Cooling Scaling

[PrezWeezy]

There is something I've been a little confused about for a while now and I was hoping someone here could help me out. Anandtech has been testing several coolers lately and I've noticed an interesting phenomena: The best cooling device at lower clocks (even at load) can't always reach the highest stable clocks. I think of cooling as linear and if they are using a chip capable of going up into the 3.94 GHz range I would think that any cooler that can cool the CPU sufficently would be able to reach at least that clock. For instance the Arctic Freezer 7 can keep the CPU to 37 degrees at 3.8, where the highest clocking cooler (the Turniq Tower 120) only keeps the CPU to 39 degrees. Anyone have an answer? I'd be very interested to understand this. Thanks in advance.

 

[tylerdustin2008]

I would have to say that the AC 7 is on idle, and the Tuniq is on load.

There is no way a AC7 will beat a Tuniq. Just because the Tuniq has more of a surface are, and thus can cool a hotter chip.

Also what exactly do you mean by clocks?

The best cooling device at lower clocks

You mean idle?

where the highest clocking cooler (the Turniq Tower 120)

The best cooler?

 

[DerwenArtos12]

Originally posted by: PrezWeezy
There is something I've been a little confused about for a while now and I was hoping someone here could help me out. Anandtech has been testing several coolers lately and I've noticed an interesting phenomena: The best cooling device at lower clocks (even at load) can't always reach the highest stable clocks. I think of cooling as linear and if they are using a chip capable of going up into the 3.94 GHz range I would think that any cooler that can cool the CPU sufficently would be able to reach at least that clock. For instance the Arctic Freezer 7 can keep the CPU to 37 degrees at 3.8, where the highest clocking cooler (the Turniq Tower 120) only keeps the CPU to 39 degrees. Anyone have an answer? I'd be very interested to understand this. Thanks in advance.

You're looking at idle temperatures and they can be quite decieving. IMHO idle temperatures mean next to nothing in the over all cooling performance of a heatink. You have to think of cooling as non linear and as a delta, not a flat number. The most important numbers you could analyze to relate directly to the efficiency of the heat sink are the deltas between ambient and full load temperatures, the lower that delta, the more efficient the heatsink is. Everything else is subject to too much uncontrollable variations. I'm not really even a fan of using a real processor to test heatsinks, a hotplate with a fixed thermal output would give more consistant results.