Motherboard Recommendations?

[junkieclown]

Hello, new member here making his virgin post...
I need some advice about _stable_ motherboards.

I am going to build a new PC soon and the one major requirement is that it must run OpenGL stuff using the Nvidia Linux driver. If possible I'd prefer to base this system on an AMD cpu and VIA KT400 or 333 chipset. However I have heard much complaint about various AMD + Nvidia with Linux combinations over the past couple of years. And I have also heard alot of what seems like spurious info given out as "the magic cure" for system instability. eg: "Compile this source while turning about in circles striking your legs and back with a hazel switch and reciting canticles from The Hitchhiker's Guide"

My old VIA mvp3 system is a crashoholic with Nvidia and Linux. Mainly because I have so many devices attached to it (scsi, ata adapter, usb, eth, lp0, ISA soundblaster) that I can't make the #%^ing board give the Nvidia card its own IRQ. It always wants to share IRQ11 with the video and some other card on it, which I understand is the most frequent factor behind Nvidia/Linux instability. (With wrong agp drivers and inadequate voltage agp slots bringing up 2nd and 3rd)

Basically, if you have true stability with a current setup --and by "stability" I mean continuous uptimes measured by the kernel package release schedule of your distro, and by "current" I mean the most recent motherboards -- I would like to hear what you're running and what peripherals (as in scsi/ ata/ raid adapters, NICs, USB , soundcards, etc.) you have in the interrupts list along with your Nvidia based video card.

If I have to go to an Intel 845x motherboard and PeeFor to get stability I will; but I am most interested in alternatives to that. including SiS chipsets for Intel and ...basically anything that's current and fast.

Thank You.

[drenyx]

not really an answer to your question, but something you may not have considered. I have a tyan s1598 (trinity atx, via mvp3 chipset) board here with my tnt2 ultra. I for a while also had in here a geforce 3 ti 200 which gave me the same issues as the tnt 2. I tracked it down to the NvAGP setting that if you set it to 0 (NvAGP = 0) in the xf86cfg it runs stable, but you are effectively running at AGP 1x. I'd be interested to know if you have similar results as far as stability, was hoping to find someone who had found a problem with the agp driver for the via mvp3 and might possibly have a better solution.

[junkieclown]

Originally posted by drenyx
not really an answer to your question, but something you may not have considered. I have a tyan s1598 (trinity atx, via mvp3 chipset) board here with my tnt2 ultra. I for a while also had in here a geforce 3 ti 200 which gave me the same issues as the tnt 2. I tracked it down to the NvAGP setting that if you set it to 0 (NvAGP = 0) in the xf86cfg it runs stable, but you are effectively running at AGP 1x. I'd be interested to know if you have similar results as far as stability, was hoping to find someone who had found a problem with the agp driver for the via mvp3 and might possibly have a better solution.

Thanks. Just at the moment I am using the tnt2ultra with kernel agpgart (nvagp=0) ...Ooops - that should be "NvAGP" "2"... (rh8.0 stock updated kernel, and agp1x selected in the bios. It's acting more stable than before. However, I think irq sharing/pci steering is a major factor. I could turn agp all the way off and _still_ get crashes. That was with the Promise ata100tx ide adapter on the same interrupt as the nvidia card. I have switched my pci cards around so that the Nvidia now shares irq 11 with my Advansys scsi adapter instead of the Promise. I can't tell if that is the only reason the system is acting more stable (in fact I can't be sure it is _really_ stable just yet but it is definitely improved). If I turn on 2x agp in the BIOS it locks during Q3. I 've checked to make sure the card's agp jumper settings are right. It probably just overheats at 2x.

&lt RANT &gt
Sure wish there was a way to force the system to share a different irq that the one the agp card is on. What device stresses the pci bus/ northbridge's stability ? the agp card What device always ends up sharing an irq with others in a full system ? the agp card. WTF ?!?! Why couldn't latency insensitive devices like usb, & eth0 share irqs, instead of getting their own unshared slots no matter what, and let nvidia, and the scsi and ide adapter have their own reserved interrupts? PC bios sucks the Devil's O-ring.

I'm still hoping to hear success stories about recent hardware. Unfortunately my research on the kernel mailinglist is turning up disturbing info about VIA KT400 and kernel agpgart. No stock kernel ships with support for this chipset. Gosh, wouldn't it be nice if chipset people like VIA thought enough of us to seed Linux kernel development ppl with reference boards for their upcoming products? Or maybe it's the other way around. Maybe it's the kernel developers who need to take emerging hardware seriously enough to ASK for reference boards so new linux distros can support new PC products when the new products come out. It's kind of pointless to ask OEM pc vendors to ipre-install Linux when the kernel work on basic things like mobo chipsets doesn't begin until _after_ the product is already shipping.
&lt /RANT &gt

-appended nov 29-

And my system _still_ isn't stable. Looks like I can expect a crash every couple of days.