I can't tell if the Model 80 could easily be maxed when it was originally released. But it does seem it could at least run OS/2 with 2MB of RAM. And it would have been super expensive. OS/2 would have come some months later, so no one was probably gonna jump on it right away for that purpose. And the 386 16 MHZ was still pretty slow as far as 386 were concerned.

Xenix would have been available around the same time as OS/2? I don't know if the 386 version was any good. But I see IBM also provided AIX for their PS/2 line a bit later. It seems interesting that it also had an option to run a real mode / dos vm for dos applications. It also had X Windows. It would have been superior to OS/2 initially. But this configuration would have been super expensive. I can see that Windows ended up bringing 32-bit computing to the masses when it was bundled with PCs.

the3dfxdude wrote on 2023-02-27, 15:18:

I can't tell if the Model 80 could easily be maxed when it was originally released. But it does seem it could at least run OS/2 […]
Show full quote

I can't tell if the Model 80 could easily be maxed when it was originally released.
But it does seem it could at least run OS/2 with 2MB of RAM. And it would have been super expensive.
OS/2 would have come some months later, so no one was probably gonna jump on it right away for that purpose.
And the 386 16 MHZ was still pretty slow as far as 386 were concerned.

Personally, I can't say much about it, either. :(
The Model 80 was built like a tank and surely impressive at the time.
I assume it didn't have the same bottlenecks as the humble AT 5170, so it really was an improvement.

But even the PCem emulation of the AT isn't feeling that slow, actually, considering all the limits: Slow RAM+CPU, "ISA" at merely 6-8 MHz; clones ran it at 10/12MHz.
If I equip the emulated PC with the same amount of RAM I had used for my 286-12 w/ Windows 3.10 in the early 1990s (4 MB total), then OS/2 v1.1 flies.
It really feels like I'm using a weird copy of Windows 2.03, with a slightly more snappy GUI.

Interestingly, that's also what an 1988 article mentioned (attached/linked), albeit with a professional background and higher expectations than me.
An DTP artist at work has another idea of "smoothness" than a hobby or calm business user, I suppose. Please notice the network use, also.
So in practice and with a snappy CPU, maybe 640KB+2MB XMA would have been sufficient for a light workload with OS/2, who knows ?
Networking consumes RAM, but a real multitasking OS has better network performance, than DOS, on the other hand. So things end up even ?

Edit: Considering that a base line PC already had not only 640 KB, but a complete megabyte:
Wouldn't have a single 2 MB expansion memory board solved the memory issues with OS/2 ? As it was sold by IBM for ATs ?
3 MB total aren't much, but not far off from 4 MB and close enough to get OS/2 1.1 as such booting nicely (~27 sec vs ~2min with bare minimum config).
Sure, that must have been an ~1000 USD investment at the time. But in relation to the what the rest of the PC already costed..
Also, there's still the tax refund option for the professional users/companies. But they never like to talk about that, I suppose (hot topic).

That's why I'm somewhat surprised or perplexed why the press was so negative about OS/2 v1.1 or v1.2.
I mean, what were they expecting? Physics can't be defeated. The entry class PCs sold in 1988 -just checked- were sold with a bare 1MB. Yes. Of course!
- Because a) plain DOS was so limited, anyway and b) it was the maximum of 8088/8086 PCs. The 1MB configuration wasn't state-of-the-art, however! In the eyes of the sellers adding more was just a waste.
And that wasn't even wrong, if we talk about casual users, or users of a single application (some maybe hated PCs, but did depend on this one, single program).
By contrast, professionals, power users or advanced users never went with the base line setup, anyway. Not for long, at least. That's why the ads proudly say "upgradeable up to 12 MB!" or "upgradeable up to 16 MB!"

Another reason for keeping the RAM count low was price, but with a twist: Actual RAM prices were under subject to change daily/weekly.
Like with the oil/fuel price, essentially. So sellers couldn't predict price drop/increase. Fine detail here.

If you were in the printing business (from word-processing to DTP), you didn't have CGA graphics and an 640KB of RAM for long, but rather went the route of EGA/VGA if not Super EGA/Super VGA or 8514/A.
And an extra 256 to 512KB for the printer spooler, maybe. You also bought a RAM and font cartridge upgrade for your precious HP LaserJet Plus (or similar).
The HDD ("fixed-disk") also was increasingly getting smaller, so you sold the old one as used, maybe, and bought a bigger one.

Personally, I think that PC history was all but a big misunderstanding. Or maybe, still is.
Young writers of "retro articles" who weren't around at the time do sometimes write quite some nonsense based on clichés. ^^
I often notice this when they do write how primitive the 80s/90s were, fully ignoring things like IBM's 8514/A in 1024x768 in 16c/256c (1987).
Or, in case of DOS games, the existence of Standard VGA (640x480) and Super VGA (800x600) titles.
The shareware/freeware scene of the 80s/90s, is also often ignored, I think.

If we try to interpret things in context or try to read between the lines (the ads with the catch phrases), then the story comes out quite different.
And not seldomly, starts to make sense. That's why I'm always glad if people ask questions and do experiments on their own. ^_^

Speaking of, here's OS/2 v1.1 in hi-res. :D

the3dfxdude wrote on 2023-02-27, 15:18:

Xenix would have been available around the same time as OS/2? I don't know if the 386 version was any good. But I see IBM also p […]
Show full quote

Xenix would have been available around the same time as OS/2? I don't know if the 386 version was any good.
But I see IBM also provided AIX for their PS/2 line a bit later. It seems interesting that it also had an option to run a real mode / dos vm for dos applications.
It also had X Windows. It would have been superior to OS/2 initially. But this configuration would have been super expensive.
I can see that Windows ended up bringing 32-bit computing to the masses when it was bundled with PCs.

I wonder the same. My father's PC software from the 1980s is pretty much 16-Bit, from what I can tell.
So I assume that's also true for the Xenix version I remember. I mean, people back then were already glad to move away from 8-Bit computing.
Atari ST, Amiga, etc. they all were 16-Bit machines that could address a couple of megabytes of continues RAM, as opposed to 64 KB.
- Which ironically is the segment size used by the 8086/8088. The 64KB issue appears again and again on the PC platform: EXE models (small, compact etc), VGA frame buffer (256KB; 4x 64KB), EMS page frame etc.
Theoretically, the 80286 could go away with it - indirectly. It uses segmentation still, but it can use pointers. 16-Bit DOS Extenders existed, too, which is less tedious than using XMS all the time.

DOS VMs of the 1980s are very interesting, I think!

Here's an excerpt fromt "DOS Merge" software for AIX PS/2.

"A PS/2 computer with the AIX PS/2 Operating System installed.
A minimum of 3 megabytes of 32-bit memory.
We strongly recommend that you have more than the minimum required memory,
especially if you intend to support multiple, simultaneous DOS users.

A fixed-disk capacity that is sufficient to accommodate the AIX PS/2 Operating System
plus the following additional space for the DOS Merge system:

- 2500 1K blocks in the root (/) file system
- 1890 1K blocks in the /local file system

If you want to support DOS running on terminals or if you want to use
applications that use communication ports, you need serial ports.
(For example, one serial port is standard on a PS/2 Model 80.)

- The AIX PS/2 operating system.
- The AIX PS/2 DOS Merge software (a 3.5-inch, 1.44MB diskette).
- The DOS 3.30 Startup/Operating diskette (a 3.5-inch, 720KB diskette).
(Ed. AIX 1.3.0 needs the 5.0 Startup/Operating diskette)
- The IBM PS/2 Reference Diskette.

The Reference diskette is not required during the initial installation of DOS Merge,
but you may need it to make optional adapter cards usable by DOS Merge.
(Refer to "Installing or Removing Plug-in Cards" in topic 2.8.11.) "
Source: SC23-2045-0 EZ12QMST.BOO

That being said, what's always confusing are the memory allegations: Is base memory ("system memory", first 640KB to 1 MB) part of it, or not ?

Personally, I think that all those requirements make sense only if they are about extra memory.
Memory, which the user/operator must install or buy optionally for the product to work.
Memory, which the computer BIOS does separately count up during POST.

Otherwise, it wouldn't make much sense, either. I mean, you don't mention something than is taken for granted, after all..
Like, the DMA controllers or PIT/PIC on the motherboard. Or an WD1003/IDE HDD controller etc. Or base memory, cache memory.

Indeed, from what I remember, 1MB simms continued to be pricey into the early 90s. Then you have the issue of software. There wasn't gonna be much native software on something completely new. So you aren't gonna start decking out a system with more memory just for the opportunity to run the same software in OS/2. I think that's why in the infoworld article, you still have companies (who could afford it) saying they were essentially going to trial OS/2 and think about a roll out a year from then. That basically puts you at the end of 1989!

I believe you that OS/2 may be quite fast, once it has enough memory to run the VM or multiple applications. I haven't seen OS/2 myself but I am curious. I remember running Windows, I believe on a 286 with 1MB, and moving to a 386 with 4MB was definitely speedier. So were both OS/2 and Windows competitive in 1990-1992? I do want to see a face off, if OS/2 had a chance at being better. And if it could run all the same software people used then. There wasn't much Windows software before 1990, but it quickly came. So this was a critical period to persuade people.

When people started getting the RAM installed, I guess OS/2 would have had a chance. But we know it came up short, probably because of the split and Microsoft bundling Windows on PCs.

Here is an article about ram prices from 1985 - 1988. In summary, they say the producers overestimated demand and produced too much. It went from $300 per MB in 1985 to $133 per MB in 1987 to $500 per MB in 1988. DOH! There is also a bit on trade practices like what's happening today.
https://tedium.co/2016/11/24/1988-ram-shortage-history/

the3dfxdude wrote on 2023-03-01, 16:16:

Here is an article about ram prices from 1985 - 1988. In summary, they say the producers overestimated demand and produced too much. It went from $300 per MB in 1985 to $133 per MB in 1987 to $500 per MB in 1988. DOH! There is also a bit on trade practices like what's happening today.
https://tedium.co/2016/11/24/1988-ram-shortage-history/

Thank you very much! That's an important piece in the puzzle, I think.

the3dfxdude wrote on 2023-03-01, 15:38:

Indeed, from what I remember, 1MB simms continued to be pricey into the early 90s. Then you have the issue of software.
There wasn't gonna be much native software on something completely new.
So you aren't gonna start decking out a system with more memory just for the opportunity to run the same software in OS/2.

That makes sense, yes. I mean, Grzyb wasn't exactly wrong with what he said about 1 MB systems being popular, also.
It's just that the basic models are rarely being usable beyond their original use case/application (say running DOS 3.x, GW-Basic, Word Star, MS Flight Sim).
I suppose that's what the different types of users sets apart. The home/private user just wants to get a PC, then leave ASAP.
The other types do consider how to integrate the PC into their business, into their application the best.
They're much more concerned and try to plan ahead. Better safe than sorry, so to say.

the3dfxdude wrote on 2023-03-01, 15:38:

I think that's why in the infoworld article, you still have companies (who could afford it) saying they were essentially going to trial OS/2
and think about a roll out a year from then. That basically puts you at the end of 1989!

That article was a fascinating read, IMHO. Ironically, I didn't even look for it, I've found it while searching for something else.
Anyway, you could really "feel" how the representatives struggled to explain themselves. 😁
The mentioning of the many 640 KB-only systems made them look much more desperate, even.

And that's kind of paradox, too.
Because, Windows 2.x or Windows/386 doesn't do solve the memory issue, rather the contrary.
It needs less memory for itself, sure, but also provides little to the applications.
Meaning, Windows applications on Windows can easily run out of memory. A horror scenario at work!

Let's image working on a document and a "not enough memory" window pops up in the middle of the work.
The clock thickens and you can't get your work done. You can't print what you have, you can't even save the active document.
On OS/2, this wouldn't happen that way. The HDD would start to roar, maybe, but you could at least finish your work.

..

All what Windows/386 could do was emulating EMS (on a 386+) and hope that both Windows/DOS applications use it.
Not so with OS/2: OS/2 had virtual memory and could provide megabytes of RAM to applications evenly (except pure DOS applications).

That's why Windows 3.0 was such a big improvement, by the way, I think.
It not only did look like OS/2 1.2/1.3, but also had both virtual memory and EMS support (in reverse).
And it could multitask DOS applications on a 386, albeit poorly at first (see Win 3.0 Read Me about compatibility issues).

Old Windows applications, with EMS support or not, could get megabytes of memory via Windows API.
Even in Real-Mode: With Windows 3.0, Windows itself was now using EMS provided by the system (via chipset, LIMulators, EMS boards, EMM386/QEMM or OS/2).

In other words, Windows 3.0 was trying to mimic both the appearance and features of the latest OS/2 at the time.
With Adobe Type Manager, some early form of True-type support was available, even. A god send for DTP users and word-processing people. ^^

The other way round also happened. Those Willow "sample applications" resembled the whole Windows 3.0 accessory group!
On OS/2 v1.3, this demo package must have been an unofficial must-have at the time.
I don't think that was a coincidence. In retrospect, this maybe was a very ingenious, if not bold move. But.. For who and why ?

Were the WLO applications made available for free to attract OS/2 users to Windows ?
Or were they made available to satisfy OS/2 users at their workplace, to make life easier to them ?
So they don't feel the urge to switch to Windows so much ?

I really wonder what the real intention of it was - maybe that's one of those mysteries that will never be solved! 😁

the3dfxdude wrote on 2023-03-01, 15:38:

I believe you that OS/2 may be quite fast, once it has enough memory to run the VM or multiple applications.
I haven't seen OS/2 myself but I am curious. I remember running Windows, I believe on a 286 with 1MB, and moving to a 386 with 4MB was definitely speedier.

Yes, that's about the same with OS/2 1.x/2.x, I think.
With the minimum RAM configuration, OS/2 runs like Windows 98 on a slow 486 with little RAM.
Except, that OS/2 itself doesn't roam the HDD if being idle. Windows by contrast constantly indexes stuff, looks for autoplay etc.
With OS/2, it takes about two minutes until the desktop appears and the HDD access stops.

With ~3 or 4 MB of memory, OS/2 v1.1 did finish booting in 27 secs (HDD activity stops),
however it was already usable a few seconds earlier. Say roughly 20 secs or so. On the second-slowest AT (8 MHz vs 6 MHz).
With a 386 emulation, OS/2 v1.1 booted up in 6 secs or so.

the3dfxdude wrote on 2023-03-01, 15:38:

So were both OS/2 and Windows competitive in 1990-1992? I do want to see a face off, if OS/2 had a chance at being better.
And if it could run all the same software people used then. There wasn't much Windows software before 1990, but it quickly came.
So this was a critical period to persuade people.

Edit: I'm sorry I forgot this one here. Um, yes, I think so.
But since Windows 3.0 has three different personalities (modes) it's tricky to make a fair comparison.
Maybe I'll do some testing in the near future and make a video once I found out! ^^

the3dfxdude wrote on 2023-03-01, 15:38:

When people started getting the RAM installed, I guess OS/2 would have had a chance.
But we know it came up short, probably because of the split and Microsoft bundling Windows on PCs.

Yes, I think the same. The split-up wasn't exactly ideal at the time.
But it's understandable, kind of. Both Apple (one of its founders, at least) and IBM felt betrayed by Microsoft.
So a split was inevitable, maybe. just the timing wasn't exactly nice. Not for the companies, not the customers/users.

That Microsoft dev essentially said that OS/2 without at least some Windows API compatibility was considered as doomed.
I suppose that even IBM realized this in secret, without being willed to give in and admit it openly.

That's why Win-OS/2 came to life, essentially. IBM wasn't willed to include little WLO or a similar API/shim,
but then simultaneously made a bold move and integrated the complete Windows 3.0 package straightaway (in OS/2 v2.0, 32-Bit).
As if they wanted to say "See, we're better than you. In fact, we own you.". Or something like that.

On the other hand.. Being able to virtualize a complete, up-to-date Windows installation.. This was a great marketing stunt/good public relations.
I mean, what else could demonstrate the superiority of OS/2's built-in DOS support ? Except the classic MS Flight Simulator test, maybe.

Also, they could integrate all those little utilities that were severely being missed by OS/2 users. Calculator, Write, Notepad, Terminal etc.
With the inclusion of Windows, they could get away with it, without needing to justify themselves to their business partners.
They could say, it was because of compatibility, while simultaneously saving face. Clever.

Oh well, it's hard to make head or tail of it. WLO was so much more elegant and quicker, if it had taken off.
Preemptive multitasking, virtual memory, no XMS/DPMI in the way..: WLO-enabled Windows applications could have had run better, while needing less resources.
If Willow had made it, it wouldn't have been necessary to run multiple, memory-hungry copies of Win-OS/2 for the sake of stability..

On the other hand.. Win-OS/2 was constantly being upgraded because MS Windows was still a living product/thing at the time.
With WLO alone, the Windows platform wouldn't have had continued to evolve, maybe. Hm.

Edit: Pictures added.

Jo22: I have just read the three pages of this forum thread, and I must say it is an absolutely awesome thread. And I hope/feel/know this thread will hopefully be ongoing 😀

WolverineDK wrote on 2023-03-03, 01:54:

Jo22: I have just read the three pages of this forum thread, and I must say it is an absolutely awesome thread. And I hope/feel/know this thread will hopefully be ongoing :-)

Thank you very much! These kind and uplifting words are motivating and mean a lot to me! ^^
I'm often worried that I write too much and do annoy people. ^^;
Speaking of, I think that I wasn't as respectful torwards Grzyb as I should have been (sorry again), which bothers me.

- I'll try to do some more experiments in the near future.
There are quite a few more interesting things/topics that come to mind,
like unusual versions of Windows 3.x no one ever has heard of..
Or Minix on a 286, Soviet clones of the 80286 CPU, EMS boards under OS/2 etc.

I can also try to run OS/2 1.x on real hardware and see how it works.
Or run some benchmarks - Windows 2.x vs OS/2 1.1 and Windows 3.0 vs OS/2 2.x..

I just hope that others will join the thread again in the future.
I'm always fascinated to see other point of views or learn about new information. ^^

Edit: Oh, and I will try to compile some Willow compatible applications in the near future.
Once I figureed out how to setup Windows 2.x or 3.0 development software..
There are some more interesting things to try, I think.
For example, well written Windows 2.x programs execute without a compatibility warning on Windows 3.0.
Once they're "marked" correctly in their header, they are both valid Windows 3.0 and Windows 2.x applications.
Such applications could run -perhaps- under both Windows 2.x and the Willow (WLO) compatibility layer.
That would be fascinating to test (are the control elements looking like on Windows 2.x or 3.x?)

Jo22: Even though I know Jack and S , when it comes to programming , and Jack has left town for quite a while. Then I still think it is awesome to read such great topics. Cause it makes me think of great articles in great old magazines of a far gone time. But still it has enough to gain some understanding from a non programming side, but understanding things around it. Especially when it is written in a language I understand, and you writing too much ? Come on mate ! what a load of rubbish to say 😀 It shows you have something to say in an intelligent way ! 😀 Also I love learning history, even if it is about computer history , so good luck on you programming some stuff too ! 😀 Keep up the smashing great work ! 😀

WolverineDK wrote on 2023-03-05, 19:53:

Cause it makes me think of great articles in great old magazines of a far gone time.

If one were to accidently search "Atari Magazines" they might find themselves at a classic computer magazine archive with 80s and 90s issues of Compute! with all those fun to try programming tips and articles.

BitWrangler wrote on 2023-03-06, 03:58:

WolverineDK wrote on 2023-03-05, 19:53:

Cause it makes me think of great articles in great old magazines of a far gone time.

If one were to accidently search "Atari Magazines" they might find themselves at a classic computer magazine archive with 80s and 90s issues of Compute! with all those fun to try programming tips and articles.

Indeed 😀 also I know the sentence is normally written as "from a bygone age" or "from a begone era" . Anyway Jo22 you are doing a smashing great job, and keep up the good work 😀

Does this Infoworld edition in 92 indicate the end of OS/2 was near?
https://books.google.com/books?id=aVEEAAAAMBA … epage&q&f=false

They were talking about the Windows support lacking and it was hard for corporate managers to justify buying it anymore. Maybe they fixed it, but if after 92, now getting much too late. Makes you wonder if OS/2 2.0 was not working well enough on release, or maybe more disinformation by competition.

@WolverineDK Thank you! ^^

the3dfxdude wrote on 2023-03-06, 16:05:

Does this Infoworld edition in 92 indicate the end of OS/2 was near? https://books.google.com/books?id=aVEEAAAAMBA … epage&q&f=f […]
Show full quote

Does this Infoworld edition in 92 indicate the end of OS/2 was near?
https://books.google.com/books?id=aVEEAAAAMBA … epage&q&f=false

They were talking about the Windows support lacking and it was hard for corporate managers to justify buying it anymore.
Maybe they fixed it, but if after 92, now getting much too late. Makes you wonder if OS/2 2.0 was not working well enough on release, or maybe more disinformation by competition.

Thanks for the link! ^^
Yes, that could well be the case. The writer surely had some sort of bad premonition here, at the very least.
Interestingly, the article on the right about the Pentium chip and its issues is also fitting. Thanks again for the link.
Because, from what I found at Gunkies, "IBM had the 615 in their pocket which was a
PowerPC CPU which was pin compatible with a 486, and could run x86 code (albeit slow..) and then switch to PPC mode. "
Very interesting, I didn't know that. So there was another chip trying to replace the 486.
The quote is from the text block about the PowerPC port of OS/2.

Edit: Some more information about the PowerPC 615..
https://everything2.com/title/PowerPC+615
https://www.cpushack.com/CIC/announce/1995/PowerPC615.html
https://www.theregister.com/1998/10/01/micros … ed_the_powerpc/

Speaking of OS/2 2.0, I did find some information on a site of an Autodesk founder.
Autodesk did apparently consider supporting OS/2 again in version 2, if it would turn out to be successful.

These two are an interesting read, I think.
My First Experience with MS Windows (archived version, also still online)
AutoCAD for Windows

Also from the Gunkies site is a (dead) link to a Willow (WLO) press release.
https://web.archive.org/web/20040905164813/ht … ory/pr/wlo.html
About 1300 companies/developers participated the beta test program (WLO v0.9 and before), if I understand correctly.
- Gupta with SQLWindows really was part of the testing program, interesting. 😁

PS: What's being missed on the Gunkies page is the following information:
OS/2 1.3 apparently has issues with IDE emulation. Once you select "AT Fixed Disk Adapter", IDE's precursor, then it boots without hanging.
The only drawback is that 386/486 BIOSes can't use auto-detection with MFM/RLL controllers - manual input works fine, though.
Max CPU performance for older OS/2 versions should be the equivalent of a 486DX-50 or so (roughly; there are workarounds for higher CPUs).

Quick update. Also interesting in that magazine issue was an article about Windows 3.1 in ROM.. 😀
But what really surprises me was an ad about an 486 EISA workstation PC.
For 1992, the advertised specs were quite high. Chipset (?) support for up to 128 MB of RAM, for example.
Even to my taste that was a bit too high. 16, 24 or 32 MB, okay.. But 128 MB on-board, without special memory modules ?
For a PC system that targets the end-user market through a PC magazine, that's quite high.
It would require 32x 4MB SIMMs (30 pin). In the CAD/CAM field or photography/DTP field, that RAM amount might be useful, sure, but..

The 386 would have opened up to these high ram systems, but as we have mentioned, ram prices were still fluctuating and crazy in the late 80s. And that really stunted OS/2, didn't it? There is another recent thread on vogons where we looked into high ram systems on the 386, and while we determined something like 128mb was possible, it would have been insanely expensive. And at least compaq, it would have been proprietary modules too. By this point in 1992, it would have been definitely cheaper, making interest possible for 2-4mb for a base system. I have a 486 that can take 128mb, and would be especially nice if it had the correct tag and cache ram for that. But yeah for the home user, 2-4mb systems were starting to become common, although there was still one low end 386sx system I saw in there that sold base with just 1mb still then! I mean that's what alot of us (including me) were still running only 1mb in 1992. But I still remember what a game changer having more ram allowed with windows, and a few new games... Anything more than 4mb then was a crazy large amount for a home user.

Jo22 wrote on 2023-03-08, 11:21:

Thanks for the link! ^^ These two are an interesting read, I think. My First Experience with MS Windows (archived version, also […]
Show full quote

Thanks for the link! ^^
These two are an interesting read, I think.
My First Experience with MS Windows (archived version, also still online)
AutoCAD for Windows

Interesting! And a bit confusing!

By late 1990, we realized that there was a lot of conflict between Microsoft and IBM about the future directions of OS/2. At one point, our Microsoft contacts, urged us to consider AutoCAD for Windows 3.1, which was about to be released. I was, however, very skeptical since Windows 3.1 was a 16 bit operating system, and on a Window system, AutoCAD really required a flat address space only available on 32 bit operating systems.

On July 5th, 1991 two months before the announced Microsoft/IBM divorce, Cameron was in the room when Bill Gates trumped Kern Sibbald's soliloquy on how to promote OS/2 by leaning forward and saying very simply: "We don't want to sell OS/2".

It was a stunning revelation. The room was filled with the MS OS/2 lead programmers. They were there to celebrate the first 32-bit application running on OS/2. When Bill spoke those simple words, you could see the blood drain from their faces as their heads filled with random white noise.

OS/2 2.0 was the first 32bit version of OS/2, right, and came out in '92? And they were there in '91 to celebrate the first 32bit application, which could not have been AutoCAD, because it involved MS programmers. I mean that part that it was coming from MS programmers makes sense, because no one else would have had prior access to the proper 32bit flat address space compilers and apis to do it. It was a very sly thing for Gates to pull this and tell the Autodesk team that they were pulling out of OS/2!

AutoCAD could not have been a 32bit application already, by this statement. I think he *hoped* that they'd finally move to that based on the announced meeting, and Gates's move in the meeting held them back if OS/2 was DOA and Windows was it. Somehow they knew, even if Gates could not publicly admit it.

Another thing said was this in '92 on the Windows version of AutoCAD:

Shoehorning a huge program like AutoCAD into the 16-bit architecture of Windows 3.0 was a Herculean task, made easier by our initial investment in the OS/2 version of AutoCAD.

This means AutoCAD was and still was 16bit application, but being 16bit was not as bad as they made it sound to be. It's just Windows was a different API It just meant they couldn't land the 32bit version as fast as they initially planned. I guess OS/2 was a bit over promised, wasn't it? It just was too late into action. I'm sure AutoCAD wanted to be on the inside to have an edge over their competition when 32bit apps finally arrived. Was there actually a 32bit version of AutoCAD before 1992? I don't think the Xenix version would have been based on our earlier conversation.

Looking at AutoCAD version history, it looked like while version 13 was for Windows 3.1, a 16bit platform at the time, they also had a port for Windows NT Alpha? So they were still moving forward with a 32bit version I guess... whenever NT was ready. And this was the only 16bit windows version. The next versions were very likely 32bit only.
https://en.wikipedia.org/wiki/AutoCAD_version_history

the3dfxdude wrote on 2023-03-08, 15:16:

The 386 would have opened up to these high ram systems, but as we have mentioned, ram prices were still fluctuating and crazy in the late 80s.

Yes, that was really a bad accident. I had no idea it was that bad, to be honest.
The PCs released in 1985/1986 all had 640KB standard (or planned as a standard) and didn't make a fuss about 1MB of RAM.
That's when my father bought a lot of DOS software and PC hardware, also.
By the late 80s, he was busy moving to another city, so he likely missed the effect of the crisis.

The 1MBit production problem really was a back-throw to the whole digital industry, I suppose.
Interestingly, East Germany had a similar issue. They were trying to upgrade production facilities from 256kbit to 1MBit.
The process was somewhat resource hungry, that it harmed the 256kbit production.
There's an interesting video about it (How Semiconductors Ruined East Germany).

If you're curious about x86 and Z80 PCs from "the other Germany", please have a look at https://www.robotrontechnik.de/
Google Translator can help - it also has screenshots of the weird East German Windows 3.0.
Funny, isn't it ? Windows 3.0 was running like a continuous (red) thread throughout all PC history. 😁
https://www.robotrontechnik.de/index.htm?/htm … systeme.htm#kws

Even to me, as an ex-western neighbor, it's quite a trippy roller coaster ride.
Personally, I find those PCs quite good! The built-quality was fine, and their "ISA" connectors were professional, too.
Some even had Super CGA, sort of. They used the same video chips used by NEC in Japan.
More information (links) in an older thread of mine: Emulation of x86 PCs of former East Germany?

Sadly, the few former East Germans I talked to do have very little interest about in own creations of the past.
They either don't believe they had those PCs, or they play down their value. Sad, I think. They had good, skilled engineers.
But for some reason, they still believe that everything in "the golden west" was better than theirs. 🙁

If they only knew that many wonderful presents they got from their western relatives were in fact their own products.
Quick backstory: Here in W. Germany, there used to be three big and reputable mail order companies: Quelle, Otto, Neckermann.
They had about everything in their printed catalogues. From what I know, Quelle (German for "source") did import products from East Germany.
After import, many of those products were re-labeled and sold in their own catalogue.

The neat thing about those companies was, that you could send things back you didn't like and that you could pay via installment payment.
This was an alternative paying method to using credit cards, essentially. Credit cards were never really popular over here.
Business people and travelers had them, of course. Many ATMs also supported them, in addition to bank cards. Mainly for travelers or tourists.

Speaking of Quelle, it was one of the earliest companies in the country with an online shop, afaik.
It started it in the late 80s and was accessible via BTX, our version of French Minitel.
Essentially, you could buy things in a similar way to Amazon or eBay in our modern day.
I remember it, because a TV show for children featured actual footage of a beige IBM PC or AT connected to BTX/Quelle.
It was this one here. The main character was constantly buying useless stuff online in incognito, because she was non-human (like ALF). 😁

the3dfxdude wrote on 2023-03-08, 15:16:

And that really stunted OS/2, didn't it? .

Um, I think so, yes. 🤷 For sure I can say it did at least delay OS/2 long enough until Windows 95 took over the market.
Here in my place, OS/2 2.x had at least gained a notable popularity among PC enthusiasts.
Which in turn continued to live on into the Warp 3 days. But I can't remember exactly when things went down hill.

I vaguely remember that OS/2 Warp was sold pre-installed with PCs made by Escom and Vobis, two major PC distributors.
Microsoft was really p*ssed about it at the time. 😜 If not a bit shocked or traumatized (since they dared to refuse paying the MS "tax").
To MS it was as if Dell suddenly started to ship PCs with the latest macOS instead or something. 😁
Unfortunately, those PCs with OS/2 did merely have the minimum 4 MB of RAM. Not a great user experience.

In the 1991-1994 time frame, approximately, OS/2 was considered a real competitor over here to Windows/DOS.
In that time, both the good old memories of OS/2 1.3 and the latest news about the great new 32-Bit support were still fresh.
To all the people who struggled with DOS and Windows 3.0, OS/2 was seen as a way out of that mess.

That's why OS/2 was being advertised as "Migrationsplattform" (migration platform) in various PC magazines around the time.
Thanks to the emulation/virtualization features of OS/2's DOS VMs, many things worked that previously failed.

For example, it was possible to give DOS-based compilers and other command line utilities up to 736 KB of conventional memory, on any PC!
All it needed was a simple mouse click in the options of the DOS support (it was done by restricting video modes to CGA, essentially).
And such a setting could be used both globally and application-specific (PIF files).

And as long as Windows 3.x applications were still #1, OS/2 was on par with its built-in Win-OS/2.
OS/2 even supported long file names at the time. On DOS-formatted floppy disks, it stored them in EA files.

When Win32s applications for Windows 3.1x appeared, OS/2 started to struggle a bit.
While Win32s 1.25 was successfully patched to work on Win-OS/2, programs with need for the more finicky Win32s 1.30 appeared soon.

That's when things got a bit unstable, I think. The increasing use of VXDs as DLLs in Windows applications also caused some headaches.
WinG and other extensions worked quite okay, though. Except DCI, maybe, have to check.

All in all, I think that the end of Windows 3.1x as the dominant PC platform also marked the end of the OS/2 days.

the3dfxdude wrote on 2023-03-08, 15:16:

There is another recent thread on vogons where we looked into high ram systems on the 386,
and while we determined something like 128mb was possible, it would have been insanely expensive.

Makes sense. Normal users (us mortals, so to say) had no urgent need for such an amount of RAM at the time.
But considering the fast PC evolution in the 1990s, no one was sure about the technological development, maybe.
So the chip sets were being made to support that amount of memory, just in case. 🤷

On the other hand, if something revolutionary or innovative like "bubble memory" really had been taken off at the time..
Then maybe we would have had gotten very cheap RAM with high capacity, passing the 1GB mark by 2000s.
https://en.wikipedia.org/wiki/Bubble_memory

..
The 90s were a weird time, I think. So many things came and went in a blink of an eye. I'm just learning about many of them, in fact. 😅
By 1992, clumsy "old" Windows 3.x was perhaps very beloved because it "just worked" most of the time and provided a form of hardware-abstraction to what
both users/developers felt to be an ever changing IT environment with an uncertain future (with a mood similar to what viewers experienced in the first two Terminator movies).
With a Windows application, you simply had a peace of mind. You knew your applications would run for the next years on the most obscure hardware, including RISC.
Interestingly, Linux/Android/Java play a similar role these days. They act as a device driver package+runtime combo to the applications.

Speaking of the late 80s/early 90s, I can't help but feel that I must say that I feel that the business climate was different, more relaxed.
Here in Germany, companies like Microsoft or Nintendo had "just" freshly opened their new outlets (okay, MS was here since '83, but big N appeared in '90).
Not sure why. Maybe because it was a good location to do business at the time, not sure. 🤷
Or maybe it was just a symbolic thing (being on very edge to west-east, then the fall of the iron curtain, the reunion).
Anyway, the companies were were friendly/helpful towards all kinds of business partners at the time.
You could ask for free advertising material, catalogues, for SDK stuff (print outs/listings of undocumented API functions) etc.
Even if you were just a one-man company, essentially. The time was a weird cross of 80s/90s mentalities, afaik.

I suppose that's why Microsoft products were so popular here, they had good "end-user" support without the need for having a lawyer at your side. 😉
Many developments, both commercial and in-house applications were made for the Microsoft platform back then. Or so it seems to me.
That's why OS/2 2.x had such a potential over here, I think. It could integrate all sorts of applications easily (in theory).

To give an idea, all of our then-new and then-fancy ISDN stuff was made with Windows-like PCs or special hardware in mind.
Here's an old 1992 video (meant for the Telekom employees ?) about the shiny new world of ISDN, if you look closely, you see Windows 3 on some PCs.
It was probably required for operating those installed ISDN cards (ISA) at the time. On plain DOS, there perhaps was no API/driver available. It's just a guess, though. 🤷
https://youtu.be/1lYFr-C_SKg?t=836

OS/2 2.x had a real chance to integrate itself into this ecosystem, I think. If it only was less stiff. Plug&Play, APM, PCI, ATAPI CD-ROM drives (just created)..
That's something that OS/2 had trouble with. Being a direct OS/2 1.x descendant, it still expected 1980s-era hardware and early interfaces (SCSI CD-ROMs, Mitsumi Lu005s interface card etc).
It catched up, eventually. But it was late. By the time OS/2 3.0 was around (1994), the betas of Windows 95 were already out and the press wrote about them.

If only OS/2 2.0 in 1992 would have had the same feature level like Windows 3.1 had right from the start (APM, MIDI/audio/video, both 16+32-Bit drivers, nice screensavers)..
But that wasn't the case. OS/2 2.0 was late due to the IBM/MS divorce and it was functionally very close to Windows 3.0 still (okay, but not great).
32-Bit graphics drivers (non-GRADD) and the MultiMedia Presentation Manager (MMPM/2) were provided as updates over the next two years (in the OS/2 2.1x life time).

By comparison, Windows 3.0 with Multimedia Extensions 1.0 (Win3 MME) was around since late 1991 (to OEMs).
And the magic screensaver, a predecessor to AfterDark, was around even earlier in the Windows/386 days.
https://www.youtube.com/watch?v=gCV8OPVY-uE

On the other hand, to OS/2's defense, it got GRADD graphics drivers by the release of Warp 4 (it had VESA VBE support before Windows).
Their capabilities were better than what Windows had at hand at the same time ('96; Scitech and other companies had worked on GRADD).
The early Windows 95 RTM used simplified 3.1 drivers, essentially, with their GDI components stripped out and no DirectX support yet.
GRADD (Graphics Adapter Device Driver) in turn was a completely new development, a left over from the PowerPC port of OS/2.
http://www.os2museum.com/wp/display-drivers-o … 16-bit-windows/
http://www.os2ezine.com/v3n04/gradd.htm

DIVE (Direct Interface Video Extension), a DirectDraw or WinG style API, also was added a bit earlier to OS/2 in 1994. It was meant for OS/2 2.11 and 3.0, afaik.
Also, the OS/2 team further optimized the OS for low-end hardware (386SX, 4MB RAM), maybe with laptop users in mind.
https://stuff.mit.edu/activities/os2/diverepo.html

the3dfxdude wrote on 2023-03-08, 15:16:

By this point in 1992, it would have been definitely cheaper, making interest possible for 2-4mb for a base system.
I have a 486 that can take 128mb, and would be especially nice if it had the correct tag and cache ram for that.

Cool! I would love to see that one day! 😀 👍

the3dfxdude wrote on 2023-03-08, 15:16:

And at least compaq, it would have been proprietary modules too.

Oh yeah, Compaq! They were always proprietary! 😁 I'm still "happy" about their hidden HDD partitions with the CMOS Setup program.
I "fondly" remember upgrading my Compaq SLT 286 laptop to 4 MB of RAM by installing those proprietary SIMMs!
It worked, but was quite some work. Windows 3.1x ran nicely on it afterwards, though.
I wonder what RAM it was (FPM ?) and how quick (70..120ns?)

the3dfxdude wrote on 2023-03-08, 15:16:

But yeah for the home user, 2-4mb systems were starting to become common,
although there was still one low end 386sx system I saw in there that sold base with just 1mb still then!

I know, the 386SX systems were really low-end by default; meant to replace existing 286 PCs.
Maybe they were former 286 PCs with just their motherboard swapped. 🤷

the3dfxdude wrote on 2023-03-08, 15:16:

I mean that's what alot of us (including me) were still running only 1mb in 1992.
But I still remember what a game changer having more ram allowed with windows, and a few new games...
Anything more than 4mb then was a crazy large amount for a home user.

That's understandable, from the perspective of the end user/private user.
PCs sold in shops were really made for running DOS, Norton Commander, PC-Tools and such things.
For running a single application at a time, in short. Okay, strictly speaking, DOS-(S)hell had a program switcher, but..
When a friend of our family bought his first PC in the 90s, the seller said to him "2MB! More than you will ever need!".

For programmers (aka software/hardware developers), which wrote the software that people at home used, it was a bit different, of course.
From what I learned from my father's Windows C programming books (1990-1993), the specs were like this (please fasten your seat belts):

Minimum: AT 80386 with 25 MHz, 2 or 4 MB main memory, 80 MB HDD (< 28ms access time), VGA graphics
"Optimum": AT/EISA 80386 with 33 MHz, 4 MB main memory, 120 MB HDD (< 18ms access time), Super VGA graphics
High-end (high comfort): AT/EISA 80486 with 33 or 50 MHz, 8MB of main memory or more, 200 MB SCSI or ESDI HDD (< 18ms access time), Super VGA graphics

If someone did raytracying in POVRay or used a good astronomoy software, such a setup wouldn't have been unwelcome, either. 😉
But of course, that's not exactly "normal", either. These are/were niche hobbies, at least in relation to the population or all PC users.

Since you mention 4MB (like in the article screenshot above).. It was one of those magic numbers, essentially! 😀
Okay, I do knowingly misuse the term here a bit, but it often appears in IT. 😁
4 MB was the maximum of memory for early 286 chip sets, it was the maximum amount Video RAM of popular PCI video cards (S3 Trio etc),
the basic amount of memory used by the 3dfx Voodoo card, the minimum memory needed by OS/2 for serious work, the size of the N64 memory pak etc.

When Windows 95 came along, many existing PCs apparently had just gotten their huge, expensive 4MB upgrade for running Windows (3.1) nicely. How mean! 😁
I noticed this when I got second-hand PCs for little money by the end of the 90s (as a long-time 286 user I was very curious about 386/486 PCs).
Many of these old PCs had between 4 and 12 MB installed. And Windows 95 on the HDD, probably due to the Windows 95 hype of the 90s.
Especially 486 laptops were rarely ever upgraded past 3 or 4 MB, at the time, which was sad. Okay for Windows 3.1, but not much else.
Unfortunately, Windows 95 really crawled on those PCs with just 4MB.

Funnily, Windows 3.1 in Standard-Mode, however, was requiring less memory for good performance than Windows 95 or Windows 3.1 in 386 Enhanced-Mode.
I noticed this when I replaced the old Windows for Workgroups 3.11 installation by plain old Windows 3.10.
When I forced it running in Standard-Mode (WIN /S) just for fun and due to nostalgia (long-time 286 user..), it was quicker than usual (WIN).
Pictures loaded faster in Paintbrush, Win File draw the directories more quickly etc.
Probably because virtual memory was not used and because Standard-Mode Windows had plenty of free space in those 4MBs.
With SmartDrive loaded, the 32-Bit HDD driver enabled etc., 386 Enhanced-Mode Windows could have had made up for it, performance wise, perhaps.

Jo22 wrote on 2023-03-08, 23:30:

In the 1991-1994 time frame, approximately, OS/2 was considered a real competitor over here to Windows/DOS. In that time, both t […]
Show full quote

In the 1991-1994 time frame, approximately, OS/2 was considered a real competitor over here to Windows/DOS.
In that time, both the good old memories of OS/2 1.3 and the latest news about the great new 32-Bit support were still fresh.
To all the people who struggled with DOS and Windows 3.0, OS/2 was seen as a way out of that mess.

That's why OS/2 was being advertised as "Migrationsplattform" (migration platform) in various PC magazines around the time.
Thanks to the emulation/virtualization features of OS/2's DOS VMs, many things worked that previously failed.

For example, it was possible to give DOS-based compilers and other command line utilities up to 736 KB of conventional memory, on any PC!
All it needed was a simple mouse click in the options of the DOS support (it was done by restricting video modes to CGA, essentially).
And such a setting could be used both globally and application-specific (PIF files).

OS/2 was the DOS killer! I do remember that was a thing that was advertised, as that OS/2 was meant to break through the 640k barrier. I mean okay, breaking through 640k was already possible, but with OS/2, you get a built in switch for it! No extra software needed.

Too bad I no longer have a 286 or 386 to try this. I guess I can run OS/2 on my 486 to get 736k for DOS programs. Yes, lot's of memory and only CGA on a 486 for applications... How is OS/2's hercules mode support? And can you really run presentation manager with these cards?
I guess this is an overlooked option because I haven't seen people talking about OS/2 as a super-DOS with lots of conventional memory in recent times. Maybe it could be a good dev system for debugging old DOS software, kind of suggested as why to have more DOS mem in this video by trixter.
https://youtu.be/Xcc_D7q9bQs

Maybe another thing to test and see how well it really works??

I've been reading the findings here for the past couple days and am piqued. Years ago I had a fascination with OS/2 and until recently seriously considered acquiring a copy to run on my own retro PC. While I have abandoned taking that step I still enjoy exploring the subject.

Growing up I was focused exclusively on consumer DOS and Windows, none the wiser that there were alternatives. I'm a gamer, not an ITer, so those remain my focus. Having just learned about it is detail I'm also fascinated by the professional multiuser systems of the 80s, a time of transition which saw the beginning in small businesses (and branches of larger businesses) of what would become our present domain-type networking while on the corporate and government side IBM mainframe support was still a necessity. I can understand their value in a working environment but from a gaming perspective they appear to have nothing to offer until dedicated gaming servers started to appear and that's well after DOS, multiuser or otherwise, became a thing of the past.

As for OS/2 my interest is revived enough to consider making the topic of my next blog post an evaluation of its history and capability. The reason I ultimately decided not to go for it (other than copies starting to fetch collector prices) was I concluded for my retro PC ("System 20" in my signature) I could get the same basic capability with superior compatibility via Windows NT4. I'd been researching for a while so I'm well aware that before NT finally matched it in 1996 OS/2 was THE essential power user operating system, able to do everything 16-bit Windows could but better. Still, going through the hassle of trying to make OS/2 Warp work with my hardware setup, not the least of which is a high-capacity hard drive, didn't seem worth a small handful of exclusive Stardock games when it would be simpler and less costly to attain the same stability and better application support while getting to experiment with the unique limits of NT4's DirectX support.

But that's no knock on OS/2 even though I'm one of those who thinks Microsoft made a good point that the 286 should've been skipped altogether (remember: consumer, not professional, perspective) and making it work for that architecture anyway frustrated Microsoft to the point that I understand why they left as soon as the success of Windows gave them the opening. OS/2 certainly deserved better than it got but, without Microsoft, IBM was ill-equipped to sell it to the general public no matter how supreme its capabilities were for the time. As hard as it is to imagine today back then the dominance of Microsoft was not a given: Windows NT3 was enduring too similar a fate as OS/2 1.x while on the 16-bit side it was telling that even Microsoft acknowledged, through the creation of BOB, the strong market for easier Windows interfaces. Before Windows 95 came along and changed everything the pieces were in place to make OS/2 a major success; IBM failed to exploit that opportunity and they deserved to be forced out of the consumer PC market for it. I for one wouldn't have minded if that part of history turned out different but I can't complain either since what we got with 32-bit Windows is still doggone good.

BaronSFel001 wrote on 2023-03-09, 03:03:

As for OS/2 my interest is revived enough to consider making the topic of my next blog post an evaluation of its history and cap […]
Show full quote

As for OS/2 my interest is revived enough to consider making the topic of my next blog post an evaluation of its history and capability.
The reason I ultimately decided not to go for it (other than copies starting to fetch collector prices) was I concluded for my retro PC ("System 20" in my signature) I could get the same basic capability with superior compatibility via Windows NT4. I'd been researching for a while so I'm well aware that before NT finally matched it in 1996 OS/2 was THE essential power user operating system, able to do everything 16-bit Windows could but better.
Still, going through the hassle of trying to make OS/2 Warp work with my hardware setup, not the least of which is a high-capacity hard drive,
didn't seem worth a small handful of exclusive Stardock games when it would be simpler and less costly to attain the same stability
and better application support while getting to experiment with the unique limits of NT4's DirectX support.

Thanks for the comparison of these two! ^^

I also fondly remember tinkering with Windows NT 4, even though I was late to the party (when Win 98 was around).
If I'm honest, then I must confess that I've spent more time with OS/2 Warp all in all.

Here's a picture of Windows NT 4, as a tribute, so to say. :)