8GB RAM Causing Windows Files To Corrupt?

[Overslop]

Hey guys, I have an issue where occasionally a Windows file will become corrupt - making the OS unusable until I repair/replace it.

It only happens very sporadically - only a few times and always months apart.

I'm trying to diagnose what is causing it and have narrowed it down to either being - 8GB RAM putting too much strain on the northbridge, my overclocking causing the instability, or my hard drive is faulty.

Hoping some of you memory experts might be able to help out.

 

[AndyOcUK]

- 8GB RAM putting too much strain on the northbridge, my overclocking causing the instability, or my hard drive is faulty.

You can rule the first two of those highly likely candidates by putting everything back to stock speeds, apart from the extra bit of NB volts to cope with 8GB.

You can test the HDD by downloading the diagnostic software from your HDD manufacturer's website.

RAM can be tested using MEMTEST.

 

[Overslop]

You can rule the first two of those highly likely candidates by putting everything back to stock speeds, apart from the extra bit of NB volts to cope with 8GB.

You can test the HDD by downloading the diagnostic software from your HDD manufacturer's website.

RAM can be tested using MEMTEST.

The thing is though, that I only get this issue very rarely. The rest of the time, the PC seems to be working fine (except an issue with the sound card I have which I believe is unrelated). So I would only find out after many months if the stock clocks were helping. And even if it was still working, it could be conincidence. I really wouldn't want to run that long on stock clocks unnecessarily. I bought the system to be overclocked, and performance without overclocking is significantly lower.

The same principles apply with the RAM if I removed 2 sticks. Although the performance difference is very minor. I would happily sell the sticks if I thought that it would solve the problem, but as I'm unsure if it would, I could be downgrading my PC without wanting to.

Will download that diagnostic software and see what it comes up with.

 

[AndyOcUK]

The thing is though, that I only get this issue very rarely. The rest of the time, the PC seems to be working fine (except an issue with the sound card I have which I believe is unrelated). So I would only find out after many months if the stock clocks were helping. And even if it was still working, it could be conincidence. And the same would apply with the RAM if I removed 2 sticks.

lol - short answer to that is 'Tough'.

Thats what you need to do to properly isolate and diagnose the fault.

Either test properly - find the fault - fix the fault.

Or

Put up with it.

There are no easy short cuts with overclocking m8

I'd test the drive with everything back at stock first - be sure to back up BEFORE YOU DO THIS!!!!!!!!!!!!!! (Don't say I didn't warn you)

 

[Overslop]

Guess I'll have to put up with it then. Going back to stock settings could only "prove" to be the cause after half a year or so, and even then wouldn't be definitive.

Hopefully that hard drive diagnostic finds the culprit...

 

[AndyOcUK]

Lol - its a pain I know.

Could even be Windows - A format & re-install with all the latest drivers and updates would rule out software issues but you could still be waiting ages for the fault to manifest itself again

 

[hernaldo]

I had some annoying thing like this happen where the odd byte would get corrupted in files every few weeks or so. Turned out it was the memory, I did a 2 hour test in memtest when it happend with no problems...

Thought I'll just put up with it until someone recommended doing a full 24 hour test in memtest.

low and behold 7 hours in I got one error, then an hour later another one...then there was less and less time between the errors.

Long story short, do a memtest run of 24 hours. Then you really will know if your memory has cooked it's goose. Also don't do it at stock otherwise you will be none the wiser as to what the problem really was if your memory runs fine at stock. You will also be able to work down to a clock that doesn't give you this annoying crap.

 

[Overslop]

I had some annoying thing like this happen where the odd byte would get corrupted in files every few weeks or so. Turned out it was the memory, I did a 2 hour test in memtest when it happend with no problems...

Thought I'll just put up with it until someone recommended doing a full 24 hour test in memtest.

low and behold 7 hours in I got one error, then an hour later another one...then there was less and less time between the errors.

Long story short, do a memtest run of 24 hours. Then you really will know if your memory has cooked it's goose. Also don't do it at stock otherwise you will be none the wiser as to what the problem really was if your memory runs fine at stock. You will also be able to work down to a clock that doesn't give you this annoying crap.

Thanks for the suggestion, will give it a go.

 

[Corasik]

I've had a similar situation with overclock testing, using Prime95, I had a system which would fail after exactly 16 hours and 12 minutes (+/- a few seconds). It was always the same test which failed. Reducing my overclock by 1mhz (on the FSB) was enough to resolve the problem, with no prime95 errors even after 24 hours of continuous testing.

Aborting the Prime95 test before it reached the fail point, and starting again, it would never fail, as it took that computer 16 hours to reach the test that failed. If it was possible to force prime95 directly onto the test that failed (other than waiting 16 hours), it would have failed right away. It was the test, not the length of time the PC was running for.