I was a network administrator back in the 90's ans worked on both OSes. I still believe that OS/2 was the better OS for servers, unfortunately the better OS doesn't always win. Thanks for sharing
Heck, back in '97 I bought my first *own* (not shared with the rest of the family - and paid by mommy and dada) PC (P200MMX, 32MB, Voodoo Rush (Hercules Stingray 128 3D, I believe 8MB), I also bought OS/2 WARP 4 - as since we got new PC last in '95, unsurprisingly it came with W95, and I was fed up with it in '97. Domn it was nice and stable as well. But in less than a year I had learned the sad truth. The software and games I needed, wanted or liked was pouring into Win95, and I had to deal 99% with DOS, 16sbit Windows and already existing OS/2 software - during that time I found no new software I wanted (well, besides web browsero, common utilities, etc. - but they weren't *new*, just _new versions_ oh what was already there). So saddened I took the Win95 installation CD of our previous shared PC and installed that on my new PC. My 2nd attempt to escape MS was around 2001 with Red Hat Linux 7.1, but although the configuration wizard for X Windows listed not even just Voodoo Rush chipsets in geneal, but my specific make and model (Hercules Stingray 128 3D), it didn't work correctly. My online friend tried for couple hours to help me using phone (call) and SSH connection to access my PC and talk to me, and he couldn't get it to work. In 2002 I finally upgraded again, made sure to ask for parts that were supported on Linux (although the nVidia card had only proprietary drivers - but thank god I didn't pick ATI. The same Red Hat 7.1 install worked like bacon on ice cream and fudge - and I never looked back. POINT! I originally wanted just to say that IMHO it was geeat on desktop as well. I wonder, if I had been able to keep using it, would I ever have switched to Linux? Coz that would be a shame if I would, because turns out no matter how wonderful OS/2 was for me, *Linux was the shiz!* LOL :D No, seriously I mean it was - like a system created and tailored personally for my needs; not that it always did what I wanted, no. That's not necessarily what I want. But how you could bend it it your will if you just knew enough. Ach, but enough about that, this is about OS/2. And I think I would have anyway, sooner or later - it would just have been much more enjoyable to stpend thi mean time with OS/2 going Warp speed tah, uhh, Win95 :D
@@THEPHINTAGECOLLECTOR Bad marketing ... Oh yes that's the favorite of IBM making the 5150 only a computer for businesses instead for general use with no graphics capability and an option for a graphics capability of 4 colors absolutely absurd. On the other hand the OS/2 was marketed as an operating system for businesses only, not for games, and that was a bad thing.
@@asanjuas Not only that, but many people mistook it for the operating system suitable only for the IBM PS/2 line of computers. Plus, the resource requirements for the initial releases were too advanced, even without the Presentation Manager being in for the very first release (even though it was promised). There were many things, which gave OS/2 a bad start from the get-go, and many mistakes along the way. But as you say, IBM for certain also underestimated the mass market needs, especially in later years with computers entering more and more into private homes. The PC clone makers took over, and Microsoft was there with the dated but widely available MS-DOS. It's a shame, as I do believe OS/2 had a lot of technological potential, especially from 2.x onwards it was really coming along. But NT is actually really nice as well, once you dissect the system architecture. Still, gaming for sure was something neither did well. That definitely something that DOS was lightyears ahead, not necessarily for what DOS did, but actually what DOS didn't do, like restricting direct hardware access.
I have a very soft spot in my heart for the advanced 16-bit Intel OSs. At first I was stunned that a 16-bit PM add-on for NT existed, but seeing how buggy it was I totally get it now. It must have been something a major customer demanded and MSFT did the bare minimum to get it out the door.
@@procta2343 totally right! but no problem. I have a "toy" machine that has hardware that almost everything likes. it was built explicitly for mid 90's fun. :) it's already installed and running well. 233MMX, Matrox millennium II, SB16 and 3com 10baseT. It also has SD cards setup for OS/2 v4, DOS 6.22/win311 and QNX neutrino 6. :)
So here’s a thing I have a distinct memory of reading about but I’ve never seen referenced elsewhere: that early NT would intercept I/O to a file named CONFIG.SYS (I think only from processes running in the OS/2 subsystem) and simulate it as though that file existed? So you would use an OS/2 editor to edit the (non-existent?) C:\CONFIG.SYS and it’d look like you were editing a real file with real settings representative of the running NT system. Then when the file was closed, it would translate the directives in it to native NT settings (and store them in the Registry, presumably). Anybody else recall that?
Never heard about it myself, but well possible still. It wouldn't be the first time that Microsoft claimed something during early development phases, which eventually would diminish into obscurity over time. Notable examples on such claims are "DOS is dead. Windows 95 does not require DOS any more." (briefly touched in my own Video at th-cam.com/video/FRzQwq-P3gU/w-d-xo.html, or, "Windows won't have a traditional file system any more, it's gonna be SQL-based" (claimed during Windows "Longhorn" development, see as well en.wikipedia.org/wiki/WinFS). So if anyone has further info about this claim mentioned above, I'd be very interested into learning more as well!
Yeah and AFAIK security and permissions never came to OS/2 or in one of the much later versons? HPFS didn't have support for it for example except for OS/2 servers version if I checked that correctly.
The NT OS/2 1.x subsystem was primarily to encourage previous Microsoft OS/2 1.x customers to migrate to NT instead of migrating to IBM OS/2 2.0. Microsoft did a pretty decent job too, with text mode OS/2 1.x apps (including their own OS/2 1.x dev tools) running transparently from within the NT Command Prompt. I tried running the WLO apps under Windows NT 4 via PMFILE. Despite being executed from the OS/2 side, they still load up on the Windows side. I didn't expect that. Particularly since PMFILE won't let you execute Win32 binaries. I also tried to force it from a Command Prompt (with PMSHELL running in the background) with "os2 /p clock.exe /c" but all the WLO apps just segfault.
Yeah, the so called NEWSHELL, available in 2 versions. I actually tried it back in the day because I hated Win95 instability and I also hated the old Win 3.x/NT 3.x user interface, so until NT4 came I had hoped I can use it. Unfortunately it was really too unstable for standard use back then.
Oh good a video on something totally rare and fringe! I'd always heard about the PM for NT but after years of searching when I finally found it, I was surprised about the 3.51 / 4.0 thing as well. I have to think there was a 3.1/3.5 versions as well. The WinOS/2 drivers were very involved to display Windows 3.0/3.1 on OS/2, so I'd have imagined it would have been just as difficult for PM on NT, and Microsoft just wasn't interested. Good stuff tho!
Yeah, I can't tell if any releases for NT earlier than 3.51 existed or not, as there is simply no source pointing to it. It's maybe worthwile to also try if the NT 4.0 version would run on Windows 2000, as the latter technically speaking still included the OS/2 subsystem. It was only with Windows XP that Microsoft expired both the OS/2 and the POSIX subsystems. As for how few love Microsoft gave to Presentation Manager, I think it was really just the bare minimum, to fulfill a requirement for winning over some customers at the time. And by the time Windows 2000 and later XP came around, OS/2 was more or less out of the (mass) market, hence Microsoft didn't have any motivation in furtherly venturing on it.
I wonder on what software and hardware microsoft was developing the FIRST windows NT (writing code, compiling, debugging, testing, etc). I wonder because, as is well known, microsoft is locked in its bubble and pretends that other operating systems do not exist. This "closed" mentality was more evident in previous decades.
They initially developed NT on OS/2 using an emulator for the i860 processor by Intel. Apparently, the tooling Intel provided for the i860 was terrible and i860 wasn’t terribly popular either so they quickly moved on. Bill Gates then wanted the NT team to focus on x86/PC architecture so a lot of stuff was done with 386/486s but Cutler insisted to also work on a MIPS port in parallel. Higher-ups at Microsoft seemingly agreed because they later published the Jazz architecture which is a weird mixture of classical PC standards like ISA buses with a MIPS processor which was able to run NT.
@@klausschmidt982 There are actually 2 survivor samples of the development boards Microsoft produced for the i860 NT early development. Them were donated by MS to the Smithsonian institution. Search for "Microsoft Windows NT Development Board, PCR1". Microsoft Jazz architecture derives from this early work, so I guess they had some hardware to test on besides emulation. But like you said, other dev tools Intel provided, along the infamous i860 emulator, weren't enough polished. This together with the processor and chipset chip shipping delays, made MS finally ditch the i860 and find a new RISC platform to continue NT development on Cutler's request (to keep NT code hardware agnostic, in contraposition with Gates views, as you said, and which paid in the long term (all the delivered ports including the current ARM64 done up to the date).
I suspect at least some of NT's early i386 development was done on the Microsoft pre-release version of OS/2 2.0 that Microsoft was working on prior to the split with IBM. OS/2 2.0 was already well advanced in mid-1990, and a very reliable OS, but the split delayed IBM's eventual release to the public by about two years. It wouldn't have been too difficult for Microsoft to modify their existing CL386.EXE OS/2 compiler to cross-compile to NT, at least until NT became self-hosting. Interestingly it was only about nine months between the split from IBM and when the first preview builds of Windows NT 3.1 were released to testers. Microsoft didn't waste any time. At least some of OS/2 1.x's development was likely done on Xenix.
Interesting that it is showing NT Version 4.0 (Build 1057) in 14:12! Do you know hat happened here? After using the last known good config it's back to 3.51.
Good spotting! I had as well tried the Windows NT 3.51 "Shell Technology Preview" to see what goes, and what not. The Shell Update replaces part of the Kernel on NT 3.51, making it report as NT 4.0. This is furtherly explained in my other video at th-cam.com/video/fAPWeeHxHrg/w-d-xo.htmlsi=cXHQgfgp1pulsfIH It had however nothing to do with the observed misbehaviour of Word for OS/2 Presentation Manager, which caused a crash either with and without the Shell Preview applied. Obviously I have mixed-up the footage and picked the wrong bluescreen, though as mentioned, it doesn't matter. Result was the same for me on either setup. I skipped the entirely shell preview topic and branched it into a separate video, as it would have added to much fuzz for this one here.
@@THEPHINTAGECOLLECTOR thanks a lot for the detailed response and btw. thanks for all those really interesting high-professional Videos! 😊 I am also very interested in all those vintage Microsoft related topics, I cannot even really tell why though. 😄
I worked with and supported the earlier OS/2 and NT packages. IBM insisted that the OS/2 HAD to work properly on a 286 machine, since they told big customers "Buy the IBM AT, and OS2 will be your operating system in a couple of years." -- Only problem was "how do we get from kernel mode to user mode?" issue. The only way to do it was have a triple fault and an entry for reset (in the dispatch table) to do that. There was no instruction to do it like there is in the 386. MS wrote NT for the 386 and newer processors... IBM demanded it work in a 286.. So they got forced to do the slow triple fault to go back to user mode from protected (kernel) mode... IBM figured that out when they took over OS2 development.... Warp was 386 happy... and didn't have the issues with a 16 meg of ram limit. (Adding the TCP/IP stack on OS2 1.3 got you down to less than 50k free, which was an issue for people wanting to run web servers....
Nice insights. I may be wrong, but wasn‘t the triple fault & reset procedure rather for switching the CPU from protected mode back to real mode? Kernel mode (ring 0) and user node (ring 3) in ny understabdibg is about privileges, but not the operations lode of thr CPU. The OS/2 kernel certainly ran in protected mode on ring 0, though not all user mode applications would necessarily run on ring 3 in real mode.
@@THEPHINTAGECOLLECTOR One of the MS developers has a tube thing about triple faults.. apparently they "fell into" that as a way to get back to normal user mode after tweaking the dispatch table... Took a bunch of microseconds. I don't remember the exact timing, but heard at one point it could be a Millisecond or so... A huge performance eater.. The instruction in the 386 did it FAR more efficiently. (Apparently Intel never thought about leaving kernel mode back in the dark ages.) The HUGE problem was dropping in the IP stack, where your 16 meg (max) on the AT would croak... Loading up an AT with nothing but Netbeui was a waste, but at least you have left over ram....
I feel like IBM wanted OS/2 1.x to run on 286 machines only because most of their PS/2 line had 286 CPUs. Meanwhile the PC market was already moving to 386 PCs. It's kind of ridiculous that regular home PC users were made to wait until 2001 for the release of Windows XP to have a reliable true 32-bit OS, when a basic version of 32-bit OS/2 could've been made available to home users with 386 machines in 1990, or even earlier if IBM hadn't insisted on OS/2 1.x being 16-bit!
@@zoomosis The problem was that IBM **promised** OS/2 to thousands of folk that stocked up on the AT system. That introduced a huge issue for the sales folk... The memory space at 16 megabytes was a problem when you added in the TCP/IP stack. If you system was a 16 megabyte sized system, you would have a limited blob of memory left for your applications. As in no room for really big applications like CAD and the like. Throw some big drivers for plotters and printers, a networking stack. and limited optimization, and OS/2 would basically not be terribly functional. That's why WARP wanted a 386 as a starting place. MS told IBM you needed a 386 to handle the switch from user mode to protected mode and back again without going through the reset interrupt, and the additional address space there to let it grow...
@@Billblom Coincidentally I was tinkering with OS/2 1.0 earlier today in a virtual machine and was surprised to see that it wouldn't install without at least 2 MB of RAM. In my mind I always thought 1 MB was enough. I know prices came down quickly but 2 MB would've been very expensive in 1988. Later I was able to get it to run in 1 MB but only with PROTECTONLY=YES in config.sys, disabling the DOS Prompt. I was also reminded of just how much version 1.0 felt like a beta, or even alpha version. Like it was released before it was ready. The "half an operating system" joke really rings true for that version since a lot of things you'd expect to be there were just missing in 1.0, like a native OS/2 text editor. Maybe 1.0 was released prematurely to satisfy a promise IBM made to some of its customers.
Correct winNT IS os/2 userland base WITH VMS kernel on top system to run multiple systems like posix, win32, os2 apps all that is completely different from ibm os2 is only limited to dos and win16 to win32 in later versions
@MrEd-qg8td I'd love to get hold of the 1.0 release for NT 3.1. If you don't mind sending it to me? Or even better, post it to winworldpc.com/product/microsoft-os2-presen/for-nt-40 If the 3.1 version still exists, it's definitely worth preserving!
@@WildwoodSubRailfan That seems to be premature judgement based on a single sentence, without actually watching the entire video, obviously. Windows NT began as a rewrite of the OS/2 operating system. And despite it eventually becoming something entirely different, to me that is „emerging from OS/2“ from a history perspective. That does not necessarily imply code transfer, and that‘s nowhere claimed in this video. On the contrary, the origins of NT, going back to VMS concepts, are mentioned as part of this video.
Hi @hypercube33, Thanks for commenting. Can you please elaborate what exactly you mean with the slide card text, is it the chapter mark titles? If so, what's wrong about them for making it hard to read? Text size, duration, something else? Will be happy to adapt, though I didn't quiet get yet the issue. Thanks!
I was a network administrator back in the 90's ans worked on both OSes. I still believe that OS/2 was the better OS for servers, unfortunately the better OS doesn't always win. Thanks for sharing
You‘re welcome 🙏
Heck, back in '97 I bought my first *own* (not shared with the rest of the family - and paid by mommy and dada) PC (P200MMX, 32MB, Voodoo Rush (Hercules Stingray 128 3D, I believe 8MB), I also bought OS/2 WARP 4 - as since we got new PC last in '95, unsurprisingly it came with W95, and I was fed up with it in '97.
Domn it was nice and stable as well. But in less than a year I had learned the sad truth. The software and games I needed, wanted or liked was pouring into Win95, and I had to deal 99% with DOS, 16sbit Windows and already existing OS/2 software - during that time I found no new software I wanted (well, besides web browsero, common utilities, etc. - but they weren't *new*, just _new versions_ oh what was already there).
So saddened I took the Win95 installation CD of our previous shared PC and installed that on my new PC. My 2nd attempt to escape MS was around 2001 with Red Hat Linux 7.1, but although the configuration wizard for X Windows listed not even just Voodoo Rush chipsets in geneal, but my specific make and model (Hercules Stingray 128 3D), it didn't work correctly. My online friend tried for couple hours to help me using phone (call) and SSH connection to access my PC and talk to me, and he couldn't get it to work.
In 2002 I finally upgraded again, made sure to ask for parts that were supported on Linux (although the nVidia card had only proprietary drivers - but thank god I didn't pick ATI. The same Red Hat 7.1 install worked like bacon on ice cream and fudge - and I never looked back.
POINT! I originally wanted just to say that IMHO it was geeat on desktop as well.
I wonder, if I had been able to keep using it, would I ever have switched to Linux? Coz that would be a shame if I would, because turns out no matter how wonderful OS/2 was for me, *Linux was the shiz!* LOL :D No, seriously I mean it was - like a system created and tailored personally for my needs; not that it always did what I wanted, no. That's not necessarily what I want. But how you could bend it it your will if you just knew enough. Ach, but enough about that, this is about OS/2. And I think I would have anyway, sooner or later - it would just have been much more enjoyable to stpend thi mean time with OS/2 going Warp speed tah, uhh, Win95 :D
Can you elaborate what made OS/2 in your opinion better for servers?
@Lofote true multi-tasking. We use Windows today because IBM didn't know what they had.
@@RudysRetroIntel But Windows NT always also had true multitasking, or what exactly would OS/2 do here better?
The personalities are better on IBM OS/2 than on Windows NT.
OS/2 also has done great work when it comes to cross-OS integration and compatibility, especially in releases 2.x and onwards.
@@THEPHINTAGECOLLECTOR of course this was when Microsoft was not coding OS/2.
@@asanjuas sure thing, IBM on the helm of 2.x+.
If it weren‘t for their bad marketing, who knows…? ;)
@@THEPHINTAGECOLLECTOR Bad marketing ... Oh yes that's the favorite of IBM making the 5150 only a computer for businesses instead for general use with no graphics capability and an option for a graphics capability of 4 colors absolutely absurd. On the other hand the OS/2 was marketed as an operating system for businesses only, not for games, and that was a bad thing.
@@asanjuas Not only that, but many people mistook it for the operating system suitable only for the IBM PS/2 line of computers. Plus, the resource requirements for the initial releases were too advanced, even without the Presentation Manager being in for the very first release (even though it was promised).
There were many things, which gave OS/2 a bad start from the get-go, and many mistakes along the way.
But as you say, IBM for certain also underestimated the mass market needs, especially in later years with computers entering more and more into private homes.
The PC clone makers took over, and Microsoft was there with the dated but widely available MS-DOS.
It's a shame, as I do believe OS/2 had a lot of technological potential, especially from 2.x onwards it was really coming along.
But NT is actually really nice as well, once you dissect the system architecture.
Still, gaming for sure was something neither did well. That definitely something that DOS was lightyears ahead, not necessarily for what DOS did, but actually what DOS didn't do, like restricting direct hardware access.
Nice video! I love this entire era of OSes, and I find OS/2 and early Windows NT particularly fascinating.
I have a very soft spot in my heart for the advanced 16-bit Intel OSs. At first I was stunned that a 16-bit PM add-on for NT existed, but seeing how buggy it was I totally get it now. It must have been something a major customer demanded and MSFT did the bare minimum to get it out the door.
Another good video.
Now I want to install NT 3.51 on one of my play machines.
-former long term OS/2 user.
Go for it 🙂
just beware that there isn't as much driver support for it compared to windows 4.0.
@@procta2343 totally right! but no problem. I have a "toy" machine that has hardware that almost everything likes. it was built explicitly for mid 90's fun. :) it's already installed and running well. 233MMX, Matrox millennium II, SB16 and 3com 10baseT.
It also has SD cards setup for OS/2 v4, DOS 6.22/win311 and QNX neutrino 6. :)
So here’s a thing I have a distinct memory of reading about but I’ve never seen referenced elsewhere: that early NT would intercept I/O to a file named CONFIG.SYS (I think only from processes running in the OS/2 subsystem) and simulate it as though that file existed? So you would use an OS/2 editor to edit the (non-existent?) C:\CONFIG.SYS and it’d look like you were editing a real file with real settings representative of the running NT system. Then when the file was closed, it would translate the directives in it to native NT settings (and store them in the Registry, presumably).
Anybody else recall that?
Never heard about it myself, but well possible still.
It wouldn't be the first time that Microsoft claimed something during early development phases, which eventually would diminish into obscurity over time.
Notable examples on such claims are "DOS is dead. Windows 95 does not require DOS any more." (briefly touched in my own Video at th-cam.com/video/FRzQwq-P3gU/w-d-xo.html,
or, "Windows won't have a traditional file system any more, it's gonna be SQL-based" (claimed during Windows "Longhorn" development, see as well en.wikipedia.org/wiki/WinFS).
So if anyone has further info about this claim mentioned above, I'd be very interested into learning more as well!
NT was multi user from the out set OS/2 was at all. A totally different OS.
Yeah and AFAIK security and permissions never came to OS/2 or in one of the much later versons? HPFS didn't have support for it for example except for OS/2 servers version if I checked that correctly.
The NT OS/2 1.x subsystem was primarily to encourage previous Microsoft OS/2 1.x customers to migrate to NT instead of migrating to IBM OS/2 2.0.
Microsoft did a pretty decent job too, with text mode OS/2 1.x apps (including their own OS/2 1.x dev tools) running transparently from within the NT Command Prompt.
I tried running the WLO apps under Windows NT 4 via PMFILE. Despite being executed from the OS/2 side, they still load up on the Windows side. I didn't expect that. Particularly since PMFILE won't let you execute Win32 binaries.
I also tried to force it from a Command Prompt (with PMSHELL running in the background) with "os2 /p clock.exe /c" but all the WLO apps just segfault.
Very interesting, didn't know that existed for GUI OS/2 applications. Do you plan to bring on the NT 4.0 version as well?
@@Lofote The NT4 version doesn‘t look much different, there‘s no value down this road.
NT3.51 also has a "Desktop Preview" which gives it the 95/NT4 style desktop. That may be worth a look if you can get hold of it
Well, actually I tried it during the making of this video. And I screwed NT the fifth time ... ;-)
Yeah, the so called NEWSHELL, available in 2 versions.
I actually tried it back in the day because I hated Win95 instability and I also hated the old Win 3.x/NT 3.x user interface, so until NT4 came I had hoped I can use it. Unfortunately it was really too unstable for standard use back then.
@OtterlyInsane @Lofote Unless you haven't seen it already, I did the video about "New Shell" in the meantime: th-cam.com/video/fAPWeeHxHrg/w-d-xo.html
Oh good a video on something totally rare and fringe! I'd always heard about the PM for NT but after years of searching when I finally found it, I was surprised about the 3.51 / 4.0 thing as well. I have to think there was a 3.1/3.5 versions as well. The WinOS/2 drivers were very involved to display Windows 3.0/3.1 on OS/2, so I'd have imagined it would have been just as difficult for PM on NT, and Microsoft just wasn't interested. Good stuff tho!
Yeah, I can't tell if any releases for NT earlier than 3.51 existed or not, as there is simply no source pointing to it.
It's maybe worthwile to also try if the NT 4.0 version would run on Windows 2000, as the latter technically speaking still included the OS/2 subsystem.
It was only with Windows XP that Microsoft expired both the OS/2 and the POSIX subsystems.
As for how few love Microsoft gave to Presentation Manager, I think it was really just the bare minimum, to fulfill a requirement for winning over some customers at the time.
And by the time Windows 2000 and later XP came around, OS/2 was more or less out of the (mass) market, hence Microsoft didn't have any motivation in furtherly venturing on it.
Yes there was an OS/2 PM plug in for NT 3.1. And I do have it. Got it straight from M$
@@MrEd-qg8td No chance of ever showing it or sharing it with the world?
@@MrEd-qg8td would you share it with us?
I am reading a book about the history of NT and this video is very good! Nice work!
Good info, this is my favourite OS period
I wonder on what software and hardware microsoft was developing the FIRST windows NT (writing code, compiling, debugging, testing, etc).
I wonder because, as is well known, microsoft is locked in its bubble and pretends that other operating systems do not exist.
This "closed" mentality was more evident in previous decades.
According to some sources like Wikipedia, they used a non-x86 platform, initially the Intel i860 (by means of an emulator), and later the MIPS R3000.
They initially developed NT on OS/2 using an emulator for the i860 processor by Intel. Apparently, the tooling Intel provided for the i860 was terrible and i860 wasn’t terribly popular either so they quickly moved on. Bill Gates then wanted the NT team to focus on x86/PC architecture so a lot of stuff was done with 386/486s but Cutler insisted to also work on a MIPS port in parallel. Higher-ups at Microsoft seemingly agreed because they later published the Jazz architecture which is a weird mixture of classical PC standards like ISA buses with a MIPS processor which was able to run NT.
@@klausschmidt982 There are actually 2 survivor samples of the development boards Microsoft produced for the i860 NT early development. Them were donated by MS to the Smithsonian institution. Search for "Microsoft Windows NT Development Board, PCR1".
Microsoft Jazz architecture derives from this early work, so I guess they had some hardware to test on besides emulation. But like you said, other dev tools Intel provided, along the infamous i860 emulator, weren't enough polished. This together with the processor and chipset chip shipping delays, made MS finally ditch the i860 and find a new RISC platform to continue NT development on Cutler's request (to keep NT code hardware agnostic, in contraposition with Gates views, as you said, and which paid in the long term (all the delivered ports including the current ARM64 done up to the date).
I suspect at least some of NT's early i386 development was done on the Microsoft pre-release version of OS/2 2.0 that Microsoft was working on prior to the split with IBM. OS/2 2.0 was already well advanced in mid-1990, and a very reliable OS, but the split delayed IBM's eventual release to the public by about two years.
It wouldn't have been too difficult for Microsoft to modify their existing CL386.EXE OS/2 compiler to cross-compile to NT, at least until NT became self-hosting.
Interestingly it was only about nine months between the split from IBM and when the first preview builds of Windows NT 3.1 were released to testers. Microsoft didn't waste any time.
At least some of OS/2 1.x's development was likely done on Xenix.
Interesting that it is showing NT Version 4.0 (Build 1057) in 14:12! Do you know hat happened here? After using the last known good config it's back to 3.51.
Good spotting! I had as well tried the Windows NT 3.51 "Shell Technology Preview" to see what goes, and what not.
The Shell Update replaces part of the Kernel on NT 3.51, making it report as NT 4.0.
This is furtherly explained in my other video at th-cam.com/video/fAPWeeHxHrg/w-d-xo.htmlsi=cXHQgfgp1pulsfIH
It had however nothing to do with the observed misbehaviour of Word for OS/2 Presentation Manager,
which caused a crash either with and without the Shell Preview applied.
Obviously I have mixed-up the footage and picked the wrong bluescreen, though as mentioned, it doesn't matter.
Result was the same for me on either setup.
I skipped the entirely shell preview topic and branched it into a separate video, as it would have added to much fuzz for this one here.
@@THEPHINTAGECOLLECTOR thanks a lot for the detailed response and btw. thanks for all those really interesting high-professional Videos! 😊 I am also very interested in all those vintage Microsoft related topics, I cannot even really tell why though. 😄
@@Jk-wt2kd you‘re welcome!
I worked with and supported the earlier OS/2 and NT packages. IBM insisted that the OS/2 HAD to work properly on a 286 machine, since they told big customers "Buy the IBM AT, and OS2 will be your operating system in a couple of years." -- Only problem was "how do we get from kernel mode to user mode?" issue. The only way to do it was have a triple fault and an entry for reset (in the dispatch table) to do that. There was no instruction to do it like there is in the 386. MS wrote NT for the 386 and newer processors... IBM demanded it work in a 286.. So they got forced to do the slow triple fault to go back to user mode from protected (kernel) mode... IBM figured that out when they took over OS2 development.... Warp was 386 happy... and didn't have the issues with a 16 meg of ram limit. (Adding the TCP/IP stack on OS2 1.3 got you down to less than 50k free, which was an issue for people wanting to run web servers....
Nice insights.
I may be wrong, but wasn‘t the triple fault & reset procedure rather for switching the CPU from protected mode back to real mode?
Kernel mode (ring 0) and user node (ring 3) in ny understabdibg is about privileges, but not the operations lode of thr CPU.
The OS/2 kernel certainly ran in protected mode on ring 0, though not all user mode applications would necessarily run on ring 3 in real mode.
@@THEPHINTAGECOLLECTOR One of the MS developers has a tube thing about triple faults.. apparently they "fell into" that as a way to get back to normal user mode after tweaking the dispatch table... Took a bunch of microseconds. I don't remember the exact timing, but heard at one point it could be a Millisecond or so... A huge performance eater.. The instruction in the 386 did it FAR more efficiently. (Apparently Intel never thought about leaving kernel mode back in the dark ages.) The HUGE problem was dropping in the IP stack, where your 16 meg (max) on the AT would croak... Loading up an AT with nothing but Netbeui was a waste, but at least you have left over ram....
I feel like IBM wanted OS/2 1.x to run on 286 machines only because most of their PS/2 line had 286 CPUs. Meanwhile the PC market was already moving to 386 PCs.
It's kind of ridiculous that regular home PC users were made to wait until 2001 for the release of Windows XP to have a reliable true 32-bit OS, when a basic version of 32-bit OS/2 could've been made available to home users with 386 machines in 1990, or even earlier if IBM hadn't insisted on OS/2 1.x being 16-bit!
@@zoomosis The problem was that IBM **promised** OS/2 to thousands of folk that stocked up on the AT system. That introduced a huge issue for the sales folk... The memory space at 16 megabytes was a problem when you added in the TCP/IP stack. If you system was a 16 megabyte sized system, you would have a limited blob of memory left for your applications. As in no room for really big applications like CAD and the like. Throw some big drivers for plotters and printers, a networking stack. and limited optimization, and OS/2 would basically not be terribly functional. That's why WARP wanted a 386 as a starting place. MS told IBM you needed a 386 to handle the switch from user mode to protected mode and back again without going through the reset interrupt, and the additional address space there to let it grow...
@@Billblom Coincidentally I was tinkering with OS/2 1.0 earlier today in a virtual machine and was surprised to see that it wouldn't install without at least 2 MB of RAM. In my mind I always thought 1 MB was enough. I know prices came down quickly but 2 MB would've been very expensive in 1988.
Later I was able to get it to run in 1 MB but only with PROTECTONLY=YES in config.sys, disabling the DOS Prompt.
I was also reminded of just how much version 1.0 felt like a beta, or even alpha version. Like it was released before it was ready. The "half an operating system" joke really rings true for that version since a lot of things you'd expect to be there were just missing in 1.0, like a native OS/2 text editor.
Maybe 1.0 was released prematurely to satisfy a promise IBM made to some of its customers.
Correct winNT IS os/2 userland base WITH VMS kernel on top system to run multiple systems like posix, win32, os2 apps all that is completely different from ibm os2 is only limited to dos and win16 to win32 in later versions
I have the PM plug in for windows NT 3.1 the 1.0 version of it. And also have the later version for NT 3.51/4.0
@MrEd-qg8td I'd love to get hold of the 1.0 release for NT 3.1. If you don't mind sending it to me? Or even better, post it to winworldpc.com/product/microsoft-os2-presen/for-nt-40
If the 3.1 version still exists, it's definitely worth preserving!
@@THEPHINTAGECOLLECTOR I sent the disk images to you last night
@@MrEd-qg8td Yeah, will have a look at it. Awesome! Thanks 🙏
@@THEPHINTAGECOLLECTOR Can you upload the NT 3.1 version to Winworld?
I wonder which user the programs run under PM system, when it is running as a service. I hope not SYSTEM like the service itself ;)
NT didn't emerge from OS/2 at all. Dave Cutler (the guy who created NT) hated OS/2 and has spoken of it fairly derogatorily.
@@WildwoodSubRailfan That seems to be premature judgement based on a single sentence, without actually watching the entire video, obviously.
Windows NT began as a rewrite of the OS/2 operating system. And despite it eventually becoming something entirely different, to me that is „emerging from OS/2“ from a history perspective.
That does not necessarily imply code transfer, and that‘s nowhere claimed in this video.
On the contrary, the origins of NT, going back to VMS concepts, are mentioned as part of this video.
capshuns trash your slide card text. perhaps slide it on up to the center of the screen so i can read both. THX
Hi @hypercube33,
Thanks for commenting. Can you please elaborate what exactly you mean with the slide card text, is it the chapter mark titles?
If so, what's wrong about them for making it hard to read? Text size, duration, something else?
Will be happy to adapt, though I didn't quiet get yet the issue.
Thanks!
Awesome video.
Thank you! Glad you liked it ^^
HPFS was made by MS not IBM.
Indeed, and noone said different. HPFS was developed by Gordon Letwin at Microsoft while working on OS/2.
Cane we get a "5 Hours of Install Progress Bars to Sleep to" video?
My new advent series features custom progress bars at the beginning, but 5 hours!? ^^
th-cam.com/video/Xe3N_j44SNs/w-d-xo.html
Comparing Windows 3.1 to NT is like comparing a bike to a Rolls Royce.
It is technicaly, but visually they are close.
@@Lofote That was probably one of the very best moves Microsoft made.
@@judewestburner defintely
Good Luck Finding Program: Windows in 340 Million Million Million Million Terabytes!