»

[scafe]

Wonder if any one can help
Recently i started experienceing segmentation faults on my AMD XP2200 PC within linux (fedora core 1) which progressivilgot worse untill the computer became unuseable to the extent of linux crashing due to kernel panics every boot! Reinstall failed to fix this, and i even resorted to usin windows but to no avail.
The hard drive is fine, but a memtest revealed faults with the memory which i then replaced with some decent Crucial memory (pc2100 - 128mb), this has again failed to fix the problem, but further memtesting has revealed that this might be as a result of the L1 cache. If this is a likely cause of my problems what do i need to replace, as i dont want to end up replaceing unnessisary items, my bank balance cant cope as it is!

base spec as follows
ASUS A7V8X-X M/board (via chipset)
AMD xp2200
1 single pc2100 128mb in slot 0
ATi Radeon 9200se 128mb g/card

Any help would be grately appreciated :-D

[KennyR]

How is your cooling? Modern AMDs are very hot beasts and will definitely crash if they can get too hot. Maybe a clean in the fan and under the cooler to get some dust out might help.

[Robert17]

try putting the ram into a different slot maybe? Just the usual like check the bios make sure ram speeds are all correctly set, you could try standard timing modes?

[scafe]

humm... I have a look at the cooling, perhapse the compound burnt off?, what i cant understand is why it happend, i hadnt added any new hardware, or made any major changes in terms of layout since i built it a last year!

[blobrana]

Yea cooling is paramount.

Try tweaking the bus speed down slightly,
and lower the Cas rating on the memory.

(in the bios you could even tweak/check the chip voltage?)

That may help the onboard chip cache.

But if its faulty then its faulty...

i don`t know enough of linux to be able to do a software workaround.
With XP for example you could change the default L2 chip cache

HKEY_LOCAL_MACHINE\System\CurrentControlSet\Contro l\Session Manager
Key: Memory Management
Name: SecondLevelDataCache
Type: REG_DWORD
Value: 0x00000000 256K L2 cache ,Value: 0x00000200 512K L2cache ,Value: 0x00000400 1M L2 cache, Value: 0x00000800 2M L2 cache ...


good luck.

[scafe]

thanx for the advice guys il give ur suggestions a try!

BTW worst case senario and the L1 cache is fried, what do i replace, i presume the processor?

[Robert17]

Those XP2200+ are pretty cheap now anyway *gulps* I paid £129 for mine ages ago lol

[kd7ota]

Robert17 wrote:
Those XP2200+ are pretty cheap now anyway *gulps* I paid £129 for mine ages ago lol

Yes!

But the best thing ever, is that one of my friends gave me his 2800+ AMD XP Barton CPU. :-D free of cost! Works great and still is on my PC.

One thing I sure did notice that even though it has an awsome cooler and heatsink, UGH, it tends to stay at 50c no matter what. :-? :-P

[Floid]

If you're 'lucky' the CPU could be fine but the motherboard could've suffered "Taiwanese capacitor syndrome," which can cause this sort of subtle and confusing excitement if something goes in the circuitry for the mainboard's voltage regulation. (Anything look subtly bulged or vented?) There's a slim chance this could also have permafried the CPU in the first place, if you want an excuse to lean towards the "test with new mainboard first" methodology.

Another possibility is dust slowly accumulating (or lubricant wearing away) and slowing the CPU fan, while at the same time dulling the effectiveness of the heatsink. If you want to bother caring, you can write down the RPMs you get off the sensor when the thing's new, and watch for signs of degradation... when I noticed my old Mini-Super-Orb (or whatever the beercan-sized dual-fan Thermaltake for Socket A was) had slowed by 1,000 RPM, I knew it was time to ditch the beast and get something more quiet.

Back to the mainboard: If your northbridge requires active cooling (or gets really freakin' hot, even if they didn't slap a fan on it), the same could've happened there, or you could've jammed the fan with a cable and not noticed all this time.

Unless you've got a spare system, or a friend's spare system to try your CPU in, it's probably less work to change the CPU first, and I wouldn't consider the risk of permatoasting the new one very high... on the other hand, since you've got two variables on the mainboard to suspect, it's worth bothering a friend for the $0 test... or finding a hole-in-the-wall shop who'll pop your chip into an open system and let you run a round of memtest for $5 or the purchase of a cable or a stack of blanks. (Hint: The one with cables everywhere, and dirty windows, on the side of town nobody bothers going to; bonus points if they can understand what you want to do on the phone.)

[Floid]

Third possibility: If the chip tests good, haul in your mainboard and let them give it a shot with a different power supply. Those go, too, and if it tests good, you can get a new supply then and there. (Might as well drag the whole system along in the back of your car, with the CPU already out for the first test.)