Does AMD/ATI, nVidia, or Intel have any sort of stable 2D graphics API/register set that their cards actually supported for a period of time? I just find it odd, especially after 3D took off, that these vendors kept changing the programming interface for basic graphics display.

That stable API is the OS driver interface (or things like Glide and ATI CIF), which unifies all the hardware for the software side of things. But all the stuff from driver to hardware is purely hardware specific and depends very much on how the particular hardware was implemented and there was a lot of ideas and evolution on that regard over the years.

How's Autodesk support on these late graphics cards?
The ADI driver existed in Real-Mode and 32-Bit Protected-Mode version, I think.
So if there's still ADI support for, say, AutoCAD R13 then an Windows driver using ADI driver might be possible.
It's same basic idea as using the TIGA driver in Windows 3.1, which relies on the TIGA graphic's cards DOS-based TIGA driver.

I wonder how difficult it would be to build a version compatible with Windows 3.0. I have yet to find a high resolution/color driver that works well with Win 3.0. My ET4000 card has the best drivers that I've tested, but the drivers for that card have a timing bug on fast computers. I can only use it if I disable CPU caches (on my Pentium 4, specifically).

fosterwj03 wrote on 2025-05-03, 18:24:

I wonder how difficult it would be to build a version compatible with Windows 3.0. I have yet to find a high resolution/color driver that works well with Win 3.0. My ET4000 card has the best drivers that I've tested, but the drivers for that card have a timing bug on fast computers. I can only use it if I disable CPU caches (on my Pentium 4, specifically).

At least Windows 3.0 can be run on fast computers at all, though doubtless it only sees a paltry amount of RAM.

John Elliott (https://www.seasip.info/VintagePC/index.html) wrote SVGA drivers for Windows 1 & 2, ostensibly by "universalizing" preexisting card-specific drivers; since the Windows 2 architecture so closely matches that of 1 software on either takes little adaptation to be made cross-compatible, as Nate Lineback's experimentation from years ago proved. Then in 3.1 we get the generic SVGA driver which accomplishes the same thing as standard. That just leaves 3.0, which theoretically should not take much with a similar method.

Jo22 indicated a while back there is a SVGA driver in a 3.1 beta that works in 3.0, but is it as consistent as the final version for 3.1? The screenshot shows real mode, probably for verification purposes, but that may be how I prefer to run 3.0 (for backwards compatibility) anyway. If it is good enough could the driver be posted for download?

¡¡This is just FANTASTIC!!

Windows 3.0o(A)(MPC) were so picky about memory ovel 16MB just locks him up and that WINA20.386 DOS driver in root. Windows 3.1x is so memory and mostly speed friendly minus few network stuff.

I don't have any memory issues with Windows MME on a system with 4GB of RAM (I think DOS 6.22 HIMEM.SYS only recognizes 64 MB). PCI graphics cards, on the other hand, have proven difficult to use without issues.

Just noticed this thread.
See sig for 3.11 iso template that includes this integrated.

fosterwj03 wrote on 2025-05-31, 02:05:

I don't have any memory issues with Windows MME on a system with 4GB of RAM (I think DOS 6.22 HIMEM.SYS only recognizes 64 MB).

Which mode, out of curiosity? I know one big advantage 3.0 offered over its predecessors was independent memory management ranges contingent on which mode Windows started in, bypassing the issue with advanced editions of Windows 2.x not being able to function concurrent with DOS memory managers. In turn, though, compatibility with older Windows programs broke on other than Real Mode (Word 1.1 & Excel 2.1 were updated to address this).

BaronSFel001 wrote on 2025-06-02, 15:15:

fosterwj03 wrote on 2025-05-31, 02:05:

I don't have any memory issues with Windows MME on a system with 4GB of RAM (I think DOS 6.22 HIMEM.SYS only recognizes 64 MB).

Which mode, out of curiosity? I know one big advantage 3.0 offered over its predecessors was independent memory management ranges contingent on which mode Windows started in, bypassing the issue with advanced editions of Windows 2.x not being able to function concurrent with DOS memory managers. In turn, though, compatibility with older Windows programs broke on other than Real Mode (Word 1.1 & Excel 2.1 were updated to address this).

"386 Enhanced", although I suspect very few software titles that work with Windows 3.0 could actually benefit from it.

fosterwj03 wrote on 2025-06-03, 01:28:

"386 Enhanced", although I suspect very few software titles that work with Windows 3.0 could actually benefit from it.

Hi, I second that. The majority probably was fine with Standard-Mode.
386 Enhanced-Mode was useful to users with low RAM but big applications, because it allowed using swap file.

The original MPC 1 specification mentioned a 286 processor, also.
With the revised MPC 1 specification it became a 386SX.

BaronSFel001 wrote on 2025-05-05, 14:45:

Jo22 indicated a while back there is a SVGA driver in a 3.1 beta that works in 3.0, but is it as consistent as the final version for 3.1?
The screenshot shows real mode, probably for verification purposes, but that may be how I prefer to run 3.0 (for backwards compatibility) anyway.
If it is good enough could the driver be posted for download?

Hi there. I don't know for sure. It's just an earlier supervga.drv, nothing too spectacular.
Adding display.drv=supervga.drv in system.ini and adding
[display]
svgamode=106
in win.ini was all I had to do. I've added the file, I hope that's okay. It's a mere driver.
It also works on my Commodore PC10, with V20 processor and my bought copy of Windows 3.0 RTM (GER).

Edit: Photo attached. Works on real hardware. PC10 and PVGA1C graphics card.
VESA VBE drivers loaded in DOS, but that's probably not so important.

Jo22 wrote on 2025-06-03, 11:10:

VESA VBE drivers loaded in DOS, but that's probably not so important.

Thanks, and yes: it likely matters not what is loaded on the DOS side because drivers needed to be Windows-specific for that support. When I finally have the chance to get my retro PC going again I plan to experiment with my PCI cards which support Windows 3.1 but were definitely not made with any earlier version in mind: a Verite (VESA 2.0) & Voodoo3 (VESA 3.0). Since Windows 3.0 offers a different desktop palette in high-resolution 256 colors that option makes for a nice-to-have.

Thanks, and yes: it likely matters not what is loaded on the DOS side because drivers needed to be Windows-specific for that support.

+1

I just mentioned it, because such SuperVGA drivers have VBE as another fallback option.
They try to use direct chip set support first, but also can use VBE.

Interestingly, both types of 800x600 16c mode. Mode 6Ah and 102h.
The difference is in timings, sometimes, which is clever.
Some SVGA cards to use slightly different timings depending if its 6Ah (old VGA compatible mode number) and official VBE mode 102h.

Also, Windows 3.0 in Real-Mode is acting like a DOS program to the OS. Okay, more or less.
So DOS and BIOS routines might be accessible, as well as DESQView APIs or QEMM/EMM386 APIs.

That's why I wondered if it's possible to write a mini webbrowser (akin to MicroWeb 2) for Windows 1/2/3 or 3.1B that uses a basic packet driver (ne2000.com) or something like mTCP.
Would be fun! Such hacky and experimental things are the positive sides of otherwise limited Real-Mode.

Jo22 wrote on 2025-06-03, 18:17:

Also, Windows 3.0 in Real-Mode is acting like a DOS program to the OS. Okay, more or less. So DOS and BIOS routines might be acc […]
Show full quote

Also, Windows 3.0 in Real-Mode is acting like a DOS program to the OS. Okay, more or less.
So DOS and BIOS routines might be accessible, as well as DESQView APIs or QEMM/EMM386 APIs.
...
Would be fun! Such hacky and experimental things are the positive sides of otherwise limited Real-Mode.

I never knew/realized that; it makes perfect sense when I think about it! Yes, experimentation is very much the fun, and I mentioned before my key interest in 3.0's Real Mode is its backwards compatibility with older Windows applications. Though I prefer the gaming over the general software side of history, I consider the latter symbiotic to the former and account for it as part of a "whole" retro experience...but have nothing against those for whom Windows 98SE suffices since, for the vast majority of notable games (both DOS & Windows), it does.