Can FSB overclocking damage SATA drives?

Uhm, increasing FSB imposes an increased frequency for _all_ devices which talks on the FSB, and SATA and PATA must surely be such devices.
I know back 10 years ago when I overclocked a PC, there was a concern for disk drives being damaged.

How can you be so sure, that a disk drive will not be damaged if you require it to talk much faster than it's designed to? Just like with memory?

These days SATA tends to chatter along the PCI-E, which is why you don't overclock it at all / that high.

I believe that increasing FSB may speed up data traffic and cause CRC errors during writing. But I´m not sure of that. I never had problems with high FSB with my PATA Harddrive but after adding a SATA one I had data corruption until I decreased FSB a little.

So there _is_ an issue, if what's being written can actually be incorrect. That's as bad as a "damaged" drive to me...

GrimReaperGuy, do you know if SATA talks on PCI-E on intels chipsets? 965 or 975 for instance? I haven't been able to find info on this... But if it's true, its frequency can be locked usually.

That was back when there where no AGP/PCI bus locks on mobo's. Nowadays, the IO buses are lockable, and therefore not a problem.

mpilchfamily, your answers are vague, incongruous and missleading. First you say, there's no problem. "Only" CPU and memory are affected. If so, I'd like to overclock quite much (because I'll have very good cooling).
Now you say I'd go too far, and have to take things one step at a time etc. How do you recognize crc errors? These can occur long after your prime/pi-stress-tests have been successful.
That's why I want EXACT answers.
1Tanker is close here;

Are you sure SATA talks on a bus that is lockable? I've read that PCI/AGP/PCI-E can be locked on some mobos, but I haven't been able to find if SATA can be locked (or if sata talks on PCI-E for instance, and therefore is lockable).

Thanks for all replies so far!

with proper mobo's you can lock (or should i say run async) all that stuff, no issue there.

Hope this helps.

On my FSB133 2.66, I get up to FSB149 with SATA, and up to FSB156 on USB PATA HDD with an adapter.

Also, my onboard RAID controller will NOT recognise my seagate @ FSB150+ (it displays a bunch of ASCII gibberish instead of the model/serial #, or doesn't find it at all). OCing seemingly hasn't damaged the drives at all (I have the system running pretty much 24/7 at a stable 147, no stresses) nor have I *noticed* CRC issues (as you said those can take a while to come out).

FSB runs at 66MHz, PCI @ 33, which makes a 2:1 ratio. Pushing the FSB up to 75MHz also pushes the PCI bus up to 37.5MHz, pushing anything attached to the bus up to 37.5 as well. The same also holds true for the AGP and PCI-E buses. If the bus can be locked down, then pushing the FSB to 75 leaves the PCI bus at 33, leaving the speed of anything attached to that bus at 33. AFAIK locking the buses isn't a new thing or restricted to PCI-E.

Not very many boards allow locking of the buses.

Thanks for your replies!
The important thing then, is to find out what mobos actually support locking PCI/PCI-E/SATA frequencies... I'll need at least 4 satas, and was thinking of the AB9 Pro, but I can go for anything actually. I'd prefer intel chipset over nvidia. Any idea of a not too expensive mobo wich supports locking, and has at least 4 satas?

Most 965P chipset boards will do you fine in both overclocking and in the SATA and IO lock departments. Personally, I recommend the GA-965P-DS3 from Gigabyte for around $140.

Sorry mate, but you are wrong. It does affect the data being written on the disk, causing huge data corruption to the effect that you have to reformat the drive. Of course it cannot damage the drive itself, but i am sure you don't like formating your hard drive and installing all your files every 5 mins!

Something is seriously wrong with your OC if that happens. I've run an X6800 at 4.6GHz for days on end without corruption errors.

It should not damage the drives.

However, if PCI / SATA clocks are out of specification, even slightly, it can cause massive data corruption.

Since the PCI clock is often calculated by dividing the base FSB

333.333 / 10 = 33.333 MHz
266.666 / 8 = 33.333 MHz
but
299 / 8 = 37.375 MHz

and the SATA clock is often generated by a ratio from the PCI clock.

If the SATA clock is out by more than 8% expect data corruption to occur.
If it is out by more than 4% data corruption *might* occur.

Check the mainboard manual, or download the PDF from MSI's website.

Look for options in BIOS called "PCI Lock" or "PCI Bus" and lock the setting at 33 instead of Auto (free roaming).

There may also be a PCI Express (base, pre-multiplied) bus speed, lock it at 100 MHz instead of Auto (free roaming).

I have a MSI K9N Neo I can try out tomorrow afternoon / night (well 10 hours from now).

You might want to research this a bit. Most motherboards now have locks on them, but some of the lower end motherboards, at least those a year ago, the locks did not work as advertised, and you would still experience data corruption. So, the answer is yes and no. If you have working locks, where you can lock the PCI-E and PCI-X buses, you will not have a problem, but if the locks are no in bios, or the locks do not work..... your done.

wes

Is it possiable yes it is. I had a WD and just got my RMA a few weeks ago but what I did'nt know at that time the IO on the HD to lock the data transfer to 1.5 mb would have prevented the HD from a meltdown.