There is no need to re-install Windows 98 in order to disable ACPI:

Re: Windows 98se reboots on shutdown

Is pretty easy:
If you will use your machine with older DOS/Win16/Win32 and pre-DirectX7 apps and games, and your hardware supports running full with only VxD drivers, then disable ACPI. Your machine will run faster and stable if the VxD drivers are well made.

But if your will use it with DirectX7+ games and modern Win98SE+ Win32 apps, your hardware only comes with WDM (.sys and .inf files) drivers, and you don't need/care about support for older games and DOS... Or you will run it with Win2000 or XP, then don't disable ACPI. WDM drivers don't like too much running in legacy mode.

Also, never mix VxD and WDM drivers, specially when it comes to Sound, USB and LAN Cards. For pure VxD-only systems focused to DOS and pre-DirectX7 games is recomendable to disable even USB and Firewire, which use WDM only drivers, and can conflict with the system VxD drivers. If you need file transfers for them, is recommendable using LAN or NullModem cable transfers.

Retromangia wrote:

As of right now, I'm currently using it. Although i don't have any real problems, I've noticed in my system settings that I have MANY devices sharing IRQ's (see pic). EDIT - actually the pic doesn't show much, but there's about 4 shared devices using IRQ 11 as well.

This is exactly how PCI is supposed to work and not something to be remotely concerned about.

I've also noticed my BIOS has many IRQ settings. Should I just leave them on Auto? (see pic)

Unless you're experiencing problems, why not?

hyoenmadan wrote:

is recomendable to disable even USB and Firewire, which use WDM only drivers

But Windows 98 FE was completely capable of using USB and had no support for WDM.

Jorpho wrote:

Retromangia wrote:

As of right now, I'm currently using it. Although i don't have any real problems, I've noticed in my system settings that I have MANY devices sharing IRQ's (see pic). EDIT - actually the pic doesn't show much, but there's about 4 shared devices using IRQ 11 as well.

This is exactly how PCI is supposed to work and not something to be remotely concerned about.

I fully agree with you. PCI has two types of interrupts, so to say.
One type is used on the physical side (INT#A,B,C,..) and another on the computer side (IRQ#1,2,3,..)

The numbers of IRQs on the computer side are much higher also, if an advanced interrupt controller (APIC) is in the system.
They were in common use since the Pentium era. To make matter more complex, sometimes the advanced configuration and power interface (ACPI)
is also required to use the APIC properly. But sadly, ACPI is very complex and intentionally buggy. I heard Linux OSes had a hard time supporting it,
because of undocumented quirks. Also, there were different versions of ACPI support in Win98,98SE and Me.

Not sure if this is for APIC or non-APIC, but to avoid excessive sharing, make sure the 4 IRQs are free - disable unused serial and parallel ports.

Typically, the PCI IRQs (#A to #D) are rotated around the slots and onboard devices, so that a card using INT #A will get whichever is a rolled over to #A position for the slot. the 4 PCI IRQs are drawn from spare IRQs not already assigned as ISA or fixed devices.
If there are not 4 available, sharing is greatly increased!

Since PCI IRQ is implemented as level (retrigerrable) rather than one-shot edge, PCI is designed so that devices SHOULD be able to share, but some devices just don't play well with others

Jorpho wrote:

hyoenmadan wrote:

is recomendable to disable even USB and Firewire, which use WDM only drivers

But Windows 98 FE was completely capable of using USB and had no support for WDM.

No really. USB support in Win98 (and Win95 by extension) has been always done via WDM, and since Win95 is present in a limited way via the "Detroit" update package (Detroit.exe). As side note, this driver model was originated in NT5 (Win2k) during its development, and then backported to Win9x series, by including a subset of the NTOS kernel as a system VxD (a complete and real heresy, which made the whole system 50% less stable). MSFN, and rloew has more information about this specific subject.

Jo22 wrote:

To make matter more complex, sometimes the advanced configuration and power interface (ACPI)
is also required to use the APIC properly. But sadly, ACPI is very complex and intentionally buggy. I heard Linux OSes had a hard time supporting it,
because of undocumented quirks. Also, there were different versions of ACPI support in Win98,98SE and Me.

APIC can be used natively if OS knows how to handle MPS specification. In NT OSs this support is accomplished via an special HAL system dll. But actually MPS is a complex way to handle it, and since it was done with servers in mind, it doesn't handle stuff like power management or BMC management. That's why ACPI was created. ACPI handles not only PowerManagement, but it also handles device enumeration, SMP support, resource management, APIC control, and custom firmware procedures for specific tasks. That's why it is so complex. It abstracts platform details from drivers and OS core. And about the bugs... Not ACPI fault, but ISVs doing stuff in a non standard way. And Linux... Well, is Linux... Them have problems supporting anything because the OS requires open code, which ISVs will never support in an active way, as their commercial secrets get revealed in the process.

hyoenmadan wrote:

No really. USB support in Win98 (and Win95 by extension) has been always done via WDM, and since Win95 is present in a limited way via the "Detroit" update package (Detroit.exe). As side note, this driver model was originated in NT5 (Win2k) during its development, and then backported to Win9x series, by including a subset of the NTOS kernel as a system VxD (a complete and real heresy, which made the whole system 50% less stable). MSFN, and rloew has more information about this specific subject.

Well, that's interesting news. I don't suppose you'd have any specific links? Was Windows ME much the same?

Jorpho wrote:

Well, that's interesting news. I don't suppose you'd have any specific links? Was Windows ME much the same?

http://www.msfn.org/board/topic/172894-window … k-of-ntkernvxd/
https://www.betaarchive.com/forum/viewtopic.p … =29274&start=25

And yep. WinME is much the same as Win98 and 95 in terms how the WDM driver model was implemented on them, but it includes support for more WDM classes, as WIA scanner support, DVD/Audio/Video kernel streaming and proper USB printing support. In fact, WDM support is so broad in WinME, that MS recommended its use with only WDM drivers and ACPI enabled systems, and discourages its use with legacy, VxD-only driven hardware with no ACPI support. Certainly this is proven by WinME being clearly less stable when used in legacy mode with VxD drivers.

Thank you everyone for the great info here. I believe Hyoenmadan put it the best way, but let me re-iterate...

Since I'm using Win98 SE with Direct X7, and only .INF WDM drivers, I can leave my ACPI settings alone, and just leave it ON correct?

Just curious, would I see any performance increase if I turned it off?

And lastly, I don't need to worry about shared IRQ's anymore right?

thanks everyone for the help! 😉

-Retro