BCLK vs CPU Ratio, 6th core failing, and does this seem stable?

[Deus Gladiorum]

Hey, I'm looking for some clarification on a few things. I've overclocked before and I was troubleshooting some potential gaming instability earlier by putting myself at optimized defaults in the BIOS. Usually I'm running a 4.6 GHz Overclock, though it's a bit unstable. My 6th core/worker had stopped working about 10-12 hours in during a Prime95 blend test. However it's not unstable enough that I've ever experienced a BSOD from it or any crash which I could directly attribute to it. However, I figured why not try OC'ing again a bit differently.

With my previous 4.6 GHz OC, I raised my BCLK to 240 MHz from 200 MHz and my CPU Ratio from 17.5 to 19. I think I upped my VCore by +0.1V so it was around 1.428 V max, iirc. I also lowered my HT Link from a multiplier of 13 to 9 and my NB from a multiplier of 10 to 9 so they were both at 2160 MHz according to CPU-Z. My Motherboard sets my 8GB DDR3-1600 RAM to 1333 MHz by default with XMP disabled, but by changing the multiplier from "Auto" to 6.66, I got to 1598 MHz while still leaving XMP disabled.

Now I'm thinking to myself, why not just raise the CPU Ratio and leave the BCLK completely alone? For this all I did was keep the BCLK at 200 MHz and raise the multiplier by 21 (for now) for a frequency of 4.2 GHz and then set the memory profile to XMP which got me to 1600 MHz as well. Everything else I've left alone. As I'm typing this, I noticed that my 6th worker/core stopped working for my Prime95 test about 5-10 minutes in. I'm guessing this is because I need to increase my V-Core so I guess I'll do so by +0.025V. So far does this all sound good?

So I was just wondering about the following:

-Could I get some feedback on the theoretical stability of this 4.2 GHz overclock? For example, it seems like a lot less could go wrong when using the CPU Ratio over the BCLK, am I right?

-Will using primarily the CPU Ratio to get a frequency of 4.6 GHz be just as efficient as using my previous method of getting 4.6 GHz? That's assuming I increase my HT Link and NB accordingly as well to match what they were in my previous 4.6 GHz OC, though I'm not sure if it's possible for those two...

-Anyone know why my 6th core always seems to be the one that goes? Is it possible to have a weak core? Everytime I've failed a stability test in recent memory, it's been the 6th core that's the first (and usually only) one to go.

-Is there any difference between an FSB and BCLK, or are they synonymous? I once read on one of these forums that they in fact aren't synonymous and I'd like to know the differences if that's the case. Same goes for CPU Ratio and CPU Multiplier, though I'm pretty positive those two are the same.

 

[Bill Reinhardt]

1. I have had more stable clocks just using the ratio over the BCLK

2. Increasing the ratio can possibly give you a more stable clock, but increasing the BLCK can help improve single core performance better than the ratio. Either way, you will be just as satisfied with only increasing the ratio.

3. As for why your 6th core fails first... not too sure on that one. The only answer I have is "not enough voltage," but it seems like you are getting pretty high with the voltage as-is.

4. There is a difference between FSB and BCLK. The BCLK is basically there just to generate clock speeds for other components to make sure everyting is in sync. What makes it different from the FSB is that the BCLK is not an actual "BUS," in a computing sense, that governs the communication speed between various components on the mobo. The two can be referred to interchangibly without too much confusion because of how similar they function... depending on the person

 

[ihog]

If a core fails in a stress test, almost 100% of the time it means you need to raise the vcore.

 

[klepp0906]

In my experience a worker stopping = memory oc or dram voltage. If it was the cpu itself failing it would blue screen.

You can do a more thorough analysis if you record which test it is that is causing the failure/ error / crash