cyclone3d wrote on 2023-06-15, 04:20:

Rikintosh wrote on 2023-06-15, 04:14:

serialShinobi wrote on 2023-06-15, 00:12:

Yes. I am learning that Pentium 4 CPUs were available with an x86 instruction set which is compatible with Win98. […]
Show full quote

Yes. I am learning that Pentium 4 CPUs were available with an x86 instruction set which is compatible with Win98.

But they all used netburst. Which is not a good fit for i686 based software organization.

As in, you know, wasting a much more advanced system on a simple i686 thread.

I wanted to say 1.0 thousand MHz earlier. I felt that it was weird that their clock rate could be exactly one gigaahertz.

Here is a good article I found that compares Netburst to it's predecessor 13 years after the launch of first Netburst CPUs, the Pentium 4.

https://chipsandcheese.com/2022/06/17/i ... -success/

It's also a bit difficult to get pci express cards working in windows 98, it requires a bit more setup.

You can even use the latest Pentium 4 3.8Ghz which will work just as well. Although, a Pentium M running at 2.4ghz is almost as fast as a P4 3.8, but using 1/4 of the energy/TDP required.

PCIe problems in Win98SE ???? didn't have a single problem getting a 7900GTX running in 98Se on a S775 board with an Intel C2D CPU.

PCIe should be transparent to 98SE.

I had problems with PCIe, the system doesn't see it correctly, on some motherboards, the system shows "Standard PCI to PCIE bridge" because the PCIe interface requires a driver to communicate with the video card. In others, there are problems with IRQ, and the way the system understands this by talking to the bios. nvidia and ATI seemed to be shitting on PCIe support in w9x, it was extremely painful to get a geforce 6200 pci-e working

Duffman wrote on 2023-06-15, 01:51:

@VDNKh […]
Show full quote

@VDNKh

There is still an unresolved VCACHE error.

I get a VCACHE protection error on my Ryzen 5950x system even with PATCHMEM and patcher9x.
I haven't found a solution yet.

Have you tired different PATCHMEM options? Or the SPLIT8MB in AUTOEXEC.BAT?

serialShinobi wrote on 2023-06-12, 04:03:

Hello. […]
Show full quote

Hello.

I am wondering if someone who has had a job in the industries of computer science and engineering could help me get my ideas in order.

Is there a cost effective way I could have developed by paid experts a means of running Windows 98 using SeaBios on a hardware target that is both supported by SeaBIOS yet too advanced to be supported by Windows 98?

I assume that a level 1 hypervisor like esxi or ms-hyperV do not have the customization that SeaBIOS could provide.

You could make a list of features supported by SeaBIOS for a given hardware target and have experts add new features to Win98.

Those hypervisors work only within a single instruction set architecture, executing opcodes in a virtual processor sharing the instruction set .

Isn't this a disadvantage?

But if you have Win 98 and SeaBIOS, SeaBIOS could emulate x86 instructions for Win98 on an IA-32 motherboard.

I am aware of aspects of Win 98 that would not be available for reprogramming. But am thinking about programming aspects of Win 98 that are available for getting it to take advantage of the system hardware.

I believe you are getting a bit confused.... SeaBIOS is just a BIOS implementation, it is not a virtualizer nor an emulator. In fact it can even be run on (some) real hardware, replacing the original BIOS from the manufacturer.

VirtualBox and QEMU are two hypervisors which happen to utilize SeaBIOS, and both can run Windows 98 reasonably fine. And as you can see from this thread, a lot of people disregard them, prefering other virtualizers such as VirtualPC which did not use SeaBIOS but some proprietary one. (Not me. I think VirtualBox is just fine.)

The BIOS implementation is going to be giving practically nothing in terms of customization or "reprogrammability", whatever that means, since Windows 9x are already protected mode OSes that try to skip the BIOS as much as possible.

As for running 9x on newer hardware , I have seen it running under the latest Rocket Lake/13th gen Intel (with patcher9x), so it is still definitely possible. That said, it is also _slower_ when running on real hardware than under an hypervisor (even VirtualBox), for many many reasons; the simplest of which is that 9x will no longer setup power states correctly.

I have a slightly controversial opinion. I believe that a good way would be to enable NT to run what 9x runs. If NTVDM were tweaked to behave a little more like dosbox, that would be great. Although there are already discussions about this and this implementation would be difficult because ntvdm and dosbox work in very different ways. VDMSound is a good start.

I refer to NTVDM as the heart of the beast, because in my day to day life, I use 9x to run everything I have, because everything I like runs inside the 9x command prompt without the need to restart or run directly in dos .

XP Sp2 has a very good compatibility with what 98 can do in my opinion, and because it is a "NeoRetro", it is closer to modern hardware than 9x. XP SP2 is by far the closest thing to a 98 replacement (in my opinion) I have a conversion project for it to behave and look 100% identical to the 98SE, but I never finalized it, precisely because I'm waiting for some kind of improved NTVDM to emerge

@Rikintosh

I very much doubt there will ever be anything like what you're looking for, the source code for NTVDM was in the Windows XP leak a couple of years ago and no one has done anything with it.

Unless you make the changes you want to see in NTVDM yourself, it isn't going to happen.

Someone did do something with it but I can't link it here.....

For something that can be linked you can try asking on the ReactOS forums. A couple of years ago there was talk of them working on NTVDM not sure what came of that if anything.
https://reactos.org/forum/

Yes, they were doing something years ago, but I don't know their current whereabouts. There was even a guy who created an NTVDM64 for windows 10.