"Normal" Windoze freezes, or improper A7V133 configuration?

[Leo V]

After being unable to crash the A7V133 with regular heavy usage, I took my stability testing a level of paranoia higher:

I inflicted unbearable pain on my PC, by running all of these at once under Win98SE:

* Huge directory copy over 100mbps LAN (1 or 2 sep. processes)
* CPU burn-in program
* CoolEdit upsampling to huge WAV file process
* Playing a CD-quality or 88KHz/32-bit (to parasite bandwidth) WAV file over the LAN (or locally)
* Exact Audio Copy top-quality audio CD ripping/testing to harddisk
* Random IE windows/browsing
* Outlook Express open

Surely enough, this soon brought hard lockups. The lockups are fairly consistent in their timing. The one consistent thing (as far as I know) is the audio playback during each crash. It could be CD quality WAV's or other, and I used different playback apps, some played from LAN and some from local HDD. I'll note that all settings are conservative & well within spec (voltage, MHz, etc).

Should I just chill and attribute these to "expected" hectic Windoze behaviour under extreme stress? (Or possibly some driver bug?) BTW I have 512MB RAM and virtual memory is disabled. I know that my 3com nic shares IRQ's with the Promise controller (which I use). It's inevitable however, and it's supposed to share the IRQ. Disabling onboard sound didn't help anything, so it's ruled out. (I use both onboard sound and a PCI card).

 

[mcbiff]

Try adding one of the items at a time and see which one makes it freeze. Then you can check for conflicts or driver bugs. IMHO you really should be running Win2k on that kind of hardware though, but it probably has nothing to do with it.

 

[Leo V]

You're right, it turned out to be a soundcard/Promise conflict. Moving the soundcard 2 PCI slots up seemed to solve those crashes. Now I'm running the supertorture test again. Had another crash due possibly to undervolting (which I reinstated). Now I bumped Vcore up slightly. We'll see how this turns out.

 

[jpprod]

Try dropping IE and Outlook out of the equation, and use a third-party browser in the stress test. I've never seen a BSOD caused by Netscape Navigator (though the browser crashes on certain situations quite a bit) on any system to date, but I've seen lots of BSODs on various systems (not mine though ) with OE and IE. They both use Windows core components, so extreme stress instability with those programs running could be a software problem.

You did mention that you're using conservative hardware settings, but have you tried disabling memory 4-way interleaving from BIOS? This is a feature not all motherboard manufacturers even feel safe enabling. Personally I'm also a bit concerned about that IRQ sharing - I just wouldn't feel safe unless each device had it's own IRQ, the old-fashioned way. That's why I've disabled ACPI on my rig alltogether.

 

[jpprod]

Damn, too slow

 

[Yzzim]

That's a lotta sh#t running there!
Do you smell anything burning? :Q

 

[Yzzim]

jpprod - now quick edit your first post saying that you think it's a soundcard/Promise error and you'll look like a stud

 

[sharkeeper]

<< BTW I have 512MB RAM and virtual memory is disabled. >>



Don't disable virtual memory. Yes, a fixed value of 128/128 is fine. I would NOT recommend disabling it, however!

Cheers!

 

[SiliconVandal]

With 512MB+ you don't really need to have VM enabled...from my testing on a system with 768MB, VM off is much faster. My 2 cents..

SV

 

[Leo V]

&quot;Try dropping IE and Outlook out of the equation, and use a third-party browser in the stress test.&quot;

What a prophecy! Outlook was the only app that screwed up this time (causing a BSOD), but it was closed successfully. My test appears on its way to completion! The CPU burn program completed its 45 minutes and reported no errors. I'm just waiting for the other processes to catch up.