Sure, there are plenty of BIOSes that will work in the AT. Quadtel, AMIBIOS, Phoenix, MR, and I think even the DTK BIOS would work if you dared to use it.
I think I've used pretty much all of them, but MR BIOS is by far the best option. It's nice to stay original, but the IBM BIOS seriously sucks.

Anonymous Coward wrote on 2020-12-02, 01:08:

That 287Turbo module is pretty insane. I have a 287 upgrade module of some sort which allows the FPU to run async to the CPU, but it's very compact and simple...nothing at all like this monstrosity. Why on earth does it require so much power? Were they afraid that the original AT motherboard couldn't handle a 10MHz NMOS 287? Where does that two lead cable go?

There is an extensive review of this product in pcmag
https://books.google.com/books?id=vcqI6sY2e-k … 87Turbo&f=false

From what I can tell the 287turbo (even in 1986) clocked up to 12mhz with 16 & 20mhz flavors not far off.

It is possible they felt there wasn’t large enough power traces to the socket.

Anonymous Coward wrote on 2020-12-02, 07:44:

You should upgrade your AT to MR-BIOS. It's a lot better than either the IBM BIOS or the AMI replacement BIOS...though either will take care of your memory count problem.

That's one thing I really don't want to change on this machine, unless I have to for some reason. I fully acknowledge the functional superiority of the other BIOSes mentioned but there's something endearing about the "clunkiness" of the original. It's part of the machine's personality - and it's a fun thing to show people too, the process of booting the IBM Diagnostics floppy to run tests or edit settings. The one thing I'd like to change is the limit on overclocking. I read somewhere that 8MHz AT's have an overclocking blockade, if you will, built into the IBM BIOS - and it would be a fun experiment to see if I could swap that 16MHz crystal out for something a little faster, especially now that I have the 386 CPU installed. It's a 20MHz 386 running at 16MHz, so I'm willing to bet it (the CPU, that is) could manage 24MHz. As for the rest of the system, the memory, well that's why it's an experiment. Perhaps even 20Mhz. I wonder if anyone ever hacked the original IBM BIOS to remove that overclocking limitation but keep everything else intact.

Jo22, there are Quadtel BIOS chips for sale on eBay that are listed as upgrades for the 5170. I could see trying something like that, in order to test overclocking potential, but I don't think I'd want to leave it in there permanently - for reasons discussed.

Update: the drivers called "CL22V143.ZIP" work great too, and include a SetRES utility for easily changing between different resolutions and color depths - and it's a full English-language driver too.
http://ftp.mpoli.fi/pub/hardware/DISPLAY/CIRRUS/

CL22V143.ZIP is described as "GD5422/GD5424", ie. non-accelerated... ugly!

I think I've read the 287Turbo article before. It's been a while though, I guess I'll have to read it again. I suspect that there were probably multiple revisions of this product. I am pretty sure a CMOS 287 uses less power than an 8MHz NMOS version. I can't even recall if a 12mHz NMOS version even exists. If it does, it's pretty rare.

The overclocking limiter is called a "governor". I believe there is a hacked version of the BIOS used by IBM employees. I think I remember reading about it on vcfed. Maybe minuszerodegrees has a copy.
Doesn't the original 286 CPU give the machine a lot of its character also? I'd rather keep the original CPU with an updated BiOS. IF you wanted to, you could even build a custom circuit board to select which ROM BIOS to boot from.

Looking at my bare NOS 6mhz 5170 AT motherboard replacement I appear to have 150ns memory

But upon testing holy waitstates Batman!

I should be able to run o waitstates at 6mhz using 150ns memory? It looks like I have clk=2 waitstates?

Is that correct?

This thread makes me sad I have the motherboard but not the rest of the system

Yes, it's a well-know fact that 5170 comes with unnecessary wait states.
Some people claim that IBM introduced the wait states to prevent the 5170 from competing with their more expensive big iron.

5170 boards all have 1 wait state for onboard RAM, but a RAM expansion card can be 0ws if it is designed to do so. I've always been curious if it would be possible to do a hardware mod to a 5170 board to eliminate the waitstate. The schematics are available in the technical manual, but I couldn't figure it out. I ended up buying a 5162 motherboard, which is is basically a AT board with 0WS.

Jo22 wrote on 2020-12-01, 07:55:

Well, the only accelerator series that I know of which could *work* is the WDC90c-something series. It's backwards compatible to […]
Show full quote

Grzyb wrote on 2020-11-29, 16:09:

Second, there was some talk about using accelerated cards in Windows on a 286, and I think the only driver known to work is that for IBM 8514/A.
So, forget about CL-GD5426, and either use some early non-accelerated SVGA, or some 8514/A clone like ATI Mach8/Mach32.

Well, the only accelerator series that I know of which could *work* is the WDC90c-something series.
It's backwards compatible to the old PVGA1A/1B series. At the core, at least.
I've successfully run some of the old Windows 2.03 drivers in conjunction with a WD90C00 and WDC90C21/31 chip (on Win 2.03 and 3.0 in Real-Mode).
This gave me 640x400 in 256c, which was not too shabby by late 1980s standards.
Unaccelerated, of course.
800x600 in 16c was another option, but that's not any better than the default Super VGA driver of Windows 3.1.
If it only could work on a 286..

Edit: Another driver to try to is the paletizzed VGA driver from Windows 3.0 MME.
It's in 16c only, but can do palette cycling.
Ifaik, it does work in Standard Mode, too.
Re: VGA games with only 16 colors

I just tried to get my WD90C31A-LR to work in Windows 3.1 on my 286. Standard VGA driver works obviously. When trying the OEM WD90C31driver the desktop wallpaper loads, then as soon as the first GUI element is drawn it segfaults. Trying the Speedstar 24X drivers was no better, both Diamond's 3.0 and 3.1 driver print a sign on message than drop back to the DOS prompt.

It would appear that there are NO GUI accelerators with drivers that work in Win3.1 Standard mode. A shame really

The title of this thread needs changing. When I saw it I thought wtf?

maxtherabbit wrote on 2020-12-24, 18:59:

It would appear that there are NO GUI accelerators with drivers that work in Win3.1 Standard mode. A shame really

You obviously haven't read this thread. A shame really.
Let me repeat:
There are GUI accelerators that work in Windows 3.1 Standard mode, even on a 286: IBM 8514/A and compatibles, eg. ATI mach8 and mach32.
A quote from ATI drivers package dated May 7, 1993: "To use the mach8 on a 286 based processor, or in Windows Standard Mode, use the supplied Microsoft 8514/A driver."

Grzyb wrote on 2020-12-25, 00:21:

You obviously haven't read this thread. A shame really. Let me repeat: There are GUI accelerators that work in Windows 3.1 Stand […]
Show full quote

maxtherabbit wrote on 2020-12-24, 18:59:

It would appear that there are NO GUI accelerators with drivers that work in Win3.1 Standard mode. A shame really

You obviously haven't read this thread. A shame really.
Let me repeat:
There are GUI accelerators that work in Windows 3.1 Standard mode, even on a 286: IBM 8514/A and compatibles, eg. ATI mach8 and mach32.
A quote from ATI drivers package dated May 7, 1993: "To use the mach8 on a 286 based processor, or in Windows Standard Mode, use the supplied Microsoft 8514/A driver."

I did read that post, but glossed over it because I didn't realise until now that the 8514 had a hardware blitter.

I assumed that by using the built in driver you would only be getting unaccelerated VGA

maxtherabbit wrote on 2020-12-25, 04:55:

Grzyb wrote on 2020-12-25, 00:21:

You obviously haven't read this thread. A shame really. Let me repeat: There are GUI accelerators that work in Windows 3.1 Stand […]
Show full quote

maxtherabbit wrote on 2020-12-24, 18:59:

It would appear that there are NO GUI accelerators with drivers that work in Win3.1 Standard mode. A shame really

You obviously haven't read this thread. A shame really.
Let me repeat:
There are GUI accelerators that work in Windows 3.1 Standard mode, even on a 286: IBM 8514/A and compatibles, eg. ATI mach8 and mach32.
A quote from ATI drivers package dated May 7, 1993: "To use the mach8 on a 286 based processor, or in Windows Standard Mode, use the supplied Microsoft 8514/A driver."

I did read that post, but glossed over it because I didn't realise until now that the 8514 had a hardware blitter.

I assumed that by using the built in driver you would only be getting unaccelerated VGA

Your comprehension of Grzyb posts is lacking...

Interesting. So even with the 1993 driver set, ATi recommended the 8514/A driver for standard mode?
Next month I'll see if I can find my 1992 user manual to see what it says. My driver disks should be from mid-late 1992.
It would be nice to be able to use 800x600 in standard mode with accellerated graphics.

BTW, the drivers shipped with Windows 3.1 also include:

TIGA (coprocessor)
XGA (accelerator)

There's a chance they also work on a 286.
Is there anybody with appropriate hardware, and willing to try?

Grzyb wrote on 2020-12-25, 10:56:

BTW, the drivers shipped with Windows 3.1 also include: […]
Show full quote

BTW, the drivers shipped with Windows 3.1 also include:

TIGA (coprocessor)
XGA (accelerator)

There's a chance they also work on a 286.
Is there anybody with appropriate hardware, and willing to try?

I'm willing to try, but I can't right now.
My room is so convoluted that I can barely walk in it.
Also, the 286es that I've got have not enough room to house that full size TIGA card.
- They are pizza box size and my Schneider Tower has no free slot.

However, I once tested that TIGA card in a 386DX40 PC (now sold).
From what I remember, that TIGA driver of Windows 3.1 is some sort of intermediate driver.
It talks to the DOS-based TIGA driver that comes with the TIGA card.
So it may work with any GFX card that provides a high level compatibility with the TIGA standard.

The 8514/A driver may or may not work that way, not sure.
But along with XGA, it's surely worth to test.

It's sad that Microsoft neglected the 286 so much, but it is understandable.
If we look back to Windows/386, we can be glad that Standard exists at all.
If memory serves, the 386 kernal evolved first from the original Windows 1.x/2.x Real-Mode kernal:
The original release of Windows 2.x was Windows/386 (OEM releases that shipped with Compaqs)
Vanilla Windows 2.03 arrived after this at the shelves.
So unlike IBM, which cared for the 286 and implemented virtual memory into OS/2 1.3, MS focused on the 386 since the 1980s.
If the 286 PCs weren't so common in the late 80s/early 90s, Standard Mode might have never made it into Windows 3.0.

I know, this was not fair to 286 users at the time, but it really seems to me that MS considered Standard Mode as a "just works" mode of Windows and Enhanced Mode as the "all inclusive" mode with all the bells and whistles.
Luckily, application wise, Standard Mode kernal is compatible with most Windows features, like MCI (MIDI/PCM), TWAIN (scanners), True Type fonts and so on.
Unlike Real-Mode kernal of Windows 3.0, it was usable by developers.
All major IDEs like Visual Basic, Delphi, TPW, dBase Fast, FoxPro etc. execute on 286 PCs running Standard Mode.
If Standard Mode just had virtual memory implemented..
On a 286 PC, someone was required to buy expensive SIMMs (by the time) and physically upgrade the amount of memory.
For a serious developer, this meant installing 4, 8 or 16MB of RAM (or in between).
Without such an upgrade, the compiler simply couldn't complete a larger project.
Like wise, an artist or architect was unable to work on a large project.
Usually, the applications simply would run out of memory.
Unless they implemented their own swap or overlay technology.

Edit : Small edit.

Caluser2000 wrote on 2020-12-25, 05:26:

Your comprehension of Grzyb posts is lacking...

A shame really.

Jo22 wrote on 2020-12-25, 11:45:

I'm willing to try, but I can't right now. My room is so convoluted that I can barely walk in it. Also, the 286es that I've got […]
Show full quote

Grzyb wrote on 2020-12-25, 10:56:

BTW, the drivers shipped with Windows 3.1 also include: […]
Show full quote

BTW, the drivers shipped with Windows 3.1 also include:

TIGA (coprocessor)
XGA (accelerator)

There's a chance they also work on a 286.
Is there anybody with appropriate hardware, and willing to try?

I'm willing to try, but I can't right now.
My room is so convoluted that I can barely walk in it.
Also, the 286es that I've got have not enough room to house that full size TIGA card.
- They are pizza box size and my Schneider Tower has no free slot.

However, I once tested that TIGA card in a 386DX40 PC (now sold).
From what I remember, that TIGA driver of Windows 3.1 is some sort of intermediate driver.
It talks to the DOS-based TIGA driver that comes with the TIGA card.
So it may work with any GFX card that provides a high level compatibility with the TIGA standard.

The 8514/A driver may or may not work that way, not sure.
But along with XGA, it's surely worth to test.

It's sad that Microsoft neglected the 286 so much, but it is understandable.
If we look back to Windows/386, we can be glad that Standard exists at all.
If memory serves, the 386 kernal evolved first from the original Windows 1.x/2.x Real-Mode kernal:
The original release of Windows 2.x was Windows/386 (OEM releases that shipped with Compaqs)
Vanilla Windows 2.03 arrived after this at the shelves.
So unlike IBM, which cared for the 286 and implemented virtual memory into OS/2 1.3, MS focused on the 386 since the 1980s.
If the 286 PCs weren't so common in the late 80s/early 90s, Standard Mode might have never made it into Windows 3.0.

I know, this was not fair to 286 users at the time, but it really seems to me that MS considered Standard Mode as a "just works" mode of Windows and Enhanced Mode as the "all inclusive" mode with all the bells and whistles.
Luckily, application wise, Standard Mode kernal is compatible with most Windows features, like MCI (MIDI/PCM), TWAIN (scanners), True Type fonts and so on.
Unlike Real-Mode kernal of Windows 3.0, it was usable by developers.
All major IDEs like Visual Basic, Delphi, TPW, dBase Fast, FoxPro etc. execute on 286 PCs running Standard Mode.
If Standard Mode just had virtual memory implemented..
On a 286 PC, someone was required to buy expensive SIMMs (by the time) and physically upgrade the amount of memory.
For a serious developer, this meant installing 4, 8 or 16MB of RAM (or in between).
Without such an upgrade, the compiler simply couldn't complete a larger project.
Like wise, an artist or architect was unable to work on a large project.
Usually, the applications simply would run out of memory.
Unless they implemented their own swap or overlay technology.

Edit : Small edit.

I think MS just saw the future in 386 on wards 32 bit x86 processors as the future and were heavily involved in the creation of OS/2 1.3s GUI. To MSs credits their OSs were very backward compatible. IIRC OS/2 1.x software was supported up to NT4. Also there was an easy way of accessing HPFS hdds in W2k done by a German chap. Great technical folk those Germans. And of XP could support NETBui as well from a driver on the installation CDRom. A open network protocol that was/is small and well suited to older kit if you didn't need internet access. A lot of developers switched over to 386DX(the DX was added when the 386SXs were introduced. 386DX systems were very expensive systems to purchase but quite a few professionals bought them like the surveyor who had the 1989 MITEK 386DX25 system I have. It originally it had a 100meg scsi hdd which was dead when he gave it to me.

Also IIRC MS Windows/386 v2.x did actually run on a 286 system with out any modifications. I suspect a 386 system ran it as what was known as Standard Mode in MS Window 3.x.

Caluser2000 wrote on 2020-12-25, 19:02:

Also IIRC MS Windows/386 v2.x did actually run on a 286 system with out any modifications. I suspect a 386 system ran it as what was known as Standard Mode in MS Window 3.x.

Yes, you're absolutely right.
Windows/386 essentially is a bundle of normal Windows 2.x and an additional memory manager.
Simply put, it loads 8086 / real-mode Windows in a dedicated VM and DOS applications are handled separately.
So Windows/386 does little for Windows applications, but a lot for DOS programs.
Windows/386 can emulate EMS, CGA graphics in a DOS window, Copy&Paste between DOS/Win applications etc.

- And that's what Microsoft had focused on at the time.
That's what I meant to express when I said 386 "mode" came first.
That memory manager kind of "fused" with the Windows 2.x kernal and became the 386 Enhanced Mode in Windows 3.0 as we knew it.
How exactly Standard Mode materalized.. I don't know. 😅
Maybe some hints can be found on YT or Beta Archive, not sure.
Interestingly, the Real-Mode kernal of Windows 3.0 is also somewhat refurbished.
It can run many, many applications if EMS is available. Even though it was marketed as a compatibility mode for old applications originally.
(It can use Windows 2.x drivers and it is compatible with the old colour palette.)

Anyway, back to Windows/386..
By running Win86 executable, the Windows 2.x environment loads on any x86 machines.

https://www.youtube.com/watch?v=OcjvgxAKiHs

https://www.youtube.com/watch?v=0lvQs1KM_cc