.09 AMD64 overclocks 45%!

[Kooldino]

Go AMD!

[Mehen]

Most P4s can get a 20% oc with stock cooling, even (especially) the prescotts, and the temp isnt going to raise much higher than it already is. I just dont see why they are basically stopping the rise dead at (below) 4ghz when they can do better than that. I'm not saying they can keeping rising and rising until they get to 5 or 6, but I KNOW they can do over 4 before stock cooling becomes an issue.

[dZeus]

increasing the clockspeed by 20% on a particular sample will also increase the heat by 20%, is it not? I would hardly call that 'the temp isnt going to raise much higher than it already is'.

Look at the thermal guideline for their 3.6 GHz parts:

http://processorfinder.intel.com/scripts/details.asp?sSpec=SL7NZ&ProcFam=483&PkgType=ALL&SysBusSpd=ALL&CorSpd=ALL

115 Watt thermal guideline for a desktop part!!! I'd say that's enough indication to let all alarm bells ring when you have plans to scale it up to 4+ GHz speeds.

[Wombat]

Originally posted by dZeus
increasing the clockspeed by 20% on a particular sample will also increase the heat by 20%, is it not? I would hardly call that 'the temp isnt going to raise much higher than it already is'.

No, it's not. One of the reasons these things are so hot is that there's a lot of leakage - power that is just being dissipated, not doing work. This waste heat isn't going to change much with clock speed.

[dZeus]

alright, but raising clockspeed is surely going to affect the total amount of produced heat, and with the 3.6 at 115 Watt, that would not be a pleasant forecast for faster CPU speeds.

[rylan]

Yeah its going to dissipate more heat because the transisters are switching faster.
Whats happened with Intel's process is the majority of their power and therefore heat is from current leakage as mentioned, which is related to the voltage you run the processor at. In the past with larger feature size current leakage was a small fraction of power dissipation, so temps were closely related to clock speed and not so much voltage.