What build of Windows 95 dropped 286 support?

[ATeamInc]

Considering Windows 3.10 can run on 286 processors, did the early Chicago builds support 286 (if any did), and starting from what build did it drop support for it?

[LangsamSpieler]

I think from 81, because on 73g are 1 MB RAM and a 286 CPU enough for runnig but i'm not sure

[TheCollector1988]

not even 73g or earlier can run in a 286...

[jimmsta]

Seeing as Chicago utilized the VMM extension on the 386 - no builds supported the 286.

[louisw3]

Windows 3.11 dropped the 286, as they had written a very aggressive protected mode disk access driver at this point on the way of moving into Chicago.

Although it's limited to the older sub 504MB IDE disks.

https://www.rigacci.org/docs/biblio/onl ... n32bda.htm

[Overdoze]

To be precise, Windows for Workgroups 3.11 was the first to completely ditch "Standard mode" for the 286. If you try to run win /s or win /2, Windows will inform you of this change.

I think Chicago/Cougar itself was forked from WfW 3.1, and then the improvements from 3.11 were gradually merged in during 1993. But I doubt it was ever intended to support the 286 anyway, since that was a pretty old and limited processor by then. Even if design considerations are ignored, the performance would be abysmally poor.

[louisw3]

To be precise, Windows for Workgroups 3.11 was the first to completely ditch "Standard mode" for the 286. If you try to run win /s or win /2, Windows will inform you of this change.

I think Chicago/Cougar itself was forked from WfW 3.1, and then the improvements from 3.11 were gradually merged in during 1993. But I doubt it was ever intended to support the 286 anyway, since that was a pretty old and limited processor by then. Even if design considerations are ignored, the performance would be abysmally poor.

The next generation of desktop was always going to be 32bit. Originally the OS/2 2.0 that fizzled up then it became Windows 3.0 expanded and retooled.

The days of the 286 had already sunset.

[ATeamInc]

To be precise, Windows for Workgroups 3.11 was the first to completely ditch "Standard mode" for the 286. If you try to run win /s or win /2, Windows will inform you of this change.

I think Chicago/Cougar itself was forked from WfW 3.1, and then the improvements from 3.11 were gradually merged in during 1993. But I doubt it was ever intended to support the 286 anyway, since that was a pretty old and limited processor by then. Even if design considerations are ignored, the performance would be abysmally poor.

To prove all this, I'll try some of the earliest Chicago builds to check whether the 286 is supported. Also, I found this: http://drivers.downloadatoz.com/vendor_ ... indows-95/

[Overdoze]

Well, be my guest. Fire up an emulator emulating a 286 machine and try to get Chicago up and running on it.

[oscareczek]

The site you linked (very random, but okay) has only BIOS and something under the mysterious name "Misc" "for" Windows 9x. It doesn't prove anything, just some bot found the files.

[ATeamInc]

And I guess I could call my tests sort-of a FAIL:
58s:
Does not support upgrading from windows 3.1. When using setup31.exe from the dos prompt, it first checks for setup.inf (which I copied from windows 3.1 /system directory), complains about networking, starts copying files, requests Windows for Workgroups 3.1 install disks (got around that by using the C:\Windows and C:\Windows\SYSTEM directories), and last errors out when trying to start complaining about vga.drv. Replacing the vga.drv from the Windows 3.1 install disk and trying to start windows (by typing win) simply exits Windows to a blue-ish command prompt.
73g:
Attempting to start from a dos 5 bootdisk complains about HIMEM.SYS being missing. Starting the setup with dos 6.22 (and HIMEM) does this: and attempting to start setup again results in lack of memory (for some reason).
Well, since it complains about missing files, I guess it can start in 286 standard mode but lacks the files to do so (hence what it did with 58s trying to copy files from Windows disks).
EDIT: I found this while trying setup31 on 73g:

Was the 286 intended to be supported (hence the 286 ega modes) or was it leftovers from Windows 3.1?

[SoftPCMuseum]

Also, I found this: http://drivers.downloadatoz.com/vendor_ ... indows-95/

This would be because of the fact that 286-based machines (such as the one that you linked to) can easily be upgraded to a 386 or higher CPU in nearly every case using a CPU upgrade board (such as the Inboard/386 AT) or adapter, meaning that a 286 ROM BIOS naturally was still supported under Windows 95, even if the machine itself required a CPU upgrade in order for the operating system itself to work.

[ATeamInc]

Well, I got progress with 73g!

The way I did it was that I started 73g in setup31 mode, allowed it to copy from the Windows 3.1 disks, but first I replaced gdi.exe from the chicago setup directory with a copy from the Windows 3.1 disk #1.

[LangsamSpieler]

I think the result will be a Windows 3.1

[ATeamInc]

That is what I thought too, so I decided to copy all the core-286 files onto the install directory and try running setup. But apparently I am still missing some files which I have no idea about. If anyone knows all the core-286 files needed to start Windows, feel free to leave them in.

[LangsamSpieler]

Let's hope the best thing for Chicago

[louisw3]

The whole point of Chicago was a consumer 32bit (ish) desktop OS.

Code: Select all

CLOCK32.EXE:  PE executable for MS Windows (GUI) Intel 80386 32-bit
NOTE32.EXE:   PE executable for MS Windows (GUI) Intel 80386 32-bit

There isn't much on 58s in the way of 32bit stuff, but they do run on Windows 10 x64.

The 286 was never the target of Chicago.

[ATeamInc]

I think the result will be a Windows 3.1

Well, the setup31 method returned a corrupted mixed file config of Windows 3.1 and Chicago 73g. What louisw3 said is entirely correct, Chicago was always meant to be 32 bit.

(You can see the corrupted program manager missing all entries, I had to create a new group and call it main just to get the Main group to appear. Also, the Unknown user is another result.)

[ATeamInc]

To add upon the last statement, upon finding the right file to add to allow 286 standard setup mode (that being krnl286.exe in the \mini directory), Chicago fails to start and exits due to a processor check which refuses to install chicago on a 286.

[retr0pimp]

Shifting gears a little, would it be possible to run any Chicago builds on an AT that has one of those InBoard accelerators permitting a 386 (or 486DLC, I suppose) to run on it? I imagine if the CPU identified itself as so, and since wikipedia alleges that the InBoard AT models would permit 3.1 to run, surely an early Chicago build would boot.

This would be one potential route around the 286 limitation, albeit expensive if you don't already have one of those boards.

[SoftPCMuseum]

Shifting gears a little, would it be possible to run any Chicago builds on an AT that has one of those InBoard accelerators permitting a 386 (or 486DLC, I suppose) to run on it? I imagine if the CPU identified itself as so, and since wikipedia alleges that the InBoard AT models would permit 3.1 to run, surely an early Chicago build would boot.

This would be one potential route around the 286 limitation, albeit expensive if you don't already have one of those boards.

Yes, but that would still be running it on an 80386-based machine or higher. The question here was whether it could be run on an 80286 CPU. The answer to that is that Windows 95 is a 32-bit operating system which requires at least an 80386 CPU or higher.

[os2fan2]

Win 3,11 runs in standard mode. It's wfw 3,11 that doesn't.