Linux OCing versus Windows OCing?

[996GT2]

Since I've done quite a bit of OCing in Windows and determined my Optimal OC to be 2.66 GHz on stock voltage, I decided to give Linux a go. I was too lazy to install so I just used a Knoppix Live CD I had laying around. In BIOS, my comp can boot at 2.9GHz (1.35 or 1.4V) but once I get into Windows it gives me a BSOD. Windows needs around 1.45-1.5V to be stable at anywhere near 3 GHz on my CPU. So I tried knoppix, booting at 1.4V and 3 GHz. *Don't know about stability yet, haven't primed it*

Here's a screenshot using Wine to emulate CPU-Z; it looks a bit buggy, but come on, it's Linux emulating a Windows program, you can't expect much more than that:
http://img230.imageshack.us/img230/3419/dsc02110ru7.jpg

Another pic:
http://img111.imageshack.us/img111/2108/dsc02111ew6.jpg

Why is Linux more stable for OCing?

 

[jose]

Don't know really, but I've experienced the same thing in the past.

 

[dexvx]

Try Prime95 in Linux and Windows, and they will top around the same with same voltage.

 

[meatfestival]

Just because it boots up doesn't mean it's stable.

 

[996GT2]

well yeah, but at least it can boot up and run for an hour without crashes, whereas windows doesnt even get to its login screen

 

[aka1nas]

Windows is probably hitting the CPU a bit harder on boot up long enough for it to generate an error.

My other theory could be that the version of Linux you are running isn't supporting all the features on the chip(think like SSE 1-3) and some circuitry is sitting idle and not generating as much heat. This idea hs been bugging me about x86-64, though I haven't had the opportunity to test it. I am willing to bet that you won't be able to OC quote as well with a 64-bit OS as there will be parts of the chip(the extra GPRs, longer registers) that would have been idle in 32-bit mode that would now be stressing the chip further.

 

[xtknight]

Maybe because Linux is running off a CD and not using the chipset's IDE controller? Either that, or Linux is a lot more solid/able to ignore some errors. I'd wager the latter. Windows already knows the capabilities of your system, while the new Linux install/LiveCD is just discovering them if you want to put it that way. If a function fails on the new Linux install, it will just assume your system does not support it, not that an OC mangled it. Far-fetched? It's a theory.

 

[Kalessian]

Yes, I've noticed this, too.

 

[IamDavid]

lol, thought it was just me...

 

[996GT2]

Hmmm, maybe I'll have to actually install linux on my HDD and test sometime

 

[TanisHalfElven]

try ubuntu. will take 15 min-20 min to install.

 

[Fox5]

Vista is more stable than XP also, I guess the more memory protection is in place, the more stable the OS is with overclocking. (64 bit is less stable than 32 bit though)