a gentleman by the name of Rudolph Loew made patches for dos windows to handle lots of memory and also talk with massive disks.
here is a link to a cache of his works for everyone to research and i have made virtual machines running massive amounts of memory and disk space.
https://www.philscomputerlab.com/rudolph-r-loew-patches.html
His works are on archive.org too perhaps a system where these patches can make bare bones run better with this project or even virtual machines. I am now making such a virtual machine and uploading to archive as demonstration.

here is a Vogon link for information: Rudolph Loew apparently passed away in September 2019

I've tried this patch, but my issue isn't with trying to use a lot of memory; although I've tried that too. The problem is the system running out of memory while still having plenty of memory available. I'm still troubleshooting this, but it's difficult to do since I'm working on this one machine and can't validate on a virtual machine.

https://msfn.org/board/topic/183777-how-to-de … n-windows-9xme/

It doesn't matter how much our how little physical memory is installed or how I configure SYSTEM.INI. I can still pressure the system into running out of some kind of invisible "memory", as reported by Windows.

@Kahenraz:

I’m afraid that C2D or C2Q CPUs are too new to run Windows 9x on bare metal. I tried years ago on 945 chipset with some low power Atom and results were very poor – 945 GPU worked with VBE drivers but all disk drivers worked only in BIOS mode and I hadn’t no luck with HDA – no sound. But main problem was stability – completely random errors and freezes.

But still – I done testing on real hardware and patch works fine even on real HW even so, I tried to increased compatibility and rewrite patch a bit. Experimental built is at attachment, please try it and report results (if original patch was applied the patcher upgrade it).

Cygwin works without problems for me on Ryzen 1400 (with 98 SE). I used this EXE:

1http://ctm.crouchingtigerhiddenfruitbat.org/pub/cygwin/setup/legacy/setup-legacy.exe

And enter this mirror:

1http://ctm.crouchingtigerhiddenfruitbat.org/pub/cygwin/circa-legacy/2009/12/27/103117

Offline installation from archive.org was unable to install base system but this isn’t Windows problem but Cygwin. Either way parse INI works all time. There is screenshot (Windows in Czech version):

The iGPU is a complete write off for the i945 chipset, I agree. I was able to resolve my hard disk issues by setting the SATA ports to "compatibility" mode. They work fine this way, but only two of the four ports are available. They aren't running as AHCI, but DMA works fine.

I'll test these new patches and let you know if it solves my issue. Thank you for the quick update.

The operating system is still locking up during "Parsing ini file". It managed to complete once, but all other times it's still locking up randomly.

I also noticed that we're using different versions of Windows. I'm using Windows ME (English) from MSDN, if that makes a difference.

I created an account here, just to say thank you for this patch. I had pretty much given up on Windows 9x platforms in VMWare since rebuilding from the 7900X to a Ryzen 5950X.
It took a lot of searching, but eventually I ended up here, and I cannot thank you enough. I'm an old school PC user and while I still have a few DOS and 98 machines, I like being able to use them on my primary PC.

Thanks for making this patch! I was looking for a solution to - windows 98 vmware 6d007eH - and this thread popped up.

Your patch is working well for me on Windows 98 SE (4.10.2222B) with VMware Player 16.2.4 on a 5600X CPU.
This is after installing the Windows "Security9_enu_15" and re-patching. The application crash situation is much improved with the 'patcher9x-0.7.45-boot' latest version from the Github.

Right now that's been a great help for me converting some documenation out of some old Windows 3.x software called FrameMaker, which would've been much tougher to do with a separate PC. 😁

---------
For the install, I got fairly far in the installer until it gave the first error message. Then I rebooted onto your floppy image, ran the patcher and Windows 98 then started up okay. After that I installed the security patch CD just to be sure it was a solid install of Win98, which it is now after re-patching.

First of all: thanks for making this patcher and for sharing it, JH64! 🙂

JH64 wrote on 2022-07-27, 23:22:

I’m afraid that C2D or C2Q CPUs are too new to run Windows 9x on bare metal. I tried years ago on 945 chipset with some low power Atom and results were very poor – 945 GPU worked with VBE drivers but all disk drivers worked only in BIOS mode and I hadn’t no luck with HDA – no sound. But main problem was stability – completely random errors and freezes.

Kahenraz wrote on 2022-07-28, 00:31:

The iGPU is a complete write off for the i945 chipset, I agree. I was able to resolve my hard disk issues by setting the SATA ports to "compatibility" mode. They work fine this way, but only two of the four ports are available. They aren't running as AHCI, but DMA works fine.

The late Rudolph R. Loew was mentioned in his thread before. One of the other things he wrote was an AHCI driver for Win9x. I couldn't get it to work in a QEMU VM, but he probably only tested it with bare-metal hardware. But Kahenraz, you could give this driver a go on your i945 chipset. It would be nice to somehow get this driver working with hypervisors as well, since AHCI not only provides higher throughput, but apparently has lower emulation overhead than Legacy IDE As well. Unlike most of his other works, the source code for this driver is not listed on his memorial site (yet).

I did try the AHCI driver, but there is some kind of conflict when I enable it. The issue, I think, is that there are four SATA ports and one IDE port. Windows 98 gets upset for some reason and bluescreens when trying to map out all of the channels.

I tired lots of different combinations, but ended up setting the BIOS to "compatibility mode" which is IDE compatible and disables half the SATA ports, so there are only two channels. This also disables AHCI, so the driver won't have any affect at that point since it's no longer recognized as a SATA device.

I don't recall the issue exactly or if it would work without the IDE channel, but I use an IDE disc drive for CD audio, which I don't think is a feature of on SATA drives. I also use the second channel for a front-accessible CF card slot, so I'm not inclined to use SATA exclusively otherwise.

Do you have any idea on this?:
On h87 with an i5(haswell) and ati x600(pcie) Windows 98 it worked(bare metal).On z690 with i7(alder lake) I get vcache protection error.
Another person had running windows 98(bare metal) on Ryzen 3900X system on a X470, he updated the bios and got the same error "vcache protection error".

Here are the threads:
https://msfn.org/board/topic/183200-z690-vcac … otection-error/
https://msfn.org/board/topic/183543-win-9598m … ew-motherboard/

Can you confirm whether you're using Windows 98 or 98 Second Edition?

Windows 98 SE

I can confirm that on my new PC with Intel Core i3-12100F and MSI PRO H610M-B DDR4 motherboard with the latest UEFI version (1.70), I am experiencing the same VCACHE Protection Error (needed to use HIMEMX to get to that stage).

I could have tried older UEFI versions but do not want to risk my new motherboard just for this, I doubt that would help anyway...

It would be interesting to see OP take a look at this issue as well, as it's a new one that has popped up with the release of the recent motherboards/UEFI versions, though this requires real recent hardware as it's not a virtualization bug, it would be true to the topic title though 😁

softice, maybe it can give some hint or an idea. After I press C to continue, "Vcache protection error" appears

If anyone ever struggles with this as well: This patch helped me getting Marvell Yukon NDIS2 drivers to work on a fast CPU.

I installed Windows 98SE on my Asus P5E3 WS Pro motherboard, which uses an integrated Marvell chipset (88E8056) for LAN support.
There are no native drivers for Win98, but Marvell provides NDIS2 drivers for DOS.

Without this patch they can be installed but won't work properly. You can only ping 127.0.0.1 and make no outside connection. Installing the NetBEUI protocol produces a bluescreen and then an error on boot until you remove the protocol again.

After installing this patch it just works. Install the driver, get online, boom.
Thanks for helping me build an overkill Windows 98 PC!

Oh wow, I never knew this existed. I have an extra Ryzen 5600 CPU that I didn't know what to do with, but may just build an overkill Win95/98 system with it. 🤣. This will be an interesting project. Too bad the Mobo doesn't have some ISA slots to throw in some really old cards for fun.

Just wanted to say thanks so much for developing and releasing this; it's a boon to researchers. Appreciated!

JH64 wrote on 2022-06-01, 22:38:

Larger problem was that code is in file VMM.VXD which is part of WMM32.VXD which is compressed archive of more drivers generated specifically for your HW configuration by installer.

Can you explain how this file works and what it contains? I sometimes use CompactFlash cards with Windows 98 on them that I move between different hardware without reinstalling it. Is there potentially a problem with this behavior from this file when moving between very different chipsets or CPUs?

Anyway to get SB16 to work? I click it and nothing. Wave works fine but not MIDI or music in games that require it.

Thanks so much for the patch, this is definitely helping me get some of the Win 9x series working in VMWare on Ryzen 5800X. I have a particular problem with Windows 98 (FE) though, and it seems to be specifically related to the Ensoniq AudioPCI drivers that I'm using as my system is Ok with the patch applied and no audio driver (no sound though, of course), but after installing the sound drivers it hangs right after the boot sequence starts. This seems to be fast CPU related because I can run the exact same VM fine if I run it in a nested VM (in particular running on an older workstation version running on Windows 7, which in turn is running in Workstation 17 on my native Win 11 host). Only when I drag the Win 98 VM out and try to run it in Workstation on the native host does it have problems. I can use this same approach (dragging between nested and native workstation instances) with other 9x VMs and they seem to have no problems.

Any suggestions to get this working?