PDA

View Full Version : Why do HD sometimes corrupt at overclocked settings



oublie
11-17-2006, 03:24 AM
Hi,

Just as the title says, i've had sata drives corrupt on me in the past when overclocking to higher fsb's etc but rarely had issues with ide drives. Why does this happen? surely on windows load there can't be that much written to the drive? an explanation would be appreciated also would using a flash drive or ram drive (the ddr ram pci drives) be a viable method to avoid this problem?

Serra
11-17-2006, 09:20 AM
If by hard drive corruption you mean data corruption on the hard drive, the answer's not that amazing and exists on all types of storage media (ie. SATA/IDE/Flash/etc).

When you boot into Windows there are a number of files that are checked for consistency and re-set if it's found necessary and some values are set to new values (ie. your last known good configuration settings). It goes without saying that if you're overclocked too high and the data gets corrupted, you'll get corrupted data written to your hard drive for those values. In addition to simply corrupted values getting written to the hard drive for things that you would expect like that, if your OC is too high there's also the possibility that the CPU will tell the hard drive controller to write to the wrong part of your hard drive, resulting in random losses.

Other issues that were a problem on some NF2 (or was it NF3?) boards included a few boards having their SATA frequency tied to the FSB frequency, which resulted in a few people getting corrupted information on just about any OC... but that's fairly rare and *shouldn't* be a problem on any board produced in the last few years or ever again.

oublie
11-20-2006, 01:34 AM
Thanks Serra that make sense.

uOpt
11-20-2006, 08:09 PM
On a number of NForce boards, two of the 4 SATA ports are supposed to be unlocked and hence you screw them up when overclocking the board. I haven't heard an explanation why this would be, but I have seen that claim several times.

What board are we talking about here?

ziddey
11-20-2006, 09:02 PM
In modern motherbaords, there's sometimes an issue of the pci-e lock not working properly. Since sata runs off pci-e in new boards I believe, that would be reason for the data corruption

qdemn7
11-21-2006, 02:54 AM
Best solution to avoid worrying about the problem is a good drive imaging program. I personally use and recommend Acronis True Image. (http://www.acronis.com/)

oublie
11-21-2006, 04:27 AM
what if i used a high speed gigaram (or whatever it is called) running off the pci bus - you know the hard drive build using ddr ram, would this make a difference as its read write speeds are so much quicker and the pci frequency can be locked or do you think the same problem would be there.

uOpt
11-21-2006, 05:13 AM
Best solution to avoid worrying about the problem is a good drive imaging program. I personally use and recommend Acronis True Image. (http://www.acronis.com/)

How do you decide when is the time to do a new image?

It silently corrupts files that you haven't even visited in that session, so you'll end up overwriting a good mirror with a bad one. Obviously go-back and similar programs do nothing in this situation.

I still want to know what board we are talking here.

uOpt
11-21-2006, 05:15 AM
what if i used a high speed gigaram (or whatever it is called) running off the pci bus - you know the hard drive build using ddr ram, would this make a difference as its read write speeds are so much quicker and the pci frequency can be locked or do you think the same problem would be there.

Using any PCI or PCIe controller would solve this problem.

Howerver, since other people are not seeing this problem (I run an overclocked A8N5X with heavy disk load on all SATA ports) you should just toss your mainboard (after checking BIOS settings).

So what board?

EsaT
11-21-2006, 07:20 AM
what if i used a high speed gigaram (or whatever it is called) running off the pci bus - you know the hard drive build using ddr ram, would this make a difference as its read write speeds are so much quickerThat wouldn't help.
It's question of PCI/PCI-e frequency, hard drive controller is connected to that and unless frequency is clocked it can rise enough causing controller to "drop from ride" and start writing erroneuous data to HD.


Other possible source is that memory can't handle frequency/latencies, in PC all data is generally transferred through RAM which can cause corruption of it.
That's why using Memtest for testing stable operation of memory is important when overclocking it.


Third option is again motherboard gradually "cooking up". Friend had one computer whose motherboard did that, starting from random unstability untill in the end you couldn't get through fresh install of XP before data on HD got corrupted.