1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Testing heatsinks...

Discussion in 'Feedback & Suggestions' started by Boscoe, 15 Mar 2015.

  1. Boscoe

    Boscoe Electronics extraordinaire.

    Joined:
    5 Jan 2010
    Posts:
    1,079
    Likes Received:
    34
    Why not just use a test rig and measure the thermal resistance of all the heatsinks as well as testing them on real world rigs? This is a much more scientific and reliable method of measuring the performance of a heatsink and is the default method in the electronics industry.

    I expect people will argue cpu/package will play a part however I believe this to be minor - just find a standard test rig. The industry copes with hundreds of packages and still rates their heatsinks with thermal resistance ratings, these are far more varied than CPU packages too.

    It's a no brainer!

    Boscoe
     
  2. Combatus

    Combatus Bit-tech Modding + hardware reviews Staff Super Moderator

    Joined:
    16 Feb 2009
    Posts:
    2,735
    Likes Received:
    59
    Hi Boscoe. It is something we've considered but realworld rigs are our priority and as we intend to include AMD testing and possibly noise testing as well, once we're testing on three systems for every cooler, we simply won't have time to do any more testing.

    I can appreciate the advantages of using a heater core for more consistent testing too, but with Intel's LGA1150 CPUs especially, the lack of solder between the core and heatspreader means that realworld testing is far more important as it often shows that some of the more high performance coolers are wasted on certain CPUs when they're overclocked, as the heat just isn't able to transfer away from the core fast enough, hence the big drops in temperatures that you see if you delid your Haswell CPU for example.

    You wouldn't see this in a 'synthetic' test rig - the heatsink would just perform very well, when in reality it may not perform any better than a much less effective cooler as the CPU core to IHS thermal transfer is acting as a bottleneck. So as realworld rigs would have to be part of any test procedure we have, it's those we'll be sticking with unfortunately! Unless we get paid more or days suddenly get much longer :) However, we will be doing a blog post on some of our new test rigs soon to see what results you want to see - for example, would you want us to delid our Haswell cooler test CPU to remove the bottleneck I just spoke about, how you'd like our results presented (having all-in-one coolers in the same graph etc)...
     
  3. Boscoe

    Boscoe Electronics extraordinaire.

    Joined:
    5 Jan 2010
    Posts:
    1,079
    Likes Received:
    34
    I understand that and that's fair enough and it's important for real world tests however you can also measure the thermal resistance between junction and case of CPUs, get the mean and standard deviation then plot this info against the thermal resistance of the cooler and you'll get a graph that shows the exponential decay of return of a lower thermal resistance cooler. This is a far more efficient and easier way of comparing coolers than trawling through reviews to find which ones which at whatever fan speed etc.
     

Share This Page