Well, Windows 2.x has a few games and utilities (easel, metric converter, terminal, ..) and some commercial packages, like Word and Excel or Page Maker.
The /386 version was the most useful back in the day. It was like EMM386 on steroids and could even run CGA games in a window.
Oddly, it also came with a win86.com executable. The 286 edition added support of using 64KiB of High Memory (HMA).
Despite this, it was similar to the base version. According to current information, the /386 version (v2.01?)
was the first one to be available (OEM, shipped with 386es from Compaq). But the first retail (boxed) version was Windows 2.03.
Anyway, the main issue is the support of EMS (-> large type). It can be only used by DOS programs normally.
So even if you use MemMaker or an EMS board, Windows application can't use it (except if they were made EMS-aware, just like DOS apps).

Windows 3.0 in real-mode is different. It supports EMS system-wide across all applications.
Both Windows 2.x and 3.x programs can use it, even in RM. You can also use large wallpapers w/ EMS enabled.
Windows 3.0 has about one thousand times the applications with lots of them beeing Freeware.
Maybe that's also because of Visual Basic 1.0, which runs in both Standard and Enhanced mode.

Zup wrote:

Are there any reason (=good software) that can be installed on that versions of Windows? Or the best thing to do is forget about Windows?

I suppose that depends on what sort of "good software" you were otherwise going to run on this computer.

I keep linking to ToastyTech, because it's the only place I've come across that's written at length about stuff like this.
http://toastytech.com/guis/win1x2x2.html
http://toastytech.com/guis/win1x2x3.html

So it seems that real fun started with Windows 3.x standard/enhanced mode.

I guess having Windows 2.x in that XT would be funny to show in parties, but not really a good choice.

I'd say so. Though a remarkable number of programs will run in any mode.
I'm speaking under correction, but Turbo Pascal for Windows 1.x also had a "286" checkbox,
so it might be possible that its compilations will by default run on XTs, too.

Edit: Just checked. TPW itself requires either Standard or Enhanced Mode.
But the Resource Toolkit and the compilations do run in RM, too!
So with default settings beeing choosen, the TPW programs are probably mode-independed. :)

Jo22 wrote:

Turbo Pascal for Windows 1.x also had a "286" checkbox,
so it might be possible that its compilations will by default run on XTs, too.

I guess so. That was the case with the DOS version of Turbo Pascal.

Azarien wrote:

Jo22 wrote:

Turbo Pascal for Windows 1.x also had a "286" checkbox,
so it might be possible that its compilations will by default run on XTs, too.

I guess so. That was the case with the DOS version of Turbo Pascal.

You're right, just checked.
Newer versions of Turbo Pascal for DOS (ie, past 4.0; maybe 6.x and later) seem to include it, as well.

Fixed your screenshot.

Zup wrote on 2017-05-15, 16:22:

So it seems that real fun started with Windows 3.x standard/enhanced mode.

I guess having Windows 2.x in that XT would be funny to show in parties, but not really a good choice.

My apologies for the necro post here, but just recently I was thinking of TPW/Win3 when doshea replied to me in this thread and then I've remembered this older thread of yours.

That's when I realized that I've created another one of myself last year, which was about old Windows versions appearing in films.

MS-Windows (1980s) in films and series

Maybe that's interesting to you, as well.
Having popular films or shows featuring our retro PCs might be an interesting tool to introduce newcomers to the hobby.
So it would truely "be funny to show in parties". 🥳

PS: A few days ago I've noticed Windows 3.1x appearing in Gun Smith Cats anime (second episode?).
I know, it's not exactly Windows 3.0, but it looks quite same to lay(wo)men.
It also uses the colour scheme of the S3 graphics drivers (light blue title bar).
Apparently, this colour scheme was popular in Japan at the time? 🤷‍♂️

Edit: Broken link fixed.

I vaguely recall a British 80186 system that could address -800kb of base ram that had some personalized stuff for various versions of windows, I don’t remember if they made a cludge so it could run Windows 3.1

Although the amount of software isn’t massive there are a number of Windows games and programs that work on 2.x or 3.0 real mode.

Word 1.0 for Windows being one

RM Nimbus? 🤷‍♂️

Edit: Picture..

Um, I'm not sure if this information is helpful, but I vaguely remember that some users at Beta Archive discussed the early days of Windows 3.1x.

If I understood correctly, early Beta versions of Windows 3.1x did still support Real-Mode.
So it *might* be possible to them to run some Windows 3.1 applications on an XT.

Provided that a) the Real-Mode kernal has the ability to use new Windows 3.1x API functions and b) that the compiler didn't make Protected-Mode binaries.

Because, if the compiler used was originally a Windows 3.0 compatible compiler,
like Turbo Pascal for Windows was, then there's a chance for Windows 3.1x applications to run unmodified on an XT.

But since no one ever had a Real-Mode capable Windows 3.1x, it's hard to check.

Best bet would be to make Windows 3.0 report back a Windows 3.1x version somehow.
That way, the application should run until an unsupported function call is made.

Anyway, it's just a mind experiment. If we had access to a Beta version of Windows, the even better.
Not sure if we can talk about this here, though.
I mean, in principle, it's harmless talk about computing history. Betas were never sold, rather the contrary.
Beta testers got their free copies back then..

Anyway, I'm just thinking out loud. In theory, there're might be hope to get some Windows 3.1 applications going on an XT.
The NEC V20/V30 chips do even have the 286 specific instructions (!), so there's really a chance.

It's merely a question whether or not the executable use features of Standard-Mode or 386 Enhanced-Mode kernal and if the Windows 3.1 Beta Real-Mode kernal has access to the extended Windows 3.1 API.

Jo22 wrote on 2023-06-26, 18:07:

RM Nimbus? 🤷‍♂️

Edit: Picture..

That must be it, I vaguely remember someone running a cludged version of windows 3.1 on the 186 version of the machine (2.x and 3.0 were personalized versions for that machine specifically with its odd more than 640k configurations)

Reading this the 3.1 beta real mode must have targeted 186+ machines

https://www.betaarchive.com/forum/viewtopic.php?t=43683

rmay635703 wrote on 2023-06-26, 20:06:

Reading this the 3.1 beta real mode must have targeted 186+ machines

Yes, that makes sense.
One of the differences between Windows 3.0 and 3.1 was performance.
While Windows 3.0 was more lightweight, it also was written mainly in plain C.

Windows 3.1, by contrast, was much more complex. The whole DDE/OLE system, WinMem32, accelerated GDI drivers, new Multimedia APIs, True Type fonts etc..

To make up for it, it got parts being rewritten in assembly.
At this point, using 286 instructions made sense, if Standard-Mode was needing it, anyway.

I suppose that the Real-Mode support simoly was a leftover from Windows 3.0.
In Windows 3.0, Real-Mode kernal rrally was primarily intended to support Windows 2 applications.

Microsoft didn't use this as a marketing stunt or to appeal XT owners secretly, I think.
Because, the Real-Mode kernal supports the older Windows 2.x graphics and printer drivers, for example.

Then, it can also handle the old resource formats properly (Windows 2.x icons in colour, correct colour palette in applications etc).

That's really fascinating, I think. Some Windows 3.0 applications, by contrast, may loose coloured icons then (in About dialog etc). But they'd still be run, ar least.

Windows 3.0 on an XT can run games like Lander 3 or Gnu Chess.
Back in 1990/1991, this must have been a nice side effect to XT owners.

While not officially supported, they could keep using their Turbo XTs in the new Windows world.

For students or other poor people, this must have had been a godsend, kind of.
They could at least participate on some of the current software of the day.

Merely the Windows 3.0 VGA driver was incompatible with 8086/88 by default m.
It used 80186/80286 instructions, probably to speed up drawing (I know that Windows 3.1 VGA driver uses dirty-rectangle,
maybe that's also true for 3.0 driver. If so, the new CPU instructions would have been really useful here).

Nowadays, there's a patch at Vcfed forums to make it run on 8088/8086 XTs.
Back in the day, however, users could simply drop in a NEC V20/V30, which was recommended anyway, due better performance.
It also provided enhanced compatibility with newer DOS utilities of the day (the 80286 was the baseline CPU by then).

All the memory operations were better handled, too, I think. The 8086/8088 had to use the ALU to form a memory address (segment/offset etc).
The 80186 and the NECs had a dedicated address decoding unit, if I understand correctly, causing a performance boost of up to 20% sometimes (estimated number).

Here's a video of Windows 3. 1 Beta running on an XT..
To me, the colourful Windows 3.1 Icons alone would have been appealing enough to favor this over Windows 3.0! 😁

https://www.youtube.com/watch?v=9PNyXFdrlkw

From what I remember you could use XT versions of drivers be they Windows 2.x or 3.0

I had VGA on one of the XT machines back in the day but all you got was VGA mono.

Another oddity was that real mode made expanded memory available to compatible applications, this could make a big difference on an XT if you set things up correctly and ran the correct applications/revs.

rmay635703 wrote on 2023-06-27, 01:47:

From what I remember you could use XT versions of drivers be they Windows 2.x or 3.0

I had VGA on one of the XT machines back in the day but all you got was VGA mono.

Another oddity was that real mode made expanded memory available to compatible applications, this could make a big difference on an XT if you set things up correctly and ran the correct applications/revs.

Hi, yes, that's right! ^^ The VGA mono driver used plain 8086 instructions, as far as I know.
EMS support is also neat. It was really useful to users who had an XT or AT with hardware-based EMS (and no XMS).
Because with Expanded-Memory, Windows 3.0 can do real work even on an XT (Excel or a dozen pages in WinWord).

The Windows 3.1 Beta seems to go even a step further and allows the Super VGA driver to work in Real-Mode.
From what I can tell, it can already run real Windows 3.1 applications like Minesweeper, too.
On the other hand, it lacks a lot of features from Windows 3.0 MME.

There seems to be no "driver" icon in Control Panel, for example.
So no sound output (wave). Unless, maybe, if that early Sound Blaster driver package is installed.
That one essentially ships with an early form of MCI/WinMM API, if I remember correctly.

oeuvre wrote on 2017-05-18, 14:32:

Fixed your screenshot. […]
Show full quote

Fixed your screenshot.

Took me about 10 seconds

Quick update. Found something about the removal of Real-Mode kernal in Windows 3.1!

------------------------------------------------------------------------------- The information in this article applies to: […]
Show full quote

-------------------------------------------------------------------------------
The information in this article applies to:

- Microsoft Windows Software Development Kit (SDK) 3.1
-------------------------------------------------------------------------------

SUMMARY
=======

Support for real mode has been removed from Windows 3.1. Many applications
designed for Windows 3.0 do not support real mode. The reasons behind this trend
toward protected mode include superior memory management and smaller, faster
application code.

MORE INFORMATION
================

The remainder of this article lists the advantages and disadvantages of removing
support for real mode from applications.

Advantages of Removing Real Mode Support
----------------------------------------

1. Protected mode code is smaller, cleaner, and more maintainable. These factors
lead to a faster, more responsive, more reliable, system. Code is smaller and
cleaner for the following reasons:

a. The 286 and higher processors can track memory locations in hardware,
which makes locking and unlocking memory objects unnecessary. An object
can be locked once when it is allocated and unlocked just prior to being
freed. Because the object can move in memory even when it is locked, it is
not necessary to bracket each access to an object with lock and unlock
calls.

b. Far functions can use simplified function prolog and epilog code. For more
information on this aspect of protected mode, query on the following
words:

" prod(winsdk) and protected and streamlined " (without the quotation marks)

c. Because protected mode code is restricted to running on 286 and higher
processors, the Microsoft C Compiler -G2 switch can be used to generate
smaller and faster application code.

2. Protected mode (both standard and enhanced mode) breaks the "640K barrier."
Furthermore, under enhanced mode, Windows uses paged virtual memory to expand
available memory by using the system hard disk as a swapping device. The
large address space allows applications to have more code and data and allows
users to run more applications.

3. Testing is easier because there are fewer Windows modes to test. To fully
test a product that runs under real mode, five separate modes must be tested.
Real mode itself contributes three of those modes: real mode with no expanded
memory, real mode using the small-frame Expanded Memory Specification (EMS)
and real mode using the large- frame EMS. The other two modes are standard
mode and enhanced mode. Because support for real mode has been eliminated,
the same amount of testing effort can concentrate on producing a better
product. It also provides an opportunity to develop and test additional
enhancements. For more information on EMS, query on the following words:

" prod(winsdk) and ems and developers " (without the quotation marks)

4. Based on a survey of Windows developers, most developers are targeting only
protected mode because Windows performance on 8086- based and 8088-based
machines is not satisfactory. Furthermore, these machines cannot address more
than 640K of RAM.

5. "Wild writes," write-accesses to memory that incorrectly modify a memory
location, can frequently be detected in protected mode through the mechanism
of a GP-fault (an unrecoverable application error). It is not possible to
detect these errors under real mode. These GP-faults provide information
about application bugs before the application is released.

For the reasons mentioned above, Microsoft is removing support for real mode from
Windows 3.1. For these same reasons, many developers have also removed support
for real mode from applications developed for Windows 3.0. Applications that are
written to support only protected mode should be marked with the Resource
Compiler's -T switch to prevent the application from loading in real mode.

Removing support for real mode also benefits the end user because applications
run faster and are more reliable. While small applications run quickly in real
mode, larger applications run slowly. However, in protected mode, large
applications also run quickly. A collection of large and small applications can
be run simultaneously without any loss in speed. For example, when more
applications are running simultaneously than can fit in the physical memory
installed in the system, the paging mechanism (only available in enhanced mode)
intelligently manages virtual memory to keep the most-frequently used memory
pages in physical memory. This management speeds up the system.

Disadvantages of Removing Real Mode Support
-------------------------------------------

1. The installed base of 8088-based and 8086-based machines cannot use the
software.

2. Code that performs segment arithmetic cannot be used in protected mode.
Therefore, some drivers and MS-DOS programs that run in real mode must be
rewritten for protected mode, or they cannot be run under Windows 3.1.

Source: https://jeffpar.github.io/kbarchive/kb/078/Q78326/

Edit: Real-Mode applications do of course still run on Windows 3.1x, in both Standard-Mode and 386 Enhanced-Mode.

Most, at least. There are a few old applications from Windows 1 and 2 days which tinker directly with memory addresses and may crash.

Those are usually the ones not marked clean/Windows 3 friendly.
Windows 3.x will display a compatibility warning dialog for those, but lets the user run them, nevertheless. In stark contrast to Windows 9x.

These are the ones the Windows 3.x Real-Mode kernal was meant for.
It would run them as good as real Windows 1.x and 2.x would.

Windows 2.03 and Windows 3.0 are very useful for bringing Windows to i80186 powered HP 200LX palmtop pc's since those versions are working in real-mode . But once they are installed there they are totally useless.

I do wonder whether we could build some cool programs for Win1/Win2 that would give it some more usefulness, but what would you gain over writing those programs for MS-DOS?

You'd probably gain more with Win2 than with Win1, given Win1 is little better than DOSSHELL.