Sorry for the double post, but I can't find how to edit my first post.
I messed up the link to Vlasks site. Here's right link: https://www.vgamuseum.info/index.php/cpu/item … 3-elsa-gloria-4

Well I wanted to add, that you have to add the GLoria-4 to the System after installing Win98.
Then on first boot you have to abort the automatical installation when it prompts you to insert the Win98 CD.

Otherwise the card will be recognized as a S3 968 and after a restart your CRT might die...
Because with normal S3 968 drivers the card wants to default to a 17kHz signal with 35Hz refresh rate, according to the out of sync message of my TFT.

Ok, some updates.
After some trouble with automatic standby, the OpenGVS RealWorld benchmark finished in blistering fast 3 hours and 20 minutes in 640x480 and 32Bit.

Sorry for the blurry picture. I've recorded the whole sequence over a VGA to HDMI adapter and a HDMI video capture card.
It works fine when recording in 640x480, but for this screenshot I had to switch to 1024x768.

To my surprise UT99 GOTY works out of the box in OpenGL.
Well with less than 1 FPS it works. 😉

I'm now trying to make a video about this card, as there are currently none.

EDIT: I was wrong here! These numbers apply to the Microsoft software OpenGL renderer, see screenshot above.

I wonder what that card is equivalent too if a modern cpu emulated the OpenGL calls.

No directx right?

Oh dear. Today I learned that I'm actually pretty stupid.
I thought for shure that the GLint 300SX was doing the OpenGL heavy lifting, because the 3DLabs demos were running fine.
Actually Vlask was right all along and the GLint does absolutely nothing in Win98.

I just now noticed, that the OpenGL Version says "GDI Generic". So it's rendered by the CPU.
To confirm my suspicion, I pulled the GLoria-4 out and plugged in a WINNER Pro with the same 2D S3 968 chip, but without the GLint.
And lo and behold: GLQuake and UT still work in OpenGL.

I never knew that Win98SE came with an OpenGL wrapper. It even supports Antialiasing.
Sorry to waste everybodys time!

Well, here at least some screenshots of the 3DLabs demos:

Revenge, oh sweet revenge!

While I wasn't able to get OpenGL to work on Win98, I was able to get it to work on WinNT4.0, using the drivers I attached in the first post.

I've uploaded a couple videos to YouTube showing the GLint 300SX in action, if you want to call it that:
3Dlabs GLint 300SX from 1994 running Quake III
3Dlabs GLint 300SX from 1994 running Unreal Tournament
3Dlabs GLint 300SX - Official OpenGL 3D Demos

Here some benchmark numbers:

GLQuake:
320x200x16Bit: 4,4 FPS
640x480x16Bit: 1,1 FPS

Quake III Arena:
320x240, 16Bit, low detail - 3.8 FPS
640x480, 16Bit, low detail - 1 FPS
1024x768, 16Bit, high detail - 0.3 FPS

Half-Life Opposing Force Benchmark on Voodooalert
640x480x16Bit: 0.728 FPS

System:
Pentium 3, 1000 MHz, 256kB cache
Asus CUBX, 440BX
Soundblaster 128
WinNT4.0

Those NT4 scores look worse than the software rendering scores you posted earlier. Maybe that's to be expected since the Pentium3 1ghz is far more powerful a cpu than would have been available in 1996.

RaVeNsClaw wrote on 2025-04-09, 19:00:

WinNT4.0

Very interisting GPU, I develop SIV and am wondering if it will report your 3DLabs GLint 300SX on WinNT4.0 and if so also report the clocks, please will you try it?

If you run SIV32L.exe the GPU should be on the initial screen and Menu->Hardware->GPU Info Info should report more detail.

There are some notes about systems requirements , etc. in SIV support for 386/486/586 class + Alpha CPUs and 3dfx + S3 + SiS + Matrox + XGI + old ATI + NVidia GPUs - Testing Help, but I expect your NT4 system will easily meet them.

Unfortunately, the same thing happens as with Everest Home 2.20: It completely locks up the system.

Here the debug log:

Here the DBGGPU log:

Just a guess:
The S3 968 and the GLint 300SX are on the same PCI port, without a PCI bridge.
I think this might cause the problem.

Also in the device manager only the VIRGE is detected, but VIRGE drivers don't work.
Only installing the ELSA drivers works.

RaVeNsClaw wrote on 2025-04-10, 16:35:

Unfortunately, the same thing happens as with Everest Home 2.20: It completely locks up the system.

Oops, and sorry about that and I will look into why. Well done for doing -DBGINI and SIV seems to be doing something to the ELSA Gloria, but I don't have enough information to make an informed guess as to what, I need to see what [ PCI Bus ] reports.

What happens if you do SIV32L -GPUS=0? This will start SIV without trying to detect the GPUs at all. Assuming it starts please do Menu->File->Save Local and post the two save files. If there is an issue with one of the sections do Menu->File->Save->Save Selected and deselect that section before pressing [Start].

Please ZIP the SIV_DBGOUT.log along with the two save files.

I tried "SIV32L -GPUS=0", no Luck.
With a GLint MX R1, SIV runs on this system, so the problem is definetly the GLoria 300SX.
Is there something else I can try?

I'm going to test a few exotic GPUs (pre 1997) in the next few days in this system, with the intention to test OpenGL performance.
I'm going to post the reports to your SIV support thread.

RaVeNsClaw wrote on 2025-04-10, 18:54:

Is there something else I can try?

Yes, there are lots of options try SIV32L -GPUS=0 -NOGPUS -DBGGPU -DBGPCI > SIV_DBGOUT.log | more (-GPUS=0 before -NOGPUS ) if SIV starts then see what happens with [ PCI Bus ] then Menu->Machine->GPU Detail->GPU VGA Dump.

When they happen lockups are quite tricky, fortunately it's not that often. Once we figure out what triggers the lockup stopping it should be easy enough. I have had these sort of issues when developing device drivers for clients and their hardware acting up, then I stepped through with the kernel debugger, found the issue and got the device hardware/firmware fixed.

Sorry for the delay!
With those parameters it starts, but it crashes when I try to save a report at this step:

Attached you will find partial SIV reports and the GPU dump.

red-ray wrote on 2025-04-10, 19:24:

If SIV starts then see what happens with [ PCI Bus ]

RaVeNsClaw wrote on 2025-04-13, 11:02:

With those parameters it starts, but it crashes when I try to save a report at this step:

Thank you for the report and progress. Looking at the last line of the save files in both cases it's [pci-bus], did you try pressing [PCI Bus], I guess not.

It's very rare of [pci-bus] to trigger a crash and when it does it's invariably questionable hardware, the next section would be [pci-all] which is more likely to and also implies do [pci-bus].

As much of the save is missing I can't check some things I need to check so please do Menu->File->Save->Save Selected, uncheck both [pci-bus] + [pci-all] then press [Start].

red-ray wrote on 2019-08-22, 10:17:

[*]If one of the save sections has an issue do Menu->File->Save->Save Selected and deselect that section before pressing [ Start ].

Ok, that worked, but I also had to uncheck [pci-bars], [io-ports] and [audio] or it would crash.

Also I ran this on the "SIV32L-V5.81-CUBX-04-MUX" version.

RaVeNsClaw wrote on 2025-04-17, 18:40:

I also had to uncheck [pci-bars], [io-ports] and [audio] or it would crash.

Thank you and yes, those sections do [pci-bus] internally, I should specified them, but forgot. I am about to pop out for a beer so will look at the save file tomorrow and report back.

Have you ever tried to run Windows 2000 on a system with this GPU installed? I wonder if PnP would trigger the issue.

Update: The SPD is back, but the AS99127F was not reported as the name ended up as ASUS * rather then ASUS CUBX , the next SIV beta should fix this.

Do you feel I need to do anything to make SIV better tolerate the GPU? I am wondering about a qualifier, -PCI-SKIP=<bus>[:<bus>]... so [pci-bus] would skip the specified buses and GPUs on those busses would not get reported.

I don't have enough background knowledge to answer that?
If it is just this this card, then no. I mean Everest also crashes.
Clearly the weird implementation of this two chip solution without PCI bridge is at fault.

If anything I would criticize the menu layout of Siv. 😉
If you didn't tell me that I should open the menu, I would never even have found out that there is one, for which I have to do a right click on the border of the window.
I thought I had to click on the buttons on the bottom, but they often only lead in endless circles.
Also I didn't notice the "Save Selected" function before, because it's not on top the context menu with the important functions, but buried in the middle.

Don't get me wrong. I think the program has fantastic functionality.
But it is difficult to use. It's like navigating regedit, but via a context menu instead of the tree on the left side.

Please Excuse my rambling, but when you ask someone unqualified for their opinion, you get an unqualified answer. 😜

RaVeNsClaw wrote on 2025-04-17, 23:39:

If anything I would criticize the menu layout of Siv. 😉

Thank you, I suspect you don't mean layout, but rather the fact that there is no menu bar. I ponder adding one, but thus far have not as on Windows Vista and later there are split buttons you can use, you can also Right/Click on the buttons and the tray icon. Have a look at the https://www.softpedia.com/get/System/System-I … er.shtml#sgal_2 screen shots.

I feel once you know there is a menu it's easy enough, have you noticed there is Menu->Help->View Menus / Menu->Edit->View Menus ?