Announcement

**Fred H** · 19 September 2001, 09:04

I should disable ACPI in BIOS, before installing WindowsANY.

Fred H

**wkulecz** · 19 September 2001, 15:51

I disagree. If a new motherboard won't run windows 2000 ACPI kernel I'd return it as defective!

--wally.

**Dr Mordrid** · 19 September 2001, 16:10

It's not necessarily if the board can or cannot run Win2K using ACPI. With new boards that's not an issue.

The real problem comes with the peripherals you add to the system. Lots of cards just won't work well with ACPI controlled IRQ assignment. NIC's are prime offenders, but many other types can also cause hell.

If you have one or more of those cards in your system all the mainboard Win2K compatability in the world won't help you.

Dr. Mordrid

**Sciascia** · 19 September 2001, 16:15

I did an install real quick to verify that my HW worked. I did not do the ACPI bypass and I decided to go ahead and do it before I got a lot of stuff setup. I did get it to install as a standard PC the 2nd time after I disabled the ACPI in the BIOS.

What is the difference between ACPI and advanced power management? That is probably a stupid question, but I disabled both. I would really like to have my PC shutdown by itself when I tell it to. Is there a problem with that?

**wkulecz** · 19 September 2001, 18:05

Any NICs that can't work with ACPI need to be drummed off the market. Name names!

A good busmastering 10/100 PCI NIC that works great with ACPI like the Netgear FA310 costs what $15-20 tops. Or are you talking about the lamers built into some motherboards which would be another reason to lose the MB!

I have strong opinions about this stuff as the market is flooded with junk right now to the point its hard to count on anything working right if you don't shop very very carefully!

--wally.

**Sciascia** · 19 September 2001, 22:06

Well, I have a few "older" components and would just like to avoid any conflicts. Doing a fresh instal is not my favorite past time, so I'll just go without the ACPI. I was worried when I saw my FT100, SC, and G400 on the same IRQ. Primarily the FT and the SC which are instumental during capture, so I wiped it and started over. Looks good so far, just can't waitr to see this thing scream.

**Dr Mordrid** · 19 September 2001, 22:41

3Com 3C905C-TX (WOL problems)
3Com 10/100 LAN + 56k Global Modem type III mini PCI
3Com Etherlink XL/XL-FC
3Com Etherlink 10/100 PCI
D-Link DE-530CT+
D-Link ProFast Ethernet
D-Link SN5000TX Fast Ethernet
most any DEC NIC's
many IBM EtherJet 10/100's
many older Linksys
many older SMC
ExpressNet PCI Fast Eathernet with VIA chipset
Realtek network cards using the 8029AS chipset
Compaq Netelligent 10/100tx
Compaq NC3134 Fast Ethernet
ADMtek AN983 Fast Ethernet
ADMtek ADM8511 Fast Ethernet
ADMTek ADM9511 Fast Ethernet
VIA (VT8231 chipped) Fast Ethernet
VIA (VT8233 chipped) Fast Ethernet

Many USB NIC's

at the least.

Dr. Mordrid

**stevenscott** · 19 September 2001, 23:51

Just a note, 3Com has kept up making really good upgraded software for their NIC's. I have dumped other brand NIC's, and my 3Com allways works. They may not be the cheapest, but then again, in upgrades and network changes, they can be kept, and re-used. IMHO.
Also, follow their install directions.

**Fred H** · 20 September 2001, 01:41

Epox boards probably have problems with Netgear NIC.
ACPI? It takes IRQ. I don't need it, myself.

Fred H

**wkulecz** · 20 September 2001, 06:36

If a driver is correctly written to support IRQ sharing I agree it doesn't matter much if you have ACPI or Standard kernel.

But, correct me if I'm wrong, the ACPI kernel assigns a high IRQ (software, like: irq 57) to the driver that demands non-shared interrupt and calls it only after all other shared drivers have delt with it so the hardware shared IRQ appears unique to the non-share supporting driver. I've only seen this behaviour in ACPI W2K kernels and at least in my experience, helps legacy compatability, non hinders it, assuming the MB ACPI is working correctly. The IRQ ACPI takes is for the "vectoring" and is why you won't have conflicts since all the shared IRQs end up on the ACPI hardware IRQ (or on a seperate one on Via chipsets I've seen), non shared get unique hardware IRQ when possible or are re-vectored to "high irqs" which is why you don't run out or have conflicts.

OTOH buggy drivers are buggy drivers.

I wish I had docs on this, but the above is my best guess based on what is going on after setting up about 20 mostly low-ball W2K systems. While I've probably had my hands on a lot more hardware than most people here, I've still only touched a miniscule fraction of the total (much of which is junk I've carefully avoided!). I've hever had to disable ACPI (although I did on my second W2K system out of ignorance, which was a mistake, I've since re-installed with ACPI enabled).

I'm sure there are more opportunities to discover new bugs in "power management" with ACPI, but I disable this on everything except notebooks when running on batteries.

I'm basically cheap and reuse things as much as practical (which is why I was so PO'd at Matrox and the Marvel drivers dropping MJPEG) and it appears that ACPI has helped a lot. Remember I started with Win98se on a PIII-500 where I was always fighting IRQ conflicts, installed OEM W2K dual boot when it came out and never saw another IRQ conflict. As I upgraded this system I was always fighting IRQ problems on the Win98se side until I eventually gave up and banished Win9x for good.

If the EPOX has trouble with the Netgear FA-310 NIC I'd blame EPOX. I've little experience with them, but they were a low-ball special at Fry's a while back. I set one up for a friend and other than a defective SIMM from Fry's, had no problems setting up W2k. She's been very happy with it, but she doesn't do video editing. But with Via, an obscure chipset revision can make a world of difference. I see many motherboards with things glued to the Via chips so you can't see the revision numbers -- I avoid these like the plague!

As to the network cards on Doc's list I've not seen problems with the 3COM cards, perhaps it was an early driver issue. I've a few Dlink cards that have also been trouble free with ACPI, but I'm not sure these are the same model numbers. I've quit buying Linksys (I love their cable/DSL routers and hubs) and D-Link NICs because they change the chipset on the boards without changing the name on the box! This has caused much grief with Linux.

I won't comment on USB "network" cards, perhaps USB2 network interfaces will make sense.

--wally.

**Technoid** · 21 September 2001, 14:09

I can coment on usb network devices:

They Suck!

**Dr Mordrid** · 21 September 2001, 14:22

I was trying to be nice about USB net devices, but you're absolutely correct.

Dr. Mordrid

**Sciascia** · 21 September 2001, 18:25

HEY HEY HEY, what is wrong with USB NICs? I have a Lynksis for my laptop that works great. Granted it is only good for about 10Mbs, but it is very reliable and I don't have to install a "real" NIC. Besides my cable modem is capped at 1240kbs anyway. Fast enough for me.

**Hpar_** · 22 September 2001, 02:05

as far as I can tell wkulecz comment on integrated nic onto mainboards and integrated mainboards in general, I think the vast majority of those integrated mainboards are poop as well with a few acceptions one being the latest tyan dual athlon board the Thunder K7 (S2462). its got 2 u160scsi controllers, a 3com network card, ide (no ide-raid which would have been nice), 64bit pci slots, 4 ram sockets for 3gig of ram, I mean its just about everything you could want in a highend expensive mainboard. the test results on a dual athlon are awsome I understand. it is the athlon board to have right now I think. its not cheep, but you get what you pay for usually (not always).

greets to all the matrox users online :P

Announcement

Req: to Athlon owners

Req: to Athlon owners

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment