Corruption of HD/Data from OC'ing?

[BoboKatt]

I have heard of corruption your data/OS installation while oc'ing on some boards.

Does this stiff apply to "newer" boards? I have an Asus P5K-E (wifi) which is a P35 chipset.
Am I correct in assuming that oc'ing my CPU and RAM DOES NOT affect the speed at which the PCI bus is running? or the SATA ports etc?

Anyway to make sure? I just remember in my AMD x2 days and having a DFI board that specified that some of the SATA ports were linked to the FSB and hence you should not use those if you were oc'ing.

What in my bios do I enable/disable to control this if anything? I mean again, in newer boards... Interl Core2 quads etc, LG775, is this removed and taken care of?

 

[Concillian]

Generally this only happens with rather extreme overclocking that involves extremely high FSBs (550+) in current intel OCing. If you're using air and running normal FSB (<500,) you don't have to worry about this.

 

[Gillbot]

Originally posted by: Concillian
Generally this only happens with rather extreme overclocking that involves extremely high FSBs (550+) in current intel OCing. If you're using air and running normal FSB (<500,) you don't have to worry about this.

I've seen it happen as low as 400fsb. You can't give a general number if you don't know all the hardware that's in a system. All it takes is one sub-par component to introduce error and lead to corruption.

 

[VirtualLarry]

I think that the primary reason for HD corruption during overclocking, besides the CPU being unstable, is RAM, or the northbridge is overheating and causing data corruption.

That being said, you should always do a test install of Windows, perform your overclocking, and then when you get it stable, do a re-format and a fresh install. That's what I do when overclocking a new rig.

 

[Comdrpopnfresh]

I never install programs, let alone an OS if I'm unsure my clock settings are stable. I certainly wouldn't do much overclocking after a stable install if I couldn't lock the pci+pci-e frequencies- which you typically see on older boards or budget ones w/ an IGP and aren't meant for much overclocking anyhow.

If you meant an overclock on an established system corrupting data.... I think it's slightly possible, but not all that likely- if your overclock was giving hdd issues, you'd most likely experience read errors and not be able to load windows @ a point in the OS load when data is mainly being read- so you wouldn't get corruption that way. If you did experience corruption for some odd reason, after being able to boot to the desktop, there's a high likelyhood that resetting the clocks and running a dskchk would resolve much of it- the dskchk may not even be necessary.

I will tell you this: on my 3rd and final s939 asus a8n32-sli deluxe, I couldn't do a clean install to save my life. Maybe it was the fact that the board was dying, or that I installed SP3 for XP- but I got errors like crazy- even when I ran dskchk from my xp install disk- it flagged random portions of my drive- different ones on each run too- the board was definitely dead for all intents and purposes. It'd run a backtrack livecd though- thats how I knew something was screwy.

in your bios setting, usually under something along the lines of 'manual settings' or some such, you want to lock the pci clock to 33mhz (or is is 133?.... maybe it's 133 by a bus multiplier applied to the 33mhz). In either case, if you reset the bios to defaults, you want to keep the pci clock @ that reported setting. And keep the pci-e clock @ 100mhz. Not only can you screw with your integrated periphs like sound+igp or any extra port controllers (such as a marvell ahci sata controller), but I remember reading that increasing the pci-e bus speed, through either the multiplier or base frequency (stock=100mhz) can fry gfx cards if you go too high. And in the 7-8th gen nvidia days, there wasn't any performance gain from increasing pci-e speeds.

in short: leave pci @ stock, whether is is 33 or 133, and leave pci-e @ 100mhz and don't touch the multiplier for pci-e bus either. Nowadays, if you can't see the pci bus settings, it's typically locked and unavailable to change. And if in doubt, run your drives off of the primary sata ports provided by the nb/sb- not the additional one's provided by a 3rd party controller such as marvell.

 

[aigomorla]

many ways your hd can go corrupt.

I get it just from swaping processors too many times and having failed overclock.

When you overclock, your tweeking everything, ram, cpu, and motherboard. If one of those is half working, it can send bad info and corrupt your hard drive.

You dont need 500FSB, ive done it at 400fsb tons of times.

 

[Gillbot]

You can get data corruption over time too. You don't need to have a "stable" oc then install. I've had rigs that ran for months and months then all of a sudden, BAM. ALWAYS keep a backup.

 

[n7]

Unstable RAM is the worst IMO, though obviously it can happen with other factors...

 

[DavidK21770]

An unstable OC will eventually cause bad data to be written to the disk which, if in a critical file, will corrupt windows or important programs. This is why I always run a full set of test programs for a long time before trusting an OC. (prime95>12 hours, OCCT>5 hours, >60 cycles of linpak, the 3D Mark series). While testing OCs, I make a full backup of my system and do a restore when I'm done -- I use drive snapshot which will make differential backups that only take a few minutes to make, combined with either bart-pe or the ultimate boot cd for the restore. I do all of this because I've been burned more that a few times and it's a real pain to recover from when you don't know how far back to go...

 

[walk2k]

I got HDD corruption trying to overclock my old AMD/939 system. I had to unplug the HDD when doing any overclocking. I never could get it to OC more than 4% either, so I gave up and ran at stock (actually 1% oc).

If you are worried, unplug your HDD and boot from a CD with test tools, ie Ultimate Boot CD (UBCD).