«

»

Matching CPU to Heatsink Cooler

PAGE INDEX

<< PREVIOUS            NEXT >>

Testing & Results

Testing Methodology

You can refer to the Argon cooler articles for more extensive details on how they were tested with the FX-8320 CPU. For getting temperatures from the Core i5-2500K, the process was pretty similar (although it goes without saying different components were involved). AIDA64’s built in stress test was utilized for both CPUs – once temperatures “plateaued”, peak temperatures were recorded and the ambient temperature was subtracted to arrive at a change in temperature (that you see in the chart below).

Test System

  • Motherboard: ASRock Z68 Extreme3 Gen3
  • System Memory: 8GB Crucial LP DDR3 1866 MHz
  • Processor: Intel Core i5 2500K
  • Audio: Asus Xonar DG
  • Video: Sapphire AMD Radeon 7970 3GB
  • Disk Drive 1: Western Digital Black 1TB 7200 RPM 3.5″ HDD
  • Disk Drive 2: OCZ Vertex 2 50GB SSD (Used as SRT Cache)
  • Enclosure: NZXT Phantom 820
  • PSU: Fractal Design Integra R2 750W
  • Monitor: Hanns G 27″ 1920×1200 LCD
  • Operating System: Windows 7 Ultimate 64-bit

Results

On the overclocked FX processor (consuming about 200W) we saw a pretty large disparity between the AR01 and larger AR03. Obviously, larger processors like the FX series and socket LGA 1366/2011 processors are the intended application for the AR03, but what about smaller CPUs like the Sandy/Ivy Bridge Core series? Are the two untouched heat pipes on the larger AR03 “wasted?”

MatchingCPUCoolerToCPUResults

I guess that depends on your definition of waste. The benefits of the larger AR03 are obvious on the larger FX CPU’s heat spreader (and increased thermal output), but you don’t really lose performance if you move the AR03 to a smaller CPU. Adjusting for ambient temperatures, the AR03 performs almost exactly the same, even with two heatpipes “wasted.”

In fact, it looks like my math wasn’t too far off. Looking at the i5-2500K result specifically, 24mm of heatpipes look like they perform very similar – regardless of how they’re packaged.

CPU Cooler Conclusion

All of this is just a way to say the obvious – a greater amount of contact surface area generally transfers a greater amount of heat. One could infer then, that a similar amount of contact surface area transfers a similar amount of heat, despite the platform. When choosing a cooler for your CPU though, contact pressure and surface area are probably the greatest factors for the ability of the heatsink to conduct heat. None of this is really new ground (as later uncovered by Olin Coles during an extensive thermal paste review) – many of these patterns have revealed themselves over the years that Benchmark Reviews has analyzed CPU coolers.

I have to wonder though, will all of these “truths” about CPU coolers slowly become irrelevant? As the physics of ever shrinking processors changes (and the nature of their heat output with it – one of the reasons I still like my i5-2500K over my i5-3570K) will we run out of space for any number of heatpipes? Cooler Master has notably experimented with vapor chambers on some of their CPU coolers – would technologies like this become necessary to extract heat from smaller and smaller CPU dies? The results haven’t always been promising, but perhaps it would fare better on a ultra-modern platform (would the faster heat-spikes of Haswell be tamed by such features)?

For now, when it comes down to it, choosing a CPU cooler might be as simple now as picking a price point – if there’s any pattern I’ve discovered, it’s generally “you get what you pay for.” I would assume manufacturers of higher-end CPU coolers end up testing these products far more in depth than I could at home. Of course, we’re always willing to test those manufacturer claims to make sure you’re getting your money’s worth – that’s why sites like ours exist!  SilverStone was right in recommending the larger AR03 for an overclocked FX-8320 processor, but even on smaller CPUs the AR03 saw a little performance improvement (it just wasn’t as necessary on the smaller i5-2500K processor).

So does any of this matter when choosing a CPU cooler for your CPU? While there are many variables still to discuss, I’m not sure I could say you could pick a “wrong” cooler – they all cool better than the stock heatsink! Obviously there are coolers that are more appropriate for certain applications, but since the latest CPUs seem to be limited by their manufacturing process anyway you can generally let your budget and aesthetic tastes decide.


SKIP TO PAGE:

<< PREVIOUS            NEXT >>

3 comments

  1. Chris

    At the moment, the current generation of top crop of heatsinks are the Cryorig R1 Ultimate and the Noctua D15. They are both large dual towers with moderate fin densities and weigh 900g-1kg without fans and perhaps ~1.3 kg with fans.

    There are other issues of course other than contact. The heatsink you showed did not solder the heatpipes to the fins. Newer coolers generally do that. The thickness of the heatpipe itself is another matter. Then of course, the fin density and surface area of air.

    Will these truths become irrelevant? Nope. LGA 2011 CPUs still put out a massive amount of heat. HIghly overclocked Haswell >4.5 GHz use a lot of power too when overclocked.

    1. Tom Jaskulka

      I was about to reply that I’ve had my eye on Cryorig’s gear since they released their R1, but unfortunately haven’t had a chance to test their lineup since they weren’t available in the US…

      …and just discovered that they added Newegg.com to their distributor list. Hopefully we’ll get a chance to test some of their coolers sometime in the next few months! You’re right though, there are definitely a lot of variables between different heatsinks that can affect their performance – the AR01/AR03 were just similar enough that I could try and identify the difference that the contact surface/heatpipes would make.

      AM3+ CPUs and the Sandy/Ivy Bridge-E processors still benefit from these higher-tier coolers as you’ve mentioned, but Haswell still makes me suspect the barrier is more of an internal factor. Looks like I’ll need to get around to updating the testbed sometime to find out… Thanks for reading!

  2. Chris

    Ideally, try to get the R1 Ultimate. The R1 Universal is a few degrees warmer, but allows for 4 DIMM slots.

    At the moment:

    – The D15 seems to win on LGA 1150 sockets and AM3 sockets
    – The R1 Ultimate seems to win on LGA 2011 sockets, so probably 2011-3 as well

    The annoying thing is right now there are no R1s in Canada.

    Oh and the R1 seems to have the best mounting system of any cooler around. You’ll see it if you get it.

    The newer heatsinks generally have:
    – Denser fin density (the R1 is interesting because it is fan-low density-high density)
    – The heatpipes are now soldered
    – Everyone is using higher speed fans

    Haswell (4 core) actually runs pretty hot. You’d have to delid to get best cooling performance. Same with Ivy. The E series though use the indium solder, so it’s not an issue.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

*