Problems from overclocking?

[MadAsHeck]

Here's how it started. Originally I was getting a hardware lockup while playing games (Bf2, CS, HOMM5). This was on a Opty 165 OC'd to 250 fsb with no increase to Vcore. The whole system would hardlock with static, hissing, and poping noises coming from the speakers (or headset if plugged in). Actually, it would come from both the speakers and the headset because its a 5.1 system and my center and rear speakers were still plugged in. So, I assumed it was the Creative X-fi that was causing the problems. Like I said, the lock ups had been pretty in-frequent at first. Maybe a few times a month. But, then this weekend it happen 3 times in 2 days. Each time while wearing the headset and it really hurt my ears. So, I decided to remove the Xfi and switch to onboard sound.

In order to do that I had to plug in the sound module for my board, DFI N4 SLI-DR Expert, because I never installed it thinking I wouldn't need it. So, while in my case I decided to re-do some cabling.

When all was done I tried to power up the system, and the power button on my case won't work. I used the built in power and reset buttons on the motherboard and got the system running that way. (my power button still doesnt work, but I don't think its related to the problem....)

Once in Windows I installed the new sound drivers (deleted the creative before I removed the card) and rebooted. This is when I started getting BSOD's un-endingly. I could be in Windows not doing anything and would get one. The first time I ran BF2 I got one almost immediately. Try to go online to check what the BSOD meant and would get one.

So, I removed the sound module and rebooted. But, I kept getting them. Finally i was able to look up one of the BSODs I was getting, "IRQL not less or equal". From what I read I saw that it was a hardware failure, usually CPU or memory.

I reset my FSB and my Vid card core and memory speeds and ran both Prime95 (2 instances for my Opty 165) and it passed for hours. I also ran ATI tools overclocking test on my vid card. It also was passing with out any problems.

Next I ran Memtest on the RAM. I let it run 3 full passes (2 something hours) and it passed each time with no errors.

I put my FSB back to 250 and tried Prime95 again and it errors out almost imediately. While I was running the test originally I was using Guardian and MBM to watch my CPU temps, at idle at 200fsb I was around 30C, and just over 40C at load. My vid card was 40C idle, 50C load from the ATI test. The only thing that is showing as hot is my chipset which is showing as 40C/50C according to Guardian.

Is it possible that I was getting relatively no problems with my 250fsb overclock for months, and then all of a sudden after removing the X-fi from my system I started getting BSOD's? When I first got the system and ran it at that I remember running all of the 3dmarks and never had any problems with them.

Anyone have any ideas on how I should proceed next?

 

[HighTest]

EXCLAIMER: I'm not against overclocking and I personally view it as a way to get extra performance when on a budget. That said one problem with OC'ing and regularly overlooked, is electron migration. I'm not as likely to OC very expensive gear do to this problem. Heck OC that

This is a feature that all chip designers have to try and overcome and usually have it handled well at the stock speeds. Esentially, EM over time will cause the substrates and connects in the chip to breakdown over time (or more acurately develop a hole resulting in a short).

Overclocking results in higher levels of EM and will shorten the life of the chip being OC'd. Usually you'll upgrade to a newer product before the failure finally happens, so most don't see this. The exception? Lately new products have been yielding extremely high OC's. The issue is that EM is still happening and at these high OC rates and EM rates, products to fail much quicker. Once EM has damaged the core/chip, returning to stock doesn't resolve the issue and if still functional, downclocking and voltage reduction are usually the only ways to hold onto it in the long term.

That said and out of the way, you shouldn't have damaged the core yet, but it's something to be aware of.

To ensure there are not any issues, I'd do the following:

1. Restore all settings to "stock".

2. Remove all "unecessary hardware", sound cards, the works not needed to boot the PC into your OS.

3. Reinstall with a clean setup.

4. Test the hell out of everything at stock. Prime, memtest, the works to ensure CPU and memory are ok.

5. Re-add "unecessary hardware" and test again. Remember stay at stock!

6. IF everything is golden, start to OC but in smaller increments and test throughly.

If you can return to your original OC results stable, you are fine. If you find that you have to be lower, then something has either degraded (gotta clean out those HSF units) or you've been the victim of EM and will need to run at lower OC or stock to be stable.

 

[MadAsHeck]

"3. Reinstall with a clean setup."

By that, you mean a full format/reinstall?

I see what your saying. I've heard of EM before, but never been witness to its effects.

I guess this project will have to wait until this weekend to start... have to study for a physics test this week.

 

[HighTest]

"3. Reinstall with a clean setup."

By that, you mean a full format/reinstall?

I see what your saying. I've heard of EM before, but never been witness to its effects.

I guess this project will have to wait until this weekend to start... have to study for a physics test this week.

Yep! Format and reinstall. I'm sure I don't need to mention about copying/backing up any information required before performing this step.

If others are reading and don't know why you need to copy/backup, you need to reformat your system when you get home Of course any information you wanted to keep is not my responsibility. I disavow any responsibility for loss of information, system performance and any other horrific issues that result from following any of my suggestions, regardless of serious or in jest.