How do I fix this IRQ issue under Win2k?

[DarkSyd]

I might have narrowed my problems down to an IRQ sharing issue. My IRQ 9 seems to want everything. I've tried disabling PnP OS, forcing IRQs, but irq9 always ends up with all of my PCI devices. Is this one of the 63000 bugs?

IRQ Number Device
9 Microsoft ACPI-Compliant System
9 Matrox Millennium G400 - English
9 Intel 82371AB/EB PCI to USB Universal Host Controller

9 Vortex Multifunction PCI Parent
9 Aureal Vortex 8830 Audio (WDM)
9 Diamond Multimedia Systems SCSI Host Adapter

8 System CMOS/real time clock
13 Numeric data processor
6 Standard floppy disk controller
1 Standard 101/102-Key or Microsoft Keyboard

12 PS/2 Compatible Mouse
14 Primary IDE Channel
15 Secondary IDE Channel
3 Communications Port (COM1)

Rather odd eh?

DS



------------------
Cel 366@550 (monster FDP32...love it!) on Asus P2b (1011), 128MB PC133, OEM G400/16 (PD5.41), Fujitsu 5.25gb, Maxtor 4.3gb, Asus SC200 PCI SCSI card, Pioneer slot load 32X SCSI, Yamaha 4416s burner, MX300 (Aureal 2040), Intel EtherExpress 16, Sycom 300va UPS, Grey Cat, Orange Cat (still MIA)

Where's my 500e?

[Scytale]

That's the way ACPI likes it. Try a search, it has been discussed by Gurm and others in this forum.

[Gurm]

*SIGH*

Ok, one last time in case any Win2k early adopter warez kiddiez are still unclear on the subject:

ACPI enables HARDWARE POLLING. This means that, despite what the computer is telling you, NONE of your devices are using ANY IRQ's AT ALL.

Instead, Windows decides when the device needs some time from the system, and doles it out as needed.

Increases stability with a very very minor performance hit.

It normal. Doesn't mean a thing. If you're having problems, it's likely something else in your system.

- Gurm

------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

[DarkSyd]

Easy Gurm, I haven't been an "early adopter warez kiddie" for about 20 years now. There's a big difference between that and a consultant. We deal primarily with CAD related hardware/software...OS's aren't my specialty
I must have missed your previous posts on the subject and have been unable to find anything on the issue as MS's site. If that's the way ACPI likes it...then so be it. Should I have the "Plug n Play OS" option enabled or disabled in my BIOS?

James

[Scytale]

I'd give an arm and a leg to be an early adopter warez kiddie again

Anyway, DS - here's a pretty informative link: http://forums.murc.ws/ubb/Forum5/HTML/007376.html

[DarkSyd]

Good stuff.
Thanks for the link.

James

[Gurm]

Darksyd,

Don't take it personally. Honest. Heh. I will try to get the link from MS that describes ACPI. You could have found it with a little poking of your own.

HOWEVER, for all OTHER WAREZ KIDDIEZ...

Hehe.

- Gurm

------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

[Chasbo]

Gurm,

In the meantime:

http://support.microsoft.com/support.../q232/8/24.asp

[PaulS]

Gurm's right, though just to clarify, ACPI doesn't disable interrupt handling, it simply enables sharing. Devices that require interrupts still generate them, and their interrupt handlers are still executed.

Normally a device generates an interrupt when it needs the processor to do something. Under NT, the interrupt handler for the device is supposed to nothing except schedule some processor time for the work to be done and clear the interrupt. The entire OS only works if devices get out their interrupt handlers ASAP, doing the actual work outside the handler.

The bunched up interrupt thing is partly irq steering gone mad. With ACPI, Windows can reassign interrupts on both PCI and ISA cards (provided they are plug and play). So on the off chance that an interrupt might be required for a non PNP card, Windows tries to keep as many interrupts free as possible.

The polling thing only happens after the IRQ has been generated, Windows then checks each card on that interrupt to see who did it. There is a kind democracy here, normally higher priority interrupts interrupt and mask lower priority interrupts, delaying their clearance. By putting everything on the same interrupt you reduce the overall handling time because you are less likely to be interrupted by a higher priority interrupt.

Paul