20 heatsinks reviewed and compared on matbe

[Slaugh]

20 heatsinks reviewed and compared on matbe
(For noise vs. temperature @ 5, 7, 9 and 12V, go to page 18)


Here are the contenders:

• Akasa Evo 120
• Arctic Cooling Freezer 7 Pro
• Asetek Vapochill Micro
• Cooler Master Hyper 48
• Gigabyte G-Power Pro
• Gigabyte G-Power Lite
• Intel Box
• Noctua NH-U9
• Noctua NH-U12
• Scythe Ninja
• Thermalright XP-90
• Thermalright XP-90c
• Thermalright XP-120
• Titan Vanessa S-Type
• Titan Vanessa L-Type
• Tuniq Tower 120
• Zalman CNPS 7700 AlCu
• Zalman CNPS 7700 Cu
• Zalman CNPS 7700 Fatl1ty
• Zalman CNPS 9500 LED

[Dirty-Harry]

The reviews are in french... me no speaky french

[Firetech]

The reviews are in french... me no speaky french

Hmmm, it's not in French when I view it in Firefox

Thanks Slaugh

[nici]

Its babelfish translated from french to english.

Anyway, the Noctua NH-U12 looks interesting.. Not that im planning on upgrading, i´ve thrown enough money in the bottomless hole of pc silencing as it is..

[jmke]

that's some very nice testing, must have taken him plenty of time; if I'm not mistaken he's an employee in a computer store; and get his hands on plenty of hardware to test.

29 tests at 4 different fan speeds, approx ~116 hours worth of tests without mounting time included; YAY! I spent close to 140 hours with "only" 89 test configs, but needed to install them inside case (motherboard removal for some) which eats a lot more time.

I have nothing to add to his findings, only remark I can make is about the test setup; testing inside a case will give certain HSF an advantage (tower heatsinks) and others a disadvantage (xp-120 and similar). In an open testbed you get very artificial results, not always a representation of real-life performance. The XP-120 is almost on par with the rest in his tests; while I had some disappointing performance in a case, the HSF was not able to keep the CPU cooled with fan @7v where the tower heatsinks (Freezer/Noctua NH-U/SHogun) did it with ease.

[Firetech]

approx ~116 hours worth of tests without mounting time included; YAY!
I spent close to 140 hours with "only" 89 test configs, but needed to install them inside case (motherboard removal for some) which eats a lot more time.

jmke, I haven't had time to check the full report (I originally posted at 3AM) but all I can say is that you are both HUGELY dedicated guys.

THANKS for doing the testing that we all wish we could do.

BTW, I know both testing methodologies have their merits & flaws but FWIW 'testing inside a case' results are what I'm looking for while I try to make my purchasing decision.

[spolitta]

thats an strange test, Why is Ninja doing so bad compared to something like the Gigabyte G power pro?

[jmke]

is it doing bad? I don't think so, look at the dBA level of the Gigabyte unit...

[Felger Carbon]

thats an strange test, Why is Ninja doing so bad compared to something like the Gigabyte G power pro?

The G-Pro has 61 or 62 fins on a 90mm radiator, closer than 1.5mm per fin (including the thickness of the fin). The Ninja has 23 fins over ~112mm, or almost 5mm per fin. As a result low air flow passes easily thru the Ninja, but the G-Pro will just stop such air. On the other hand, at high air pressure, meaning a noisy fan, the G-Pro will excell. And this is what the data shows. The G-Pro is terrific with a noisy fan.

[halcyon]

I think the Tuniq Tower 120 is really interesting, although big and heavy.

Other sites have showed that it retains it's cooling potential to very low air volumes and that the matbe graph could perhaps not be the only truth about it's performance at lower fan speeds.

It consistently pulls lower temps than others, only rising (somewhat too sharply, I might add) at lowest air volumes.

Maybe something to consider for those with highest speed X2 processors and extra voltage?

[Felger Carbon]

I think the Tuniq Tower 120 is really interesting, although big and heavy.

Other sites have showed that it retains it's cooling potential to very low air volumes and that the matbe graph could perhaps not be the only truth about it's performance at lower fan speeds.

It consistently pulls lower temps than others, only rising (somewhat too sharply, I might add) at lowest air volumes.

The Tuniq Tower is two banks of fins with a fan in the middle. The Sonic Tower is two banks of fins, separated by 1 1/8", which obviously could have a fan in the middle. In fact, the EU site fan-x reviewed the ST exactly that way here and also got good results. Interestingly, the original review used the word "sandwich" for the central fan position, but the translation renders that as "sand-yielded".