@Scali Thank you very, very much for the help! 😁
More than often, I write quite a lot but have difficulties to express what I really want to say. 😊

Jed118 wrote:

This looks like some debug level *ish.

Apparently, yes. But let's also keep in mind that the 80286 was released in '82.
At the time, I think, neither the original IBM PC or DOS had much sofware yet
(beware, my opinion is only mainly based on old magazines of the time. 😅) :

Quite a few programs were still beeing made for BASICA/ROMBASIC and some systems had as little as 64K RAM.
So at the this time, Protected-Mode was barely used. In addition to this, the 286 was a complex design for its time.

It was made using a structure width of 1.5 Microns and had a transistor count of 134.000 (8086, '78: 29.000 transistors, 3 Microns).
This makes me believe that development of the CPU took at least a year, maybe more (speaking under correction).

So we have to imagine we were in the year 1980 when the specs for CPU design were set in stone.
That's about four years before the official release of the first IBM AT.

(By the way, later versions of Himem.sys used an undocumented instruction of the time,
LOADALL, on 286 PCs to avoid the need for switchting between PM and RM.)

Jed118 wrote:

I tried win /r and it would not take. If there's a version of Windows 3.1 out there that will run real mode

Hi, could it have been that you had a preview/alpha release of Windows 3.1 ? 😀
I believe I once read at BetaArchive (?) that early development versions of Windows 3.1 still had the /r switch functional.

As far as I know, the normal releases of Windows 3.1 only had two kernals, though. One for the 286, the other one for the 386.
If I'm not mistaken, the 386 kernal is always prefered on 386+ machines independently of the actual mode (Standard/Enhanced).

I *guess* this is also why some people have reported that they succesfully ran WfW 3.11 (not plain 3.11) with the /S switch,
if they used win.com or other files from vanilla Win 3.1 (haven't checked myself yet; did anyone else ?).
I don't know for sure where this design change came from, since Windows 3.0 used the 286 kernal if /S was used.

Jed118 wrote:

goddamnit I'll redo the experiment, but to be completely honest, I'm going with the notion that my Compaq 386/20e actually had 1Mb RAM on it and I simply forgot it. More and more things are pointing to that, internally and externally.

Please don't be mad about it. This thread was really interesting so far, I think.
Personally, I never had thought that Windows can run with less than 1MB of available memory, yet it does.
Unfortunately, none of us non-Compaq 386 owners really knows your old setup. 🙁

I mean, the original Deskpro 386 was a special piece, anyway, since zt was the first PC with a 386 CPU.
Maybe there was sort of an XMS or Extended Memory simulator included that you used.
For EMS, in comparison, I believe third-party LIMulators were sold commercially that used a swap file on a HDD even.

AFAIK I was using standard Windows 3.1 and I seem to recall DOS 5, later 6, and 6.22 when I got a handle on some diskettes. It's really quite impossible to verify exactly what version of Windows it had on there.

I'm happy with the in depth-ness of this discussion - I am learning quite a bit about memory management, between this thread and consulting my MSDOS manual. Screwing around with my 386 was fun.

Well, I edited the video: It's not going to be super-in depth and the editing is quick and dirty. I'm releasing this as a pilot for my new channel. I don't expect it to get to LGR's fame, or even Nostalgia Nerd, and it isn't as super thorough as Phil's Computer Lab, but there will be some neat retrospectives, Canadian Content (in French too!) and builds.

Please let me know what you think - I have pretty tough skin so give me some criticism, I can take it 😉

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

No critics ? Well, okay, then I'll reply first. 😁

Pros:
- Speech was clear and not too fast
- Video was well explained
- Configuration of the chipset was shown
- Usage of a tripod (no free hand recording)
- Transitions between videos was fine
(unlike how it is done in some commercial movies,
where they happen every 2 secs)
- light was fine (room not too dark/bright)
- DR-DOS was mentioned

Cons:
- The video seemed slightly stretched
- No 5.25" floppies were seen 😉

Well, for your entertainment and humoring of me - I can say that one time when I was just starting this back in 2001, my source of machines was an IT guy in the military in the 90's and he was on the MSDN lists. I remember the first time he told me he had some older Windows versions to give away and I thought 640K was all the RAM I had in my 486 at the time (I was VERY green - very new - at that time, like just barely graasp the concept of "Form factor" new) - well, he gave me Windows 3.0, which was wholly different han I'd thought it would be. And I can confirm 3.0 DOES run on 640K - my Tandy 1000 has 640K and runs Windows 3.0 on top of DOS 6.22, slow as molasses in january sure, but it runs. Windows 3.0 was the last version to use "Real Mode" - Windows 3.1 and 3.11 require a 286 or better, and 3.1x for Workgroups requires a 386 or better or it won't even install as the network stack used some i386 instructions IIRC.

creepingnet wrote:

And I can confirm 3.0 DOES run on 640K - my Tandy 1000 has 640K and runs Windows 3.0 on top of DOS 6.22,
slow as molasses in january sure, but it runs. Windows 3.0 was the last version to use "Real Mode" - Windows 3.1 and 3.11 require a
286 or better, and 3.1x for Workgroups requires a 386 or better or it won't even install as the network stack used some i386 instructions IIRC.

Windows 3.0 with its Real-Mode kernal also supported Expanded Memory.
If my memory serves, EMS was a requirement for desktop wallpapers in Real-Mode.
With EMS enabled, Windows 3.0 was actually usable on a PC/XT class machine.

Unfortunately, not many Windows 3 programs were able to run in Real-Mode.
For example, both Visual Basic and Turbo Pascal for Windows required one
of the Protected-Mode kernals (Standard/Enhanced).

Though at least TPW was able to produce pure 8086/Real-Mode applications.
Not sure where these restrictions originally came from.
I wonder whether windows 3.0 RM was ever meant to be seriously used or -as advertised-
if it only was merely included for maximum backwards compatibility with older applications.

The inclusions of the Windows 2.x colour palette and support for Windows 2.x device drivers
hints to that. But then why did MS even bother to implement a sophisticated EMS mechanism ?

The older Windows 2.x line also supported EMS, but only for DOS programs or EMS-aware Windows programs.
Even Windows/386 who was basically an EMS-emulating memory manager with a built-in Windows 2.x VM,
did not provide memory for Windows programs. Windows 3.0 in turn did. Even for Windows 1.x and 2.x programs.

No more critics ? Come on guys, take a heart!
Jed118 put some work into it and he deserves at least a little bit of feedback.
A few words about Pro/Contra shouldn't hurt anyone. 😀

Thanks!

I'm working on another one based around de-motoring (making everything solid state) on my 386, but because of some SCSI2SD issues (as well as, naturally, other IDE hurdles when trying to put new tech into a 25 year old computer) it's taking forever for the content to come together.

I've got PLENTY of ideas and even rough scripts, including French content (because Canada) but there's this thing called a life (aka wife), work, education, a project car, and trying to start a family that all get in the way! Lately my home network took up the bulk of my weekend...

I do some of the crude editing and even capture some footage of projects at work during lunch in our shop. I'll be there tomorrow soldering some pin extensions to the back of an old ISA I/O card. I can't wait to have the project completed, but so far it has been 15 years in the making....

You're welcome.

Good luck with your project! 😀