Overclocking corrupting hard drive

[ric449]

How is this possible with modern motherboards? The hard drive and everything else used to be connected to the PCI bus, which is usually overclocked when you are overclocking. But since nothing except PCI devices are on the PCI bus anymore why would overclocking the FSB corrupt the hard drive data?

[Cricket]

The IDE controllers are still part of the PCI bus (to communicate with the CPU) and since upping the FSB can speed up the PCI bus (unless you can lock it), there is the chance that the accelerated bus can cause data corruption.

:) Cricket

[ric449]

Really? What else is part of the PCI bus? I thought everything had moved off that now because it isn't suited to the faster speed of ATA nowadays.

[ric449]

This explains what I learnt: http://www-tus.csx.cam.ac.uk/techlink/workshops/2001-11-21_inside_the_pc/sld022.htm

*edit* Just found another example which makes it more clear and detailed, as you can see the only place for the PCI bus is with the PCI slots: http://www.anandtech.com/showimage.html?u=http://images.anandtech.com/old/chipsets/intel/815/blockdiag.gif (I know its a 815 chipset, quite old, but the same rules apply.)

[Cricket]

I got my info from PC Guide, but the info may be old...but the fact is the IDE controller still has to communicate with the CPU through the southbridge and system bus...it's the system bus that gets speeded up when you're overclocking the FSB and this is what's causing the data corruption. I don't know exactly how it happens, but I have seen instances where the data on the hard drive has gotten corrupted when the system is overclocked too high.

:) Cricket

[ric449]

What exactly do you mean by the system bus? Do you mean the connection between the northbridge and southbridge?

[Cricket]

Originally posted by ric449
What exactly do you mean by the system bus? Do you mean the connection between the northbridge and southbridge? In your link to the diagram at Anandtech, up at the top of the page under the CPU..."System Bus (66/100/133 MHz)".

:) Cricket

[ric449]

Ah, the FSB. Well, when overclocking, people raise the FSB, which is the connection between the CPU, northbridge, and AGP. That explains why the AGP, CPU, and memory are forced to run at different speeds, but the PCI? That is on the southbridge now, unless when raising the FSB somehow the bus between the southbridge components is raised too.

[gosu mike]

hard drive corruption is due to unstable memory when overclocking. My asus had pci lock and i still would get corrupt data because my memory was pushed too hard. later discovered running it at 2.8v instead of 2.6 led to no more data corruption.

[ric449]

Aha, so because the memory has the corrupted data, the corrupted data is written to the hard drive?

[Cricket]

Originally posted by ric449
Ah, the FSB. Well, when overclocking, people raise the FSB, which is the connection between the CPU, northbridge, and AGP. That explains why the AGP, CPU, and memory are forced to run at different speeds, but the PCI? That is on the southbridge now, unless when raising the FSB somehow the bus between the southbridge components is raised too. I believe the PCI bus does get affected when the FSB is raised...or why would they bother to put the option to lock the PCI at 33MHz on some overclocking motherboards? I think everything on the motherboard is affected when the FSB is raised, not just the connections on the northbridge.

:) Cricket

[glc]

Whenever you run *anything* out of spec, there's a chance for data corruption, just like when you have a component failure or flakiness. I can't count the times I've seen a hard drive get scrambled from a flaky ram module with everything running stock. Everything has to interact properly or you stand a chance of getting errors. Smart overclockers either have a spare hard drive for experimenting - or have an image ready to restore.