• Please review our updated Terms and Rules here

Windows 3.1 w/ Intel Inboard 386/PC - VxD Issue

Another bit of information just found on here:

I hadn’t seen this before, I know this is for the 3.0 version but maybe relevant?
2004-2007 does seem quite late for this though?

betawiki.net

Windows 3.00a - BetaWiki

betawiki.net betawiki.net

Due to some changes introduced with bug fixes, Microsoft Virtual PC 2004/2007 users have to include the line EMMExclude=A000-CBFF in SYSTEM.INI's [386Enh] section.

​

 
harrison.frazier said:
It was suggested that I post here- I have no clue what I did 2 years ago to get it working but I am using device=*vpicd. I definitely followed what a lot of you have already tried and maybe I got lucky. Maybe I replaced some files from 3.0 to 3.1. I know I did the upgrade on top of 3.0. I have linked an image of my 2GB CF image, hopefully this helps you all! I have an ATI Mach8 video card, you may have to re-run C:\windows\setup and remove them first.

https://imgur.com/wHNDIDK

Hardware: XT, original bios
Cards (L to R)
External CD-ROM (looks like scsi)
3COM Etherlink II TP
ATI Mach8
Inboard (With FPU and Cyrix Cx486DRx2, and my custom 4MB piggyback card)
XTIDE
Microsoft Bus Mouse

2GB CF Card Image Here

https://imgur.com/TTU0jWc
https://imgur.com/7CR0rOn
Click to expand...
hello - my first attempt , did not work.
Took your windows directory onto my SCSI-drive. (my os is still PCDOS 7.0) Probably hardware differences, i can not boot without device=ibvpicd.386.
Perhaps later IBM pc xt are easier getting to run accelerated software ? then the older pc 5150 ???
/cimonvg
 
Cimonvg said:
hello - my first attempt , did not work.
Took your windows directory onto my SCSI-drive. (my os is still PCDOS 7.0) Probably hardware differences, i can not boot without device=ibvpicd.386.
Perhaps later IBM pc xt are easier getting to run accelerated software ? then the older pc 5150 ???
/cimonvg
Click to expand...
Hi Cimon, perhaps this is a difference with the IBM 5160 and 5150. Are you running the stock BIOS? Perhaps it is something with the differences with the board. It would be good if someone with a 5160 could validate Harrison’s build.

Good that it confirms my attempts as well.

Mike
 
Something just crossed my mind @harrison.frazier how do you have the switches set on your board.

There is some settings that determine the amount of ram.

I think in my 5150 I set it to 640kb against the instructions for the inboard. I wonder if by addressing the ram in my system board along side the inboard if it is causing any issues.

I did this partly due to making the maximum amount ram available under dos, whatever I configured the hardware to in terms of available ram directly changed what was shown under dos when running mem.

I’d be interested to know how your ram is configured also for your 5150 on the board switches @Cimonvg

Mike
 
Mike1978 said:
Something just crossed my mind @harrison.frazier how do you have the switches set on your board.

There is some settings that determine the amount of ram.

I think in my 5150 I set it to 640kb against the instructions for the inboard. I wonder if by addressing the ram in my system board along side the inboard if it is causing any issues.

I did this partly due to making the maximum amount ram available under dos, whatever I configured the hardware to in terms of available ram directly changed what was shown under dos when running mem.

I’d be interested to know how your ram is configured also for your 5150 on the board switches @Cimonvg

Mike
Click to expand...
168B7CBC-3FF3-4CA2-A682-37B22DBE98EE.jpeg

So you will all know these pages but my machine is the pc1 without the upgraded bios (no B marking on the rear of the case) yet I used the full 640kb using the system board settings, in place of setting 544kb. I’m considering trying GLaBIOS but as 640kb was working I’d almost forgotten I’d set it this way. Was curious why 544kb would be necessary, as the board itself only actually has 256kb unless a Bios limitation.. but may rule this out.
 
@Mike1978
Mike, I did follow the PC XT instructions when installing and disabled the conventional memory and co-processor. I also actually removed all the mainboard ram chips except for the first bank.

Unfortunately I don't have a 5150 to test with, otherwise I would!
 
@harrison.frazier can you please provide us with a directory listing of both the c:\windows and the c:\windows\system directories with the dates of all the files. From there we should be able sort out whats going on. Thanks.
 
harrison.frazier said:
@Mike1978
Mike, I did follow the PC XT instructions when installing and disabled the conventional memory and co-processor. I also actually removed all the mainboard ram chips except for the first bank.

Unfortunately I don't have a 5150 to test with, otherwise I would!
Click to expand...
Interesting I don’t recall having to disable copro but mine did have one on the inboard.

Unfortunately my system is not in a good place I don’t know if my inboard is dead or if my 5150’s both have an issue as I’m no longer able to fire up my system anymore. But looking into that on a separate thread. Interesting I hadn’t even considered removing ram from the mainboard. Wonder if knowing how the 5160 board switches are set and what ram remains is part of the key to this? Taking the mainboard ram out of the operation is likely a good call as the inboard ram is addressed much faster.

Good insights further perhaps if and when my board runs again I’ll try further!

Perhaps that’s something else I can try to diagnose my board also remove the additional ram chips in my time ram has been a major cause of issues for older kit.
 
harrison.frazier said:
Mike, I did follow the PC XT instructions when installing and disabled the conventional memory and co-processor. I also actually removed all the mainboard ram chips except for the first bank.
Click to expand...
On an IBM 5160, it is not possible to disable bank 0, and so the IBM 5160 motherboard provides at least one bank's worth of conventional memory. Presumably that is why when removing chips (which is optional) that you left the bank 0 chips in.

I see in the IBM 5160 section of the Inboard installation instructions, "You must disable conventional memory on the system board down to 256K bytes. Use a ballpoint pen to set switches 3 and 4 on the system board to ON." Enabling only bank 0. But the first sentence is inaccurate because the figure of 256K is correct for a 256-640KB type of IBM 5160 motherboard, but not for 64-256KB type.

The Inboard installation program should be able to work out the bank 0 size (64 KB or 256 KB), simply by read/write tests, and then configure the Inboard to provide conventional memory onward from there (to address 640K).

Then, at power-on, the POST in the 5160's motherboard BIOS (a POST that sizes conventional memory via read/write testing) will find 640 KB of conventional memory.

Mike1978 said:
Unfortunately my system is not in a good place I don’t know if my inboard is dead or if my 5150’s both have an issue as I’m no longer able to fire up my system anymore. But looking into that on a separate thread. Interesting I hadn’t even considered removing ram from the mainboard. Wonder if knowing how the 5160 board switches are set and what ram remains is part of the key to this? Taking the mainboard ram out of the operation is likely a good call as the inboard ram is addressed much faster.
Click to expand...
In regards to motherboard RAM, the IBM 5150 is different to the 5160. The 5150 motherboard's RAM sockets are permanently enabled - the two RAM switches in SW1 do not disable/enable RAM banks. And removing RAM chips does not disable RAM banks. Some technical information about that is at [here].

That is why the IBM 5150 section doesn't have the user set the two RAM switches in SW1 for only bank 0 (because those switches don't disable/enable motherboard RAM banks).

The Inboard installation instructions instruct that conventional memory provided by cards be removed or disabled (because the Inboard will now provide that memory).

The SW2 switches would need to be set for how ever much conventional memory ends up being present (motherboard + Inboard).

Because of the issue described at [here], I am puzzled as to why the Inboard installation instructions don't indicate a requirement that the motherboard RAM banks be fully populated !!!

With the motherboard RAM banks fully populated, the Inboard installation program should be able to work out motherboard provided conventional memory, simply by read/write tests, and then configure the Inboard to provide conventional memory onward from there (to address 640K).

NOTE: I see in the IBM 5150 section of the instructions, "Because the Inboard 386/PC provides the full 640K bytes of conventional memory," but I find that hard to believe - data bus contention would result.
 
modem7 said:
On an IBM 5160, it is not possible to disable bank 0, and so the IBM 5160 motherboard provides at least one bank's worth of conventional memory. Presumably that is why when removing chips (which is optional) that you left the bank 0 chips in.

I see in the IBM 5160 section of the Inboard installation instructions, "You must disable conventional memory on the system board down to 256K bytes. Use a ballpoint pen to set switches 3 and 4 on the system board to ON." Enabling only bank 0. But the first sentence is inaccurate because the figure of 256K is correct for a 256-640KB type of IBM 5160 motherboard, but not for 64-256KB type.

The Inboard installation program should be able to work out the bank 0 size (64 KB or 256 KB), simply by read/write tests, and then configure the Inboard to provide conventional memory onward from there (to address 640K).

Then, at power-on, the POST in the 5160's motherboard BIOS (a POST that sizes conventional memory via read/write testing) will find 640 KB of conventional memory.


In regards to motherboard RAM, the IBM 5150 is different to the 5160. The 5150 motherboard's RAM sockets are permanently enabled - the two RAM switches in SW1 do not disable/enable RAM banks. And removing RAM chips does not disable RAM banks. Some technical information about that is at [here].

That is why the IBM 5150 section doesn't have the user set the two RAM switches in SW1 for only bank 0 (because those switches don't disable/enable motherboard RAM banks).

The Inboard installation instructions instruct that conventional memory provided by cards be removed or disabled (because the Inboard will now provide that memory).

The SW2 switches would need to be set for how ever much conventional memory ends up being present (motherboard + Inboard).

Because of the issue described at [here], I am puzzled as to why the Inboard installation instructions don't indicate a requirement that the motherboard RAM banks be fully populated !!!

With the motherboard RAM banks fully populated, the Inboard installation program should be able to work out motherboard provided conventional memory, simply by read/write tests, and then configure the Inboard to provide conventional memory onward from there (to address 640K).

NOTE: I see in the IBM 5150 section of the instructions, "Because the Inboard 386/PC provides the full 640K bytes of conventional memory," but I find that hard to believe - data bus contention would result.
Click to expand...
I shared the same skepticism but only due to whatever I set by switches on the 5150 resulted in what showed in mem in DOS as you would expect.

Prior to adding software to manage the memory between inboard and system ram when not using the included inboard ILIM driver. This came up elsewhere why the ram wasn’t mapping in DOS.

There physically was always the same amount of ram in the machine irrespective of how the motherboard switches were set. The inboard driver (ILIM) couldn’t be used alongside Windows 3.1 to manage the ram for DOS. So setting to 544kb as suggested in the manual this limited DOS without the memory manager to 544kb, setting the motherboard to 640kb gave the full 640kb.

Once adding something that managed the ram in DOS and Windows such as 386Max this would match up the ram on the motherboard and inboard collectively and make it fully available to DOS. But the switches on the motherboard was already mapping the ram.

As an aside 386Max was written by the same author Bob Smith of Qualitas as the original ram software provided by Intel with the inboard. I’ve also recently spoken with him over email about it.

Though in all actuality there is only 256kb on the 64-256kb board, the remainder made available by the inboard. So to me it made sense irrespective of the install instructions to set it to the full 640kb and you could do so without a memory manager.

My 256kb board didn’t have more physically and you remove all top up ram boards before placing in the inboard.
 
Mike1978 said:
I shared the same skepticism but only due to whatever I set by switches on the 5150 resulted in what showed in mem in DOS as you would expect.
Click to expand...
MEM (like a lot of information reporting software) is probably getting the figure from the BIOS. The BIOS data area is populated by the POST, and in the IBM 5150, the POST gets its figure by reading the switches.
 
BTW.

At [here] is the section of IBM 5150 motherboard circuitry concerning the two RAM switches in SW1. The signals from the two switches go to the 8255 chip to be read by software.

In contrast, at [here] is the same related circuitry in the IBM 5160. The two signals additionally head off to other circuitry (at [here]) to enable/disable the RAM banks.

At [here], is an example of the kind of information seen in the manuals of some RAM expansion cards.
 
I have an idea about the VXD issue. I have an Intel Inboard 386/PC with a 2 Meg add on memory and an IBM 5150 to install it in but I have not done the install yet, I wanted to get networking working first.

The Intel documentation for the Inboard version of Windows 3.0 mentions the case of installing over plain Windows 3.0. Most of Intel files have new names except for 3 .INI files and DOSX.EXE DOSX.EXE might be the program which calls the VXD for the interrupt controller. My idea is to replace the Windows 3.1 DOSX.EXE with the DOSX.EXE from the Inboard version of Windows 3.0. It will probably be 2 or 3 weeks (I am slow) before I try this so feel free to try it first.

Windows 3.1 probably fixed one or more bugs in DOSX.EXE but I think a DOS box with bugs is better than no DOS box at all.
 
Last edited:
Cimonvg said:
hello - my first attempt , did not work.
Took your windows directory onto my SCSI-drive. (my os is still PCDOS 7.0) Probably hardware differences, i can not boot without device=ibvpicd.386.
Perhaps later IBM pc xt are easier getting to run accelerated software ? then the older pc 5150 ???
/cimonvg
Click to expand...
hello
one more run
well same vga,hd,floppy but now in IBM5160 and it works! I did copy the Harrison.frazier WINDOWS library to my SCSI disc. And my IBM5160 XT now runs windows with DOS box ;)
My IBM XT is a 64-256k "6181696" system board.
Did also a Windows95 test - but no go..
Thanks harrison :)
/Cimonvg
 
Cimonvg said:
hello
one more run
well same vga,hd,floppy but now in IBM5160 and it works! I did copy the Harrison.frazier WINDOWS library to my SCSI disc. And my IBM5160 XT now runs windows with DOS box ;)
My IBM XT is a 64-256k "6181696" system board.
Did also a Windows95 test - but no go..
Thanks harrison :)
/Cimonvg
Click to expand...
Seems it must be a feature of the 5160! Fantastic a second successful test!

Great work @Cimonvg & or course well identified @harrison.frazier whatever is present in your image suggests will work on a 5160.

Now just to get my inboard working again / find another unicorn.

Fantastic :)

Love the community finds for these dinosaurs, some life in the old girls yet :)

Mike
 
You must log in or register to reply here.
Back
Top