IB owners, read this

[MPiland]

Well from experiance . It looks like your M/b is at fault and not your CPU. I will keep up with this thread until you reinstall the new cpu.

This wasn't my chip, it was the guy from that thread. He already RMA'd the chip and is waiting on the new one to show up. Why would Intel tell him the chip is bad if it's the motherboard? Either way, something to watch for.

 

[Ferzerp]

One thing of note, if your OC appears stable but you are getting parity errors, see if you can increase the frequency with which your motherboard samples load and changes the voltage to compensate. I was able to drop my offset voltage by a few notches by putting this on the max rate. I didn't need the voltage I was supplying, but in order to ensure that it always had enough, I had to give it slightly too much. By making it repond to the load faster, I was able to drop it down a little.

 

[MacGyverSG1]

One thing of note, if your OC appears stable but you are getting parity errors, see if you can increase the frequency with which your motherboard samples load and changes the voltage to compensate. I was able to drop my offset voltage by a few notches by putting this on the max rate. I didn't need the voltage I was supplying, but in order to ensure that it always had enough, I had to give it slightly too much. By making it repond to the load faster, I was able to drop it down a little.

I remember reading about breaking in a CPU when overclocking. Instead of going from default speed to max OC in one day, only increase the speed by steps. Run the CPU for a week for each step, making sure 100% stability before going to the next step. I guess it would let the CPU get used to the voltage increase gradually. Maybe it made more sense back in the day, but not now.

 

[Emo]

I found that I had WHEA 19 errors when I run 3D Mark 11 and Prime 95, just had to bump the voltage a couple of notches and the errors disappeared. Oddly IBT and AIDA64 wouldn't trigger the errors. If I hadn't checked the event viewer I would have thought that I was 100% stable since I never saw any BSODs.

 

[Idontcare]

I remember reading about breaking in a CPU when overclocking. Instead of going from default speed to max OC in one day, only increase the speed by steps. Run the CPU for a week for each step, making sure 100% stability before going to the next step. I guess it would let the CPU get used to the voltage increase gradually. Maybe it made more sense back in the day, but not now.

Yeah that school of thought doesn't really jive well with the basic device physics at work inside a CMOS chip like the cpu. I can see where such ideas can be generated, it seems like something that ought to make sense, but that just isn't really how voltage and device physics work.

There really is no "let the CPU get used to the voltage" factor involved in solid state physics.

I found that I had WHEA 19 errors when I run 3D Mark 11 and Prime 95, just had to bump the voltage a couple of notches and the errors disappeared. Oddly IBT and AIDA64 wouldn't trigger the errors. If I hadn't checked the event viewer I would have thought that I was 100% stable since I never saw any BSODs.

Remember there are over 700 different instructions that your CPU can be asked to execute. Any given stress tester is only testing a really really small subset of the total instruction set for stability and consistency.

It should come as no surprise that when you run a different program, which is going to execute different instructions (it is a different program afterall) that you stand a pretty good chance at discovering some of those now used instructions aren't actually stable with your current OC.

This is why it is imperative to employ a bevy of stress testers when attempting to isolate stability issues as well as confirming stability in the actual apps you use with the computer.