AMD system 'random' crashes/reboots

[Tiberius13]

**** EDIT: Problem resolved. See last post. ****

I have had, and know others who have had this problem!! If you (or ANYONE out there) knows of the root case, PLEASE let me know!! Thank you in advance!!

Here is a breakdown of my computer, and the problem my AMD computer has:
AMD 1.333, Asus A7M266 MB, 128 DDR 2100 RAM, 40 Gig hd, 300W power, ATI AIW Radeon video, SB Live Value, DVD drive, floppy, network card, Logitech wireless keyboard and mouse, Win2k (also tried WinXP with a different 100 Gig HD... same problems occurred)...

The issue: The computer randomly crashes... it will stay up for anywhere between a couple minutes to a couple days. The crashes are not the system 'freezing up' due to use. They are sudden instantaneous events. The computer blinks out and does a hard reboot. If I change the setting in Windows to not reboot automatically on a crash, it goes a blue screen of death where it gave a wide variety of messages (I don't have my notes with specifics handy, but there were at least 10+ different types of errors I have seen) If I leave a program such as Bearshare (ie. Napster type program) running overnight, it is pretty much a gaurantee the computer will crash before I wake up. Doing things like playing back movies, video capture, or gaming don't seem to trigger higher instances of crashes. Many times the crash occurs when I return to the computer after a period of time away from it. As soon as I click on a file to run it... it's crashes... (it should be noted that the computer is usually always running mirc, and often downloading files)

So basically I think I have eliminated the hard drive (by trying 2 different drives), and a bad OS install (by trying Win2k and XP).
That leaves: CPU, RAM, Motherboard, power supply, cables/wiring, and finally - possible weird hardware/software conflicts between components I have

Has anyone else experienced problems like this?? I have heard from a bunch of AMD users and they have reported the same problem, so I'm sure SOMEBODY out there must have figured this one out by now?? Is it a bad CPU? bad RAM? bad MB? HELP!!!


<P ID="edit"><FONT SIZE=-1><EM>Edited by Tiberius13 on 02/14/02 02:35 PM.</EM></FONT></P>

 

[ZER0]

so it crashes all the time u got a great deal eh?

 

[Kelledin]

Sounds like a memory error to me. I had the same problem with my first Slot A Athlon, and the problem went away one I got some decent memory.

Try running some default tests in WinBench. If they bring it crashing down in less than ten minutes, chances are pretty good you have memory errors. Power supply issues or heat issues will tend to cause hard lockups more often than random reboots. Also get yourself a copy of memtest86 (d'oh, it was posted the other day, but I can't remember the URL for it. I never use it anyways).

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[svol]

First check if your temps aren't too high and that you don't have Voltages that differ to much (which indicates a bad powersupply).
Check if your memory CAS latency is set to 2.5.
If this still doesn't solve the problem check if you got the latest BIOS en drivers.
You also might want to try the 'Via Latency Patch' from this site:<A HREF="http://www.networking.tzo.com/net/software" target="_new">http://www.networking.tzo.com/net/software</A>.

Good luck.

My case has so many fans that it hovers above the ground .

 

[Tiberius13]

Ok.. some follow up info for you guys:

1) I have the most up to date BIOS updates for the motherboard
2) I am 90% sure I installed the Via Latency Patch a long time back. However, I will check that tonight to be sure.

Also, I will run memory and psu tests and see what kind of results I get, along with checking bios settings for CAS latency and RAM timing for FSB Rate which I believe should be 4:4:1 (133/133/33).

My biggest suspects now seem to be the psu, or RAM... with the motherboard and cpu still in the running....

I originally thought this problem was heat related, and within 2 weeks of buying the computer had the heatseak/fan changed at a different location. The guy looked at the current heatsink/fan and said it 'should be good enough'... he then put in the Super-Orb I currently have. It improved the temp by a couple degrees, but didn't fix the crashing problems. I also run my computer with the case off most of the time to reduce heat. I will try setting up a fan to blow into the case starting tonight. I want to eliminate heat as a cause 100% with certainty (although I am quite certain heat is not the problem at this point).

I appreciate all of the ideas and help, and will get back to you guys with results from ram and psu tests tonight.

 

[Civilized]

The first thing that popped out ot be is lots of different erros...most of the time this is memory related....i justhad similar probs with one of my ram sticks and figuring it out was a bitch!....especially when you have two sticks of ram and it was running fine before it happened...so i just thought i had hardware conflicts wiht my sblive card so itook it out and i still had the probs ...so i reformatted and still had probs...by this time i was clueless and took out everything except my video card...and eventually took out each stick of memory one at a time and finally it worked when i took out the supposably bad stick...and it works fine.....good luck

 

[Tiberius13]

Thanks for the input Civilized! I plan on upgrading my RAM soon anyways (I'm running with just 128 because DDR cost so much back when I first got the computer.. spent almost $200 for the 128) If the place I bought it from jerks me around when I take it in for 'service', I will just buy a new 256 stick of RAM and try it with that by itself.... if it magically becomes stable, I'll know the original 128 is the culprit. But, I'm hoping that the place I got the computer will swap the RAM anyways to see if that fixes it.... we'll see...

 

[Tiberius13]

If you come across that link for memtest86, let me know. Otherwise I'll poke around a bit for it tonight.

Thanks

 

[Oracle]

My guess is that you barely have enough RAM. Add at least another of those 128Mb DDR babies and the show will go on!
I used to have this problem. Win2000-XP is a RAMeater! I upgraded from 128 to 256, and my problems were solved.

 

[Kelledin]

<A HREF="http://www.teresaudio.com/memtest86/#download8" target="_new">http://www.teresaudio.com/memtest86/#download8</A>

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[Tiberius13]

I'm not sure where to go to set the CAS Latency for my RAM.... I couldn't find a setting for it in my BIOS settings... :/

I am currently logging my psu info. Here is what I have so far:
It's a Turbolink ATX-300 (300W)
+3.3V = 14 amps
+5V = 30 amps
+12V = 10 amps

The +3.3V reading seems to always be 3.34V
The +5V reading is usually 4.95 or 4.97, and stays in the range of 4.92 - 4.97 (so far)

My CPU temperature is showing 46 C (this is with the case off, and a large (16") fan blowing into the case at close range)
The motherboard temp is showing 24 C

** Does this info seem to rule out a power supply issue??

I am going to do memtest86 on it now... and see what that tells me....

 

[blah]

what kind of NIC do you have?

..this is very useful and helpful place for information...

 

[texas_techie]

First, did you install the via 4-in-1 drivers? A via based mobo is crap without them. I think XP comes with them now, but not sure.
Second, make sure your front side bus speeds are set correctly. I had one board come with them set wrong (too high) and the system had a hard time posting.
Finally, you said leaving the system going for a while garauntees a crash. Sounds like your power managemnt is screwed. That little program in bios (forget the name, acpi?) that tells the comp when to power down, power up etc.. I had a Gigabyte board that the power mgmt. part was screwed up on. So i removed the power scheme, it helped. Instead of crashing daily, it crashed a couple of times a week.
I am having my doubts about via. But then again, i have a FIC AD11 that is stable as hell. never had a single crash.


year 2010: Intel? Whose that?

 

[Tiberius13]

I have a Realtech NIC... I'm sure it's a cheapo one....

Just got another crash this morning... computer survived the night, but, as a couple minutes after I started clicking around... it crashed.

Here is a bit of the blue screen info:
<standard WinXP message... "A problem has been detected and windows has been shut down to prevent damage to your computer...." etc> This message includes recommendations to check for BIOS updates (done), disable BIOS memory options such as caching or shadowing (shadowing is off), etc

Technical information:

*** STOP: 0x0000008E (0xC0000005, 0x805596B9, 0xF41D1BF8, 0x00000000)

Beginning dump of physical memory
Physical memory dump complete
Contact your system admin..... etc

There is no driver identified or additional info this time.

Some of the messages I got under Win2k however were:
ntoskrnl.exe - A device driver has corrupted the executive memory pool. MEMORY_MANAGEMENT
REGISTRY_ERROR
KMODE_EXCEPTION_NOT_HANDLED
IRQL_NOT_LESS_OR_EQUAL
NTFS_FILE_SYSTEM
PAGE_FAULT_IN_NONPAGED_AREA
PFN_LIST_CORRUPT

When I ran memtest86 last night, it seemed to be passing everything. The psu is still showing a +5V range of 4.92 to 4.97. Also I have a large house fan blowing into the open case to keep things very cool and eliminate heat problems.

.... I'm still stumped...

 

[Kelledin]

That probably means exactly what it says--a device driver is blowing your O/S install. What I'd suggest:

Format C: and do a fresh install. For WinXP, just install it without installing the 4-in-1's (WinXP includes Microsoft's reference VIA drivers). For Win2K, install it, then apply Service Pack 2 (SP2 includes Microsoft's reference VIA drivers).

Apply the AMD+Win2K regpatch, or make sure it's applied. IIRC it applies to WinXP as well as Win2K.

If Windows has its own reference driver to run your NIC, use that. It will probably autodetect and install its own driver if it has one. Upgrade only the drivers you absolutely need to run WinBench or Sandra (probably just video drivers).

Install and run your benchmark/burn-in utility. If it passes, install a manufacturer-supplied driver for something.

Run benchmark/burn-in again. If it passes, install another driver.

Keep doing this until you hit the one driver that blows the benchmark test. That's probably your problem driver.

Another possibility is that the CPU core is chipped. T-birds have a somewhat brittle ceramic packaging and a rather delicate core; this was improved considerably in the AthlonXP. You'll have to take the HSF off to check for damage, of course.

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[Tiberius13]

FSB speeds? I might need some guidance in how to tweak that. Everything I have looked at in my BIOS settings seems to indicate that all is set correctly. My understanding is that if my FSB was set to 100, my computer would run at 1000? But since it is set at 133, it is running at 1333. The good old formula of cpu clock multiplier x bus speed = computer speed ? Is there another bus setting I need to worry about?

I had installed the 4 in 1 drivers on my win2k install - although it was likely just the version from the cd I got with the motherboard (I bought the computer in April, 2001)

I just put the 4 in 1 drivers on my XP install last night (I have only been running this XP install for a couple of weeks so far) However, according to the VIA website XP only needs 1 of the 4 drivers anyways (whereas win2k required 3 of the 4) Whatever, the install program supposedly detects what you system is and puts the right stuff on it.

I also got the AMD 760/761 chipset drivers and installed them last night (although it didn't change anything.. seems I had the most recent drivers already)

I have not put the "Microsoft Windows 2000 Patch for AGP Applications" on my XP install yet - as I wasn't sure if the issue had been addressed for XP (I love how XP is never mentioned on the AMD site.... leaves everyone guessing for now) I am going to install it tonight though since XP is supposed to be 90% Win2k core code anyways.

 

[Kelledin]

You haven't by any chance installed the VIA AGP driver, have you? Make sure you have <b>not</b>--you want the AMD AGP driver, not the VIA AGP driver. I'm pretty sure having the VIA driver in there would screw a few things around...

<i>If a server crashes in a server farm and no one pings it, does it still cost four figures to fix?

 

[Oracle]

Stop wondering and expressing theories. YOU DON'T HAVE ENOUGH RAM! 128Mb is barely enough to get the system started under Win2000.

<font color=red>Floppy disk?!? What the heck's a floppy disk?!?</font color=red>

 

[Tiberius13]

Here is the link to what I installed:
http://www.viaarena.com/?PageID=2
And this is directly to the HTML guide for the 4 in 1 drivers.
http://downloads.viaarena.com/installguides/DriverInstallationGuide.htm

When I ran the install program, it actually showed 2 drivers it wanted to install (I'm running WinXP, so the documentation said it should only install one driver.. the Via Inf Driver.. but it also wanted to installed the VIA IDE Bus Master Driver (IDE miniport) - I deselected that driver and did not install it)

I did Not install the via AGP driver from the 4 in 1 drivers. (unless XP installed it on it's own by default - which I doubt)

I will install the AMD AGP driver (for Windows 2000) tonight... unless I hear otherwise from someone. (I'm sure Microsoft didn't bother to fix this issue in WinXP... that would just be asking too much I guess)

 

[khha4113]

I remember a while back when someone mentioned incorrect RAM voltage (2.8V, sorry I don't have it) on Asus A7M266 manual that caused his system unstable. He changed to 2.5V by removing jumper (<i><font color=red>V01</i></font color=red>) and his problem solved.
<b><i>Edit:</b></i>
OK, I found <b><A HREF="http://forumz.tomshardware.com/modules.php?name=Forums&file=faq&notfound=1&code=1" target="_new"><font color=blue>it</A></font color=blue></b>

Good or Bad have no meaning at all, depends on what your point of view is.<P ID="edit"><FONT SIZE=-1><EM>Edited by khha4113 on 01/30/02 02:03 PM.</EM></FONT></P>

 

[Tiberius13]

Thanks for the link! Interestingly enough, it seems that going 'jumperless' on VIO1 actually sets it to 2.9V (he said he was at 2.8V... and I believe he thought that was "too high"... heh)

I will try some different settings for this VIO1 voltage as it seems that AMD machines like the higher voltage, as long as the ram/system can handle it I guess....

 

[kief]

Most of the advice here seems good. Its gotta be one of three things:

bad ram (or aggressive bios settings on the ram)

bad power supply

device driver conflicting with OS or another driver

You can try to set bios to "failsafe defaults" or go in and manually set each setting for the lowest value (so it does not change other stuff)
For power supply you could try another.
For device driver update all the available drivers, and also check the system logs for info on which driver caused the error!

Jesus saves, but Mario scores!!!

 

[kief]

PS to my last post, I DID have this problem on one of my many systems when I installed a slew of new hardware and the driver for my 3com managed ethernet card was the problem. I swapped out the NIC for an old Intel I had sitting around and it has been up and running for 77+ days now with only one reboot (when I installed a modem to backup my cable connection)....

Jesus saves, but Mario scores!!!

 

[lhgpoobaa]

best way to test for PSU stability is the following.

run just your montioring program, nothing else
(i use motehrboard monitor 5)
watch the voltages, see if they vary or oscilate at all on idle.

then, while keepin an eye on voltages, run a highly cpu intensive applicaiton like prime95, seti@home or my favorite, TOAST.

tell us how much the voltages dip, expecially on the 5v line.

if u get fluctuations and/or excessive dip, chances are the PSU cant cut it.

other 'poor psu' indicators are:
excessive PSU heat output (my 300W generic was putting out almost double my Enermax unit does)
Audable fan speed changes, particularly when going from CPU ide to max load.


The lack of thermal protection on Athlon's is cunning way to stop morons from using AMD.

 

[Tiberius13]

Interesting solution to your problem... I just might pull my Realtech NIC and SB Live card from my PC soon just to see if that might help....

I'm trying to download some of those CPU intensive programs being suggested here, and some graphics benchmarking suites, etc. If there is a video card related issue maybe running some intensive graphics benchmarking might flush them out.

The real world 'tests' I have given the computer are fairly decent however. I have capped tv shows using the ATI AIW Radeon, and I have used virtual dub to encode those huffy caps (about a 10 gig file saved for a 1 hour show) over to a 200 mb divx with mp3 sound. I have also captured and encoded at the SAME time... very cpu intensive!

I have had crashes during this process only a couple of times. If I reboot my machine just before I start doing it, I can usually do both at the same time for a couple of hours without a crash. Putting that strain on the computer without it resulting in higher instances of crashes has always puzzled me... (but leaving Bearshare running overnight with a long queue list and up to 10 downloads going almost gaurantees a crash overnight, or as soon as I go on the computer in the morning and try to launch a file or something... )

I think tweaking some voltage settings on the motherboard will be my next step as I have seen people say that it tends to make AMD systems more stable with my motherboard.... we'll see...