Explain this please: Temperature difference during OC

[Trace]

When I run my i7 at 191 * 21 = 4009Mhz at 1.3vcore, I load around 70*C on my water.
When I run my i7 at 215 * 19 = 4080Mhz at 1.31vcore, I load around 55*C on my water.

Whats up with that???

[x88x]

I'm definitely not an expert on overclocking by any means, but I would guess that the multipliers generate more heat...or something... I seem to remember hearing that it's better to increase the base clock than the multiplier, though I suppose that would depend on the CPU. I haven't actually done any overclocking yet, though once I get Zeus up and running I'm gonna be shooting for your 4GHz club

[Trace]

Well when it was the good ol' FSB having a higher FSB was good so you had more memory bandwidth, however, I thought that was a thing of the past with the new QPI

[mDust]

I know there is a difference between the old architecture and the i-series chips, but I don't know off the top of my head how it all works (even though I've read about it several times). I guess I learn better by 'doing'. But I would assume that you should still run the clock speed up as high as it will go while maintaining stability. Once you max a stable clock speed, bump the multiplier up until it crashes, then back it down a notch...or get better cooling and continue bumping the clock up.
The multiplier does make the CPU work harder each cycle so it should probably get hotter. But excessive heat from OC'ing is usually due to increased voltages that are needed to support higher clock-speeds.
It looks like the i7 prefers higher base clocks though. Which chip do you have?

Well when it was the good ol' FSB having a higher FSB was good so you had more memory bandwidth, however, I thought that was a thing of the past with the new QPI

So the chip talks directly to the memory now instead of through the Northbridge...why would memory bandwidth not be important anymore? Intel finally got rid of a major bottleneck, it wouldn't make sense to constrict it again.

[Trace]

I believe it to have tons of bandwidth now
and I think I found the issue: speedfan. Lol
I tried with core temp and real temp and they reported same to each other and different from speed fan with the dropped multiplier. It seems speed fan does not recognize the chip correctly with the dropped multiplier or something

I have the 920 d0 revision

[mDust]

Hmmm...that's -1 for speedfan in this thread too.

[dr.walrus]

I'm having similar issues with speedfan. Temperatures simply not reporting with any accuracy.