A study on twin-tower architecture

[burebista]

Oh well, my friend Ramiro left me speechless. After 4 months of testing he put together a comprehensive article about twin-tower architecture. Unfortunately it's in Romanian but here you can go through Google translate.
You have all kind of tests, in case/open stand, from low TDP to extreme TDP, 775 vs. 1366, from fanless to extreme CFM, a study of each heatsink, a study about each heatsink mounting system, an extreme test NH-D14 vs. TRUE Copper with MX-3 and Liquid Pro and more.
If you have some questions about the article shoot and I'll try to answer/translate.

 

[xLAMBIEx]

Very, very nice. shall have to read the whole thing when I have the time. I find the english translation just isn't good enough throughout the whole article though.

 

[burebista]

I find the english translation just isn't good enough throughout the whole article though.

Unfortunately yes, Google is not so good. I have a promise from my friend that the article will be translated in English but I don't know when. Now after 4 months of testing and writing he wants a break.

Meantime I've tried to translate myself last page with conclusions. Hope it's better than Google but bear with my Engrish though.

Conclusions

We reach the end of article, important things were said and results speak for themselves. I'll try to conclude on twin-tower architecture generally speaking and for each heatsink.

If until now I have doubts about this architecture performance or efficiency (doubts because poor performance of IFX-14 on Socket 775 but also because market lack of good heatsinks with twin-tower architecture) all my doubts vanished when I tested NH-D14. I saw a very good performance at extreme categories and this is hard to achieve. It's first time in “modern” history when a heatsink is good at low/fanless RPM but also at extreme TDP/CFM. A huge contribution was from Noctua's quality and experience but a very good performance for IFX-14 on i7 platform reinforce the conclusion that twin-tower architecture is one of the best solutions for making a very good heatsink.
I also can remark a very good behavior in case in a low airflow but with condition that that heatsink must be at least a little optimized for low airflow.
Using three fans is also a strong point of this architecture and it's proven in my tests.

I always said that it's almost impossible to build a heatsink which can win at all categories (from fanless to high TDP) but those from Noctua almost made impossible possible.
Magashadow beats Noctua at two categories (1x500 RPM and 1x800 RPM) but IMO this is the weak point of twin-tower architecture: not so good efficiency with one fan with low/very low RPMs. Nevertheless NH-D14 is very close to a perfect heatsink.

We can speak about heat generated by those two CPU's. It's very easy to see that for Q6600 (because low TDP) we have almost equal results. It's clear which CPU we must use to see the differences beween heatsinks and it's also clear that “ancient” CPU's don't have a place in a modern heatsink test. IHS size, two dies vs. single die, number and density of transistors, VRM's efficiency for 45nm CPU's, die area compared with IHS area, mounting system with good pressure, different behavior of top heatsinks with different TDPs, all those are decisive arguments for choosing the right platform for testing. So now LGA1366 platform is the only choice for a heatsink test with relevant results for any user.

ASUS TRITON 88

Asus is not a big name in heatsink market but I can say that Triton 88 was decent especially if we take account of very select company that was included. This heatsink is a step forward for ASUS and I hope that they'll improve build quality and mounting system. I can say that Triton 88 have potential but it was shadowed by those drawbacks above.
I'm expecting from Rampage series makers new heatsinks with better quality and a mounting system as solid as VRMs on Rampage 2 Extreme.

THERMALRIGHT IFX-14

For me the surprise of the tests. Probably IFX is the most disputed heatsink because many reviews shows a very good performance but some reviews shows quite opposite. The “mistery” was revealed testing it on two platforms. On LGA775 it wasn't a top performer but on LGA1366 managed to beat Megashadow from medium RPM's and up.

Among the things that lead to this good performance is better mounting system (we saw the problems on 775) and i7 IHS preference for heatsinks with concave base.

I can say that IFX-14 could be a good choice for those on 1366 platforms and fans with 1200 RPM and above. Also it can be use for benching sessions on 1366.
Things discovered in this article strengthen the fact that for conclude about a cooler you must test it thoroughly because we have coolers that behave different in different testing conditions (I can point to CM Z600 which loose performance on Nehalem and high/very high TDP).
One more reason to emphasize again the need of using a representative test platform for current technology.

PROLIMATECH MEGASHADOW

This Taiwanese tank surprised me several times in this test. He manage to defeat the giant Noctua at 1x500 RPM on Socket 775 and 1 x 800 RPM on 1366, impressive victories if we take a look at his size compared with Noctua. If we count that in many categories was very close to Noctua we can see an exceptionally heatsink, the most complete until Noctua arrives.

It's a good choice for all purposes but especially for low RPMs and for those without big cases.
I can say that Megashadow is one of the best heatsinks until now.

NOCTUA NH-D14

I don't know what I can say now at the end because I already say all about this amazing heatsink. Things were pretty clear since then chapter “WOW factor” but I must test it and compare it to draw some complete conclusions.

I've always wish to see an impressive heatsink who can lead especially on extreme conditions but I didn't expect to see it soon. NH-D14 is the most complete and best performer heatsink at the moment. It managed to cool fanless CPU's with huge TDP's (was impossible until Noctua), it managed to keep cool a Nehalem at 1.6V with only 1600 RPM, you can see that in this article. Limits on air cooling under normal conditions, close to the average user, have been seriously shaken by this cooler.

I don't believe that I have more to say after I've seen Noctua's behavior in my tests, all I want to say is congratulations to Noctua for making the best heatsink in air cooling history and I'm waiting from them a extreme heatsink based on tower architecture.

In the end I can say that I wish to make more studies because I didn't do all what I want to do and I thank all who helped me and were close to me in those tests.

I hope you enjoyed this article and that you found useful and it's possible that in near future I'll have more heatsinks to test.