0 setup. I'm going to try a DOS 6.2 install next and see if that changes anything.

I've also been experimenting with RAM timing which might be responsible for some of the Doom instability (?). I'm going to keep testing things and see if I can get this system to behave stability.

I also plan to source an earlier VLB controller card to keep this system more 'period correct'. And since the PAS16 card has a SCSI interface, I'm also going to look at getting an early SCSI CD-ROM drive (circa 1993) and see how that goes.

Shponglefan wrote on 2023-12-24, 19:46:

2) Speedsys won't work and locks up at the Y2K test. Some Googling reveals this is a problem with some setups and there doesn't appear to be a workaround.

Out of curiosity, does it make any difference if you set the date in the BIOS to something pre-2000? Maybe mid 1994 or some such?

3) Doom v1.2 would randomly crash. Typically it would crash an invalid sprite number error. A couple times it just locked up and required a hard reboot. It seems more prevalent with the PAS16, although it did happen a couple times with the GUS. I haven't figured out exactly what is causing this yet.

I vaguely remember reading somewhere that Doom stopped supporting the GUS hardware mixing after a certain patch. If you have access to it, try using the retail release of the game and see if it makes any difference with regards to the crashing.

EDIT - fount the relevant thread: Re: DOOM 1.1 GUS hardware mixing

Joseph_Joestar wrote on 2023-12-24, 19:53:

Out of curiosity, does it make any difference if you set the date in the BIOS to something pre-2000? Maybe mid 1994 or some such?

No. I'd left the default date of Jan 1, 1980. Changing the date to other dates didn't make a difference.

I also tried doing a clean boot (no CONFIG.SYS or AUTOEXEC.BAT) and that didn't matter either.

I vaguely remember reading somewhere that Doom stopped supporting the GUS hardware mixing after a certain patch. If you have access to it, try using the retail release of the game and see if it makes any difference with regards to the crashing.

EDIT - fount the relevant thread: Re: DOOM 1.1 GUS hardware mixing

Most of my testing wasn't using the GUS. I was using the PAS16 for digital sound and Roland SCC-1 for music.

The only versions of Doom I have at the moment are 1.2 and 1.9 (Ultimate Doom). I'm currently re-installing everything with DOS 6.2 so I'll do further testing once that is done.

Shponglefan wrote on 2023-12-24, 21:48:

The only versions of Doom I have at the moment are 1.2 and 1.9 (Ultimate Doom). I'm currently re-installing everything with DOS 6.2 so I'll do further testing once that is done.

From what I can see, the shareware version of Doom 1.1 is still available for download on Doomworld. Might be worth trying that.