0.27C/W?!! Isn't that better than the SLK-800?

[jarsoffart]

Anandtech Comdex review


Are those just inflated specs?

 

[Goi]

There are many ways to measure C/W, and without proper knowledge of how the values are obtained, we can't compare them among different HSFs. In fact, unless all other things are kept equal, its useless comparing the values. Its worse than horsepower/torque ratings of vehicles, its about as useful as MIPS/Mflops ratings.

 

[Mikewarrior2]

exactly what Goi said. unless you're going by one set method, you'll see C/W numbers all over the place 😉.


Mike

 

[jarsoffart]

Okay, good. I spent big bucks for my SLK-800 and I was quite shocked to see it killed so quickly. I was comparing it to the 0.60 C/W ratings of Dansdata.com.

 

[Lizardman]

Just wait a couple months for some comparative reviews to come up. Like the ones Hardocp has every once and a while. Those are the only true way to see what hunk of copper and gold is the best.

 

[Goi]

Last I checked the SLK800 is still head and shoulders above everything else in the Socket A scene. The closest competition are the ThermalRight SLK600, AX7 and the Alpha PAL8045. In the Socket478 scene, the recently discontinued MCX4000 still rules, with the Alpha PAL8942 a close 2nd.

 

[Awakened2002]

Whats a good price for the SLK800? I found it at SVC for $33.99...

 

[jarsoffart]

That's the best price I'm pretty sure. I bought mine for 43. Most places have it for around 45.

 

[HeroOfPellinor]

Actually, the Swiftech MCX462+ would probably beat the SLK-800 in a comparison. It also bolts to the MB, is heavier, and isn't made of paper thin copper wafers. Probably not worth the extra $15-20, though.

 

[Goi]

Well, in Dansdata's huge cooler roundup, the MCX462 didn't do so well, and was pretty far from matching the SLK800. With the fan on full power it just barely edged out the Alpha PAL8045 and Thermalright AX7, matching the SK7.

 

[Nerdwannabe]

Originally posted by: Goi
There are many ways to measure C/W, and without proper knowledge of how the values are obtained, we can't compare them among different HSFs. In fact, unless all other things are kept equal, its useless comparing the values. Its worse than horsepower/torque ratings of vehicles, its about as useful as MIPS/Mflops ratings.

I second that.

C/W also depend greatly on contact area.

A heatsink mounted to a P4 chip will have better C/W value than K7 chip due to larger contact surface. (example)

 

[Mikewarrior2]

LIkewise, DansData uses a huge contact surface, invaliding its C/W ratings versus any modern day CPU. 😉



Mike