When does a HDD get corrupted while OCing

[Quad Master]

Hi guys

I have seen the HDD getting corrupted when your PC is overclocked to extremes.

What is the reason behind this and is there any way to prevent it.

Is the PCI/AGP bus lock that prevents the HDD from corruption.

[Aphex_Tom_9]

I think it's a combination of RAM instability and abrubt shutdowns due to power fluctuations, ive broken 2 windows isntalls with pushing too hard.

[Quad Master]

So will a better powersupply help solve this sort of problems.

Also is there a PCI/AGP lock that is to be applied while ocing using the NF4 motherboards.

Note: I dont have a Nf4 motherboard so cant check in BIOS.

[Aphex_Tom_9]

Also is there a PCI/AGP lock that is to be applied while ocing using the NF4 motherboards.

Note: I dont have a Nf4 motherboard so cant check in BIOS.
I think all NF4 boards are automatically locked.

[Quad Master]

Ok thats what i wanted to confirm , I had heard about auto agp/pci locks.

Thanks for the info Aphex_Tom_9

[uOpt]

Every kind of memory corruption can and will corrupt the contents of your harddrives.

Short version:

because the filesystem buffer cache in the OS (the one in RAM, not the one on the harddrive) will be screwed up and it will write wrong blocks at the wrong places on the disk, because the information where a block belongs is changed. Then you write -say- a data block for file "foo" on the location where the allocation blocks for "bar" live, assigning a bunch of random blocks to file "bar", losing the original blocks.

Long version:

http://cracauer-forum.cons.org/forum/ecc.html

[Ugly n Grey]

you should also turn of write caching while running unstable speeds...

[uOpt]

you should also turn of write caching while running unstable speeds...

That is the cache in the harddrive, which isn't affected in first place.

I don't think you can turn off the filesystem buffer cache in RAM in any modern OS.

[Ugly n Grey]

Martin, you're wrong...

caching writes to the drive while the computer is unstable often leads to drive corruption, it leaves unwritten data in the drives cache when the computer crashes. The cache on the HD is indeed affected by an unstable system.

UnG

[uOpt]

Martin, you're wrong...

caching writes to the drive while the computer is unstable often leads to drive corruption, it leaves unwritten data in the drives cache when the computer crashes. The cache on the HD is indeed affected by an unstable system.

UnG

Uh, OK, I see what you are getting at.

However, the cache on the harddrive will be flushed even if the software crashes, the harddrive is autonomous. The risk there is power fail which supposedly is handled by the drive using its rotation energy to flush the cache but I doubt this works as advertised.

In any case, the filesystem buffer cache in the OS is the far bigger risk for filesystem integrety when overclocking, because it can be much bigger than the drive cache and it directly suffers from memory corruption when memory or CPU screw up.

[Quad Master]

I really didnt knew all this Thanks a lot for ur inputs Martin & Ugly n Grey.

[uOpt]

Basically, you never want your "production" OS installation harddrive in the machine when verifying an overclock.

It is less harmful if you use Prime95 to verify because there is few disk activity. But disk activity is never zero in Windows. In Unix you can mount all your filesystems readonly which would be mostly safe but not terriby practical either. I have a scratch harddrive to verify overclocks, and it wiped out the contents real good a couple of times.