Originally Posted by PhilWrir
Im not sure I would agree with this.
For instance, Many low end MSI boards are terrible for Overclocking in any real amount because they have crappy and unreliable power phases.
Hands down the chip is more important than the Motherboard, but if you want to go far you need a good board. That being said, when an i5 2500k can hit 4Ghz with the stock cooler without having terrible temps the board you are using looses some of its importance at low overclocks.
My first board was a $120 g45, had very limited voltage options but would still get a 2500k to 5.1Ghz. Thinking it might be the board I went & got a ud4 instead, only to discover it was the limit of the chip more than mobo limitation.
For maxing clocks & benchmarking a better board is very nice, but for guys shooting for maybe a 5Ghz validation & 4.5 - 4.8ghz 24/7 clock pretty much any board should do the trick, although a better board still may do X clocks at a slightly lower voltage (& temps), better memory overclocking, etc.
Originally Posted by Cyclops
I'm guessing you mean the CPU catche not the RAM. I've had systems (1155-1366) stable at over 4 ghz with 48 hours of blend no less that would fail Large FFTs in matter of minutes. So I think Large FFTs is the way to do it.
Blend is when it tests the RAM, Large FFT is mainly CPU.
I could easily be wrong, I rarely do any stability testing.