The Equalizer: Our Core i7-4770K Review Sample
One small detail that went largely overlooked in that already-mentioned launch article was voltage. Our source, whose retail processors were topping out in the 4.3 to 4.4 GHz range was only using 1.2 V to maintain low heat levels. After all, he's selling Haswell-based systems that need to remain stable through multiple years of warranty coverage. Meanwhile, I was pushing my press sample harder with 1.3 V. Was he being overly cautious? Was I too aggressive? Could my cooling situation really be that much better? Or might differences in stress testing account for everything?
One of our contacts recommended LinX to test the Hasewell architecture's advanced AVX 2.0 pipeline, suggesting it'd yield temperatures significantly higher than Prime95. Although we didn't see the delta he did, LinX does get somewhat hotter. It's even more taxing than the Intel stress test I used in our first Z87 Express motherboard round-up.
Switching to LinX broke my previous overclocking thermal limit at 1.3 V, forcing a drop to 1.25 V. The reduced voltage in turn dropped my “ragged edge” maximum to 4.6 GHz. Rather than trying to perfect that 4.6 GHz and worry about what might happen to it in the future as the sample started degrading, I informed today's invitees that we would test at 1.25 V and 4.50 GHz.
I also informed invitees that any cooler that allowed our CPU to bump the processor’s 100° Celsius thermal limit would need to be disqualified from this article’s conclusion.
Like the source Chris Angelini queried for his launch article, I could have also used 1.2 V to play it safe and limit my own overclocks to 4.4 GHz. But why would I do that? If I use my old MUX-120 review sample as a starting point, shouldn’t all of today’s top coolers be better?