lizard78 wrote on 2024-02-10, 23:01:

Hi all, So I've been developing a game with retro hardware support and I added a separate path (mostly for 'completeness', the p […]
Show full quote

Hi all,
So I've been developing a game with retro hardware support and I added a separate path (mostly for 'completeness', the performance is terrible) for OpenGL 1.0. The main renderer uses 1.1 / 1.2 features. Problem is, I'm having a hard time finding something to test 1.0 out on. I'm wondering if there are any video cards that had a complete OpenGL 1.0 ICD? I'm not talking about MiniGL or OpenGL 1.1 here. A few cards I have:
Nvidia Riva --> 1.1
3dfx Voodoo 3 --> 1.1 (full ICD, not minigl)
ATI RAGE Pro --> 1.1

I read somewhere that some drivers only reported 1.0 even though they mostly supported 1.1, but no mention of which cards these were specifically.

Assuming this accurate, https://stackoverflow.com/questions/63065312/ … -0-not-even-1-1

Anything released before 1997 had to be 1.0, AFAIU .

Cool, didn't know that!

How about the Elsa Gloria XXL?
It had that Permedia/GLINT processor and an S3 Virge 325.

The Virge was doing VGA on DOS and provided Windows 95 acceleration (2D+Direct3D).

On Windows NT, the Permedia/GLINT proccesor was in use (OpenGL support fir CAD, no Direct3D).

Speaking under correction, though. 😂

Edit: Correction, GLiNT MX + GLINT Gamma..

At the point of 1.0 would mean professional industrial 3D on NT (i.e. big Intergraph cards in the 5 figures).... basic texturing was an extension

If you wanted to go "below 1.1 spec" for the consumer space, you'd have to figure out how to target miscellaneous MiniGLs made for first gen 3d hardware and have fun workarounds on the lack of blending functions, depth, clipping, etc. 😀 Quake3's public test version was the game that got every vendor scared straight into 1.1 compliance in spring 1999, MiniGLs died right then.

Jo22 wrote on 2024-02-11, 00:11:

Cool, didn't know that! […]
Show full quote

Cool, didn't know that!

How about the Elsa Gloria XXL?
It had that Permedia/GLINT processor and an S3 Virge 325.

The Virge was doing VGA on DOS and provided Windows 95 acceleration (2D+Direct3D).

On Windows NT, the Permedia/GLINT proccesor was in use (OpenGL support fir CAD, no Direct3D).

Speaking under correction, though. 😂

Edit: Correction, GLiNT MX + GLINT Gamma..

Never heard of this card - it's some sort of professional graphics card I take it? Reminds me, I also forgot to mention I have a Permedia 2 as well in my original post... also 1.1.

leileilol wrote on 2024-02-11, 03:22:

At the point of 1.0 would mean professional industrial 3D on NT (i.e. big Intergraph cards in the 5 figures).... basic texturing was an extension

If you wanted to go "below 1.1 spec" for the consumer space, you'd have to figure out how to target miscellaneous MiniGLs made for first gen 3d hardware and have fun workarounds on the lack of blending functions, depth, clipping, etc. 😀 Quake3's public test version was the game that got every vendor scared straight into 1.1 compliance in spring 1999, MiniGLs died right then.

Right, 1.0 has some big issues. The lack of texture objects is a problem and the only solution in 1.0 (without using extensions) is plugging it into a display list and hope the driver does something intelligent with it. Another issue is the lack of any option for batching dynamic geometry. It works.... the performance just sucks. Testing on my Voodoo 3 the performance of a 1.0 compliant renderer vs 1.1 is about half the framerate. I was hoping there was something out there I could test 1.0 on mostly out of morbid curiosity. I'll probably just leave it around as a sort of compatibility layer just in case because it seems like there isn't really anything floating around out there.

MiniGL- nah don't think it's worth it for me. None of those 3dfx .dlls seem to support any of the vertex4f functions (providing geometry in clip space coordinates like you can in glide). Plus I already have a mostly finished glide renderer anyway- not much point in messing with the MiniGLs. Haven't looked into the minor vendors MiniGLs.

What about the OpenGL software renderer included with WinNT 3.51?

bakemono wrote on 2024-02-11, 13:10:

What about the OpenGL software renderer included with WinNT 3.51?

It's a good idea - but my engine needs at least Windows 95 to run. I wonder if the original release of Windows 95 included the 1.0 software renderer ? This would've pre-dated the 1.1 spec. I'm going to try that out later tonight.

lizard78 wrote on 2024-02-11, 14:16:

It's a good idea - but my engine needs at least Windows 95 to run. I wonder if the original release of Windows 95 included the 1.0 software renderer ? This would've pre-dated the 1.1 spec. I'm going to try that out later tonight.

Windows 95 didn't have one until MS released the OPENGL95.EXE addon, but I think it is the 1.1 API, similar to the software renderer in NT 4.

bakemono wrote on 2024-02-11, 14:49:

lizard78 wrote on 2024-02-11, 14:16:

It's a good idea - but my engine needs at least Windows 95 to run. I wonder if the original release of Windows 95 included the 1.0 software renderer ? This would've pre-dated the 1.1 spec. I'm going to try that out later tonight.

Windows 95 didn't have one until MS released the OPENGL95.EXE addon, but I think it is the 1.1 API, similar to the software renderer in NT 4.

You're right, I checked the original release and it doesn't have an implementation although OSR2 does (the 1.1 implementation).

lizard78 wrote on 2024-02-11, 15:22:

bakemono wrote on 2024-02-11, 14:49:

Windows 95 didn't have one until MS released the OPENGL95.EXE addon, but I think it is the 1.1 API, similar to the software renderer in NT 4.

You're right, I checked the original release and it doesn't have an implementation although OSR2 does (the 1.1 implementation).

It's actually much more interesting than you both believed!
While researching the old 3dfx OpenGL ICDs for Voodoo in the context of this topic, I suddenly came across the following paragraph in RelNotes.PDF for the Alpha2 version (section "System Requirements"):

Microsoft OpenGL must be installed to supply needed library files. This is the default on Windows NT, but only on some Windows 95 systems. To verify that OpenGL is installed on a Windows 95 system, check for the existence of the file named \WINDOWS\SYSTEM\GLU32.DLL. OpenGL for Windows 95 can be downloaded from the Microsoft web site from ftp://ftp.microsoft.com/softlib/mslfiles/oglfix.exe. For further information read the Knowledge Base article Q124556.

Before I was also sure that there was only one edition of OpenGL for the retail release of Windows 95 - the aforementioned OPENGL95.EXE redistributable with version 1.1, backported from NT 4. So what confused me here was another file name - OGLFIX.EXE, which I had not heard of earlier. But first I decided to look at that KB124556 article. And became intrigued even more, because it turned out to be noticeably different from the KB154877 article, which dedicated specifically to OPENGL95.EXE and is more well-known.

I was able to find the authentic OGLFIX.EXE file on Microsoft Technet Archive June 1997, which also occasionally contains OPENGL95.EXE for some reason. Then I unpacked it - and immediately revealed that it dates back to 1995, which means that the OPENGL95.EXE we already know was released six months later! And yep, it was OpenGL 1.0, which implies porting from NT 3.51, not NT 4! What's even more fun is that it also contains those famous OpenGL screensavers!

But what happened next surprised me even more. In the included ReadMe I saw this:

This release must be used in conjuction with the original OpenGL(R) for Windows(R) 95 release shipped in the MSDN Level 2 October CD.

???!!!
I immediately started to dig. So eventually I found that it fits the description of MSDN U.S. Setup (October 1995) - luckily, also available on the Internet Archive.
And yes, there is indeed a WIN95/OPENGL95 folder present with the very first release of OpenGL 1.0 for Windows 95, which predates OGLFIX.EXE by about three months. BTW, it contains those screensavers as well, unlike OPENGL95.EXE, where they were removed for some reason.

Hope this helps. That's all, folks!

Matrox Millenium II and Mystique 2 IIRC had OGL 1.0 on Windows NT 4 only. You need to limit the screen resolution and bit depth to 800x600 @ 16-bit colors to allow HW OGL to kick in. Tested on 3dsMax and the OpenGL screen saver.
As others mentioned, this is a professional setup. Those cards don't offer the OGL driver on Win95.

BlackDoomer wrote on 2025-01-23, 00:18:

It's actually much more interesting than you both believed! While researching the old 3dfx OpenGL ICDs for Voodoo in the context […]
Show full quote

lizard78 wrote on 2024-02-11, 15:22:

bakemono wrote on 2024-02-11, 14:49:

Windows 95 didn't have one until MS released the OPENGL95.EXE addon, but I think it is the 1.1 API, similar to the software renderer in NT 4.

You're right, I checked the original release and it doesn't have an implementation although OSR2 does (the 1.1 implementation).

It's actually much more interesting than you both believed!
While researching the old 3dfx OpenGL ICDs for Voodoo in the context of this topic, I suddenly came across the following paragraph in RelNotes.PDF for the Alpha2 version (section "System Requirements"):

Microsoft OpenGL must be installed to supply needed library files. This is the default on Windows NT, but only on some Windows 95 systems. To verify that OpenGL is installed on a Windows 95 system, check for the existence of the file named \WINDOWS\SYSTEM\GLU32.DLL. OpenGL for Windows 95 can be downloaded from the Microsoft web site from ftp://ftp.microsoft.com/softlib/mslfiles/oglfix.exe. For further information read the Knowledge Base article Q124556.

Before that, I was also sure that there was only one edition of OpenGL for the retail release of Windows 95 - the aforementioned OPENGL95.EXE redistributable with version 1.1, backported from NT 4. So what confused me here was another file name - OGLFIX.EXE, which I had not heard of earlier. But first I decided to look at this KB124556 article. And this intrigued me even more, because it turned out to be noticeably different from the KB154877 article, which dedicated specifically to OPENGL95.EXE and is more well-known.

I was able to find the authentic OGLFIX.EXE file on Microsoft Technet Archive June 1997, which also occasionally contains OPENGL95.EXE for some reason. Then I unpacked it - and immediately revealed that it dates back to 1995, which means that the OPENGL95.EXE we already know was released six months later! And yep, it was OpenGL 1.0, which implies porting from NT 3.51, not NT 4!

But what happened next surprised me even more. In the included ReadMe I saw this:

This release must be used in conjuction with the original OpenGL(R) for Windows(R) 95 release shipped in the MSDN Level 2 October CD.

???!!!
I immediately started to dig. So eventually I found that it fits the description of MSDN U.S. Setup (October 1995) - luckily, also available on the Internet Archive.
And yes, there is indeed a WIN95/OPENGL95 folder there with the very first release of OpenGL 1.0 for Windows 95, which predates OGLFIX.EXE by about three months. But what's even more fun is that it also contains those famous OpenGL screensavers!

Hope this helps. That's all, folks!

Wow, thanks for the update! I'm actually still working on this project, and it's come a long way since I made this post a year ago (even added Glide support in the meantime). This is awesome, really want to test out the OpenGL 1.0 renderer if nothing else besides just curiosity.

Maybe move this thread to Video? I believe it fits better there than in Milliways.

lizard78 wrote on 2024-02-10, 23:01:

I'm wondering if there are any video cards that had a complete OpenGL 1.0 ICD? I'm not talking about MiniGL or OpenGL 1.1 here.

The SiS 6326 OpenGL ICD is the only one that I'm aware of for Win9x that reports itself as OpenGL 1.0 compliant - more specifically, you need to use the AOpen driver combined with the Java ICD, both of which can be found at https://vintage3d.org/driver.php (6326 1.28 OpenGL and 6326 OpenGL Beta respectively) as the AOpen driver reports itself as OpenGL 1270.

https://en.wikipedia.org/wiki/RIVA_128
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#Pre-GeForce
These pages claim that RIVA 128 and RIVA 128ZX both had only OpenGL 1.0 support. How much is it true, does anyone know?

DEAT wrote on 2025-01-31, 05:28:

lizard78 wrote on 2024-02-10, 23:01:

I'm wondering if there are any video cards that had a complete OpenGL 1.0 ICD? I'm not talking about MiniGL or OpenGL 1.1 here.

The SiS 6326 OpenGL ICD is the only one that I'm aware of for Win9x that reports itself as OpenGL 1.0 compliant - more specifically, you need to use the AOpen driver combined with the Java ICD, both of which can be found at https://vintage3d.org/driver.php (6326 1.28 OpenGL and 6326 OpenGL Beta respectively) as the AOpen driver reports itself as OpenGL 1270.

Thanks! I just grabbed one of these off eBay. When it arrives and I test it out I'll post an update here.

BlackDoomer wrote on 2025-01-31, 06:30:

https://en.wikipedia.org/wiki/RIVA_128
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#Pre-GeForce
These pages claim that RIVA 128 and RIVA 128ZX both had only OpenGL 1.0 support. How much is it true, does anyone know?

I don't think that's accurate, these cards ran GlQuake, Quake II and other OpenGL 1.1 or newer titles.

DrAnthony wrote on 2025-01-31, 15:03:

BlackDoomer wrote on 2025-01-31, 06:30:

https://en.wikipedia.org/wiki/RIVA_128
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#Pre-GeForce
These pages claim that RIVA 128 and RIVA 128ZX both had only OpenGL 1.0 support. How much is it true, does anyone know?

I don't think that's accurate, these cards ran GlQuake, Quake II and other OpenGL 1.1 or newer titles.

When I originally made this thread I had tested all the drivers I could find for my RIVA and they all reported 1.1.

BlackDoomer wrote on 2025-01-31, 06:30:

https://en.wikipedia.org/wiki/RIVA_128
https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#Pre-GeForce
These pages claim that RIVA 128 and RIVA 128ZX both had only OpenGL 1.0 support. How much is it true, does anyone know?

Blatantly false, just like half of the things on Wikipedia.

See attached ZIP for reports I've done with 27 different OpenGL-capable chipsets with AIDA64 v4.60.3100 - RV100 and R100 do have support for different OpenGL extensions.

DEAT wrote on 2025-02-02, 00:08:

See attached ZIP for reports I've done with 27 different OpenGL-capable chipsets with AIDA64 v4.60.3100 - RV100 and R100 do have support for different OpenGL extensions.

Very interesting collection, thanks.
Wouldn't you mind to help me, please, with another question regarding old OpenGL ICDs?
Any examples of OpenGL ICDs that had DrvCreateDirectDrawContext function?
I just need someone who has a lot of drivers for ancient video cards on hand.