Bad L2 Cache?

[jcv0085]

Is it possible for a cpu's L2 cache to go bad, but still be able to function under a "light" cpu load? My system is always BSOD'ing and hanging when I play a game or put the cpu under a heavy load. I've memtested my ram with no errors and tried swapping ram with the same results. I've also tested my power supply and voltages are within acceptable ranges. Prime95 testing small FFTs (little to no ram tested) causes system hangs or bsod's within seconds.

 

[jcv0085]

I guess I should post my specs too.

Rosewill Xtreme Series RX750-D-B 750W
ASUS M4A79XTD EVO
Phenom II X4 965 rev. C2
A-DATA Gaming Series 2 x 2GB DDR3 1600
500GB WD Caviar Blue
ASUS Radeon HD5850

 

[warmon6]

hmm... i never heard of bad L2 cache before.

I'm gonna guess it's probably something else as you say it when you stress the computer and if the L2 cache is bad, you probably wouldn't be able to boot into windows without problems, Yet alone using it.

Whats you're cpu vcore? If this is to low, it could explain all you're problems.

most of the time when a cpu vcore is to low for the GHz of cpu, people will get problems like the one your having. Restarting under high stress like games, prime95, ect.

Most of the time though, this only happens when overclocking but it's not impossible for the bios to have wrong setting.

 

[jcv0085]

I have my vcore @ stock right now (1.45v) I've upped it to 1.5v with the same results and don't really want to go much further than that because temps start getting pretty scary beyond 1.5v. You can supposedly disable L2 cache in some systems but my bios doesn't have the option for it. There is a huge performance penalty with a disabled L2 cache, though, but would be a great feature to have for troubleshooting.

 

[warmon6]

I have my vcore @ stock right now (1.45v) I've upped it to 1.5v with the same results and don't really want to go much further than that because temps start getting pretty scary beyond 1.5v. You can supposedly disable L2 cache in some systems but my bios doesn't have the option for it. There is a huge performance penalty with a disabled L2 cache, though, but would be a great feature to have for troubleshooting.

Hmm.... very strange. This is gonna be a hard thing to figure out.

Well guess it's time to start process of elimination.

Are you're drivers on all your hardware up to date? If not, try doing that. Maybe it will clear up a drive conflict problem thats causing it.

Even though your ram came up fine after testing, Is the memory voltage set for the ram? It is probably not the problem but the voltage requirement for your ram is a little higher than normal ddr3 ram. 1.5v is standard while your ram is 1.55 to 1.75.

http://www.adata.com.tw/?action=product_specification&cid=5&piid=91

Last thing i can think (at the moment) of is maybe the PSU is not giving enough wattage to the system or maybe not enough amps/even amount of amps on the 12v rails. Which there maybe a chance of that.

while cpu aren't known for not creating amps issues, the multiple 12v rails supping power to the entire system (mainly the graphic card and cpu combined) maybe proving to much for all the rails. It a semi problem where having multiple 12v rails can create issues. This is where a lot of guys say get single 12v rail PSU if you were builting a new computer.

If this is the issue, try removing the gpu. If the problem clears up, maybe you'll need to look for a different psu.



Im not say any of these are the problem, these are just guesses that could be causing the problem.

 

[jcv0085]

I recently flashed my bios with the latest version from the ASUS site and reformatted my HDD with the latest drivers for everything. My PSU has been tested with a PSU tester and was tested as 'good'. I'm having a buddy come over later with a multimeter so I can test voltages under load. The two PSU connectors powering my video card are from two seperate rails and have tried my 8800GTS on the system with the same results. I have tested with various voltages for my ram to no avail.

I know that in most instances cpu's either work or they don't. The only reason I'm pointing fingers at mine is because I've troubleshot everything else and Prime95 fails within seconds, testing small FFT's, and I can't do anything cpu-intensive. Watching videos is probably the most I can throw at my computer right now without it crashing.

 

[warmon6]

I recently flashed my bios with the latest version from the ASUS site and reformatted my HDD with the latest drivers for everything. My PSU has been tested with a PSU tester and was tested as 'good'. I'm having a buddy come over later with a multimeter so I can test voltages under load. The two PSU connectors powering my video card are from two seperate rails and have tried my 8800GTS on the system with the same results. I have tested with various voltages for my ram to no avail.

I know that in most instances cpu's either work or they don't. The only reason I'm pointing fingers at mine is because I've troubleshot everything else and Prime95 fails within seconds, testing small FFT's, and I can't do anything cpu-intensive. Watching videos is probably the most I can throw at my computer right now without it crashing.

Well, this is the second time scene i been on toms forums where i been stumped at a problem (and i can guess why nobody else has pitched in to help.)

and it not easy for me to be stumped on hardware issues.....

Even though testing say PSU is fine. it just for some reason and i dont know why, im still having that nagging feeling it is PSU. There just been time where other peoples strange problems cant be solved or doesn't look like the PSU would be the fault but yet when a different PSU was used, those problems just vanished. It amaze even experts like me from time to time when a problem doesn't even remotely look like PSU problem.

I dont know. that just the feeling i have.

Probably the reason why to me it just doesn't sound like a L1/2/3 cache problem though is because the OS is using them when the cpu is idle and at load (it treats it like uber fast/frequently used ram), which would normally mean it should of BSOD a long time ago if there a issue with the cache. It just the problem your describing sounds like to me a power issue giving the nature of the problem.

Just wish there was someone else here helping trying to figure out whats going on that can think what i haven't though of yet. There more talented minds here than me but yet there not helping out with this complex problem.

Maybe you can try:

using a friends cpu on you're motherboard (providing they have an AM3 cpu),

if one of your friend's as an AM2, AM2+, or AM3 140W montherboard that can use your cpu, try that. (providing you have any friends that have AMD system and can run a 140w cpu)

and/or try someone else PSU thats powerful enough to run on your system.

*just be sure to have plenty of cleaners to clean the cpu of thermal compound off and new thearmal compound*

See if the results are replicated with a different setups.

If thats not possible then ,as much i hate to say this (because i dont like to throw money at a problem without knowing where it source is), you may have to buy a new cpu and see if the problem persist.

If it doesn't apper, then Horray! problem solved.

If not, then it something else.


This certainly a strange and complex problem that going keep bugging me until it's solved. [:isamuelson:8]

 

[jcv0085]

hmmm... I did accidentally drop my cpu when I was first putting my rig together. It was sitting on the work bench and it fell, bounced off a PC case and hit the floor. A half row of pins were bent but they were straightened out, using a credit card.

My computer was running fine for about a month before it started acting up. The problem has been getting worse over time. I don't know if it could have anything to do with my issue, but I thought I'd throw that out there.

 

[jcv0085]

Well, I just came back from a buddy's house to test my power supply again with a more accurate multimeter (The one I used before was analog). I tested it under load and all the voltages were within acceptable range again.

The funny thing is that my system didn't freeze under load, running the same prime95 test. The problem is temperamental. I don't know if I'll ever figure out what's wrong with this thing.

 

[warmon6]

Well, I just came back from a buddy's house to test my power supply again with a more accurate multimeter (The one I used before was analog). I tested it under load and all the voltages were within acceptable range again.

The funny thing is that my system didn't freeze under load, running the same prime95 test. The problem is temperamental. I don't know if I'll ever figure out what's wrong with this thing.

Well, thats certainly strange. certainly make it harder for me to figure it out.

Now inquired with more info from you last 2 comments,

For the bent pins Where you able straighten out all the pin or did some of them brake? I doubt that it thats the problem though as the system was acting fine for a while..

Hmmm...... Also thinking of my grandma computer , Is there a chance that you have electrical problems with you're house or (primarily) your neighborhood that you're aware of?

I know it seem an odd random question but been think though out the day, my grandma had has trouble with the neighborhood electricity and it's sometime mess up her computer (between strange, unknown problems to hardware failure) that appear randomly without a real question. Failed MB's, graphic cards drivers go bad, Blow PSU, ect.

The way my dad solve the problem was adding a a UPS (or also called backup power suppy) and problem was finally solve. Only replace the PSU (just 2 months ago) once scene the the UPS was added but that was due to age for the reason why it failed.

Maybe this isn't the solution either but just an idea as it's starting to sound more and more like what i just explained.


Last thing i can think of for know, check you're motherboard and see if there any strange marking, like burn marks. Maybe 1 of VRM or solid capacitors has fried but the rest of them working just fine to give enough power to the cpu but not enough for continuous full load. Unlikely but possible.

well, im gona keep brainstorming for now. One way or another, as long as You and other people (like i) keep on this, the problem will be found and solve. It's only a matter of time.

 

[jcv0085]

It turns out that the integrated memory controller was bad on the cpu. All the problems disappeared when I installed a new cpu. I'm glad this nightmare is over.

 

[warmon6]

It turns out that the integrated memory controller was bad on the cpu. All the problems disappeared when I installed a new cpu. I'm glad this nightmare is over.

Well glad you were able to solve you're problem. To bad i was no help.

 

[cjl]

It turns out that the integrated memory controller was bad on the cpu. All the problems disappeared when I installed a new cpu. I'm glad this nightmare is over.

That doesn't mean it was the IMC, just that it was something about the CPU.