Mega-Heatsink Roundup - 0407 @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,532
Mega-Heatsink Roundup - 0407 / Big bucks for big units! We pull out the stops and put 10 big bad cooling systems and heatsinks to the test. What coolers deliver you the best temperature drop for your hard earned dollar? We spell it out for you on a unit by unit basis.


We saw a lot of coolers, ran some tests and got quite a bit of data. But what does it all mean? Does declaring one heatsink the winner give you the entire picture? We don't believe it does. In fact we think that calling one single unit “the winner” can cause more confusion and spread misinformation than it serves to help you with your buying decision. So rather than just point out who had the lowest temperatures, which anyone can do by looking at the graphs, we are going to break it down into a few categories that will hopefully be positioned to help our readers out a bit better since what we have here is a field of good heatsink equipment.
 
Excellent article! I still have a lot to read into though. One thing that caught my attention was how well the Thermalright heatsink performed close to the water-cooling unit used in this round-up under stressful conditions.
 
Good read.

Typo Fixt, Thanks. - Kyle

Although the CoolIT Freezone did perform the best it was of a different caliber. I think the performance crown should go to the Thermalright's Ultra 120 due to it's exceptional performance when the CPU was o/c to 3.4 and fully loaded.
 
Although the CoolIT Freezone did perform the best it was of a different caliber. I think the performance crown should go to the Thermalright's Ultra 120 due to it's exceptional performance when the CPU was o/c to 3.4 and fully loaded.

time after time people have said that the thermalright is one of the best. They are my favorite brand of heatsink.
 
A good review, but I think you should have done a couple of additions and substitutions in the article. I would have much rather you have used an Intel Quad processor as they are much harder to cool than the E6420 used for testing, especially when overclocked. This would have really maximized the heat dumping capability of the heatsinks tested and I imagine that you would have found the Coolit Freezone to be simply overwhelmed by an overclocked Q6600. The other addition I would have liked to have seen would have been the inclusion of a regular, non-extreme version of the Ultra 120. I have found that it performs much the same as the TT-120 when using either the Tuniq fan or a Scythe S-Flex, but I have never directly compared it to the U-120 eX.

One other thing I fault you for with your testing methodology is the fact that you didn't use the included fans with several heatsinks but used that Thermaltake fan and I also think you should have tested the Gemini 2 with 2 fans on it. After all, these heatsinks should be tested in their as-delivered configuration for out-of-the-box performance if they have a fan included with the heatsink. If you wanted to test them as much as possible with 1 type fan, then that should have been done in addition to the stock fan setup.

Other than that, good article.:)
 
great article, and I am happy to have the best rated heat sink right now, not sure though what is in store 2008 though.
 
@ muddock: I agree with your input but as with most things in life their must be a compromise. The one fan was used throughout testing to limit any variable and to test the performance of the cooler not the fan. This was done since many people have their favorite brand of fans and will often times swap out any included fan for a quieter or more powerful fan as their application dictates.

Using two fans on the Gemini would have given it an advantage over the other coolers as well.

Testing the "out of the box" experience is incredibly valuable though this would mean some heatsinks would be tested passively since they include no fan. This would certainly skew the results.

After all of that to include all the possible combination would have inundated the reader with information overload. This was the compromise. To give you the best (scientific) results for the performance of each heatsink.

If you care to donate a q6600. Please PM me and I will be more than happy to retest the heatsinks for you :D
 
Nice. The price / performance graph really kicks ass...

I heard OCZ is making a phase change unit that is about to hit the shelves. Any chance the [H] might review something like that?
 
Yeah, I realize that it would have thrown a whole lot more testing and variables in the equation to test everything as delivered and I do understand why you did the testing with the same fan where possible. And I have done some heatsink tesing myself and know just how much time it takes to go through such a marathon round of testing as you did here; my hat is off to you. :D

I figured you could get Kyle to loan you a Quad for the heatsink roundup. And while I do have a Q6600, I plan to hold on to it myself for now,as it is a Seti crunching monster, just a lot harder to cool than a C2D. But after the new year I might have a further upgrade (Penryn) planned so I might have the Q6600 available for testing if you want to go through the testing again with a certified heat beast. But that depends on how bad or good Christmas treats the bank account.;)
 
These results also fit what was expected prior to testing.

hmm , stop that. :p

agree with the Quad core, its where we are all being dragged, sooner or later.

Gemini clearly designed for 2 fans at a minimum, not an option, no problem swapping out to a standard fan on single fan designs (even if a second fan is an option) but to cripple the base design in terms of "fairness". Dont see it.

Forget shiny, we know shiny does not matter a rats ass. Is the base FLAT, that is the question. Even just a razor blade check would be interesting.

second the price/performance graph, very nice.


Good stuff for first time out of the gate . Appreciated the commments on cooling the VRM etc. IMO lots of boards have died early deaths by overlooking this. hmmm, watta bout measuring the temp of a mosfet (or heatpipe ) on the reference platform and providing some accessment of what if any, ( I know one of the towers is susposed to have some bottom fins bent down to provide airflow) VRM cooling is provided.
 
...Although the CoolIT Freezone did perform the best it was of a different caliber. I think the performance crown should go to the Thermalright's Ultra 120 due to it's exceptional performance when the CPU was o/c to 3.4 and fully loaded.


Agreed.
 
The Coolit systems both lose the advantage of a TEC because the power usage of the CPU is overshooting its capabilities, and you're reduced to crappy water cooling. I have an Ultra-120 Extreme and its a great HS, highly recommended.
 
The Coolit systems both lose the advantage of a TEC because the power usage of the CPU is overshooting its capabilities, and you're reduced to crappy water cooling. I have an Ultra-120 Extreme and its a great HS, highly recommended.


I am fairly sure this is exactly what the testing points out. :) I think we have put together enough data for any enthusiast to make their own call. I think the Value graphing put everything into perspective.
 
i've been waiting for some old fashioned [H] heatsink reviews :p

Again it was a situation where we have been looking for a person again for a while that had the time and the passion for the products and hopefully Marc is that guy. I am very satisfied with his first article for us.

Please leave your feedback so we can mold the content to deliver what you guys want.
 
Good review, very thorough but to the point. Just reinforces that I should upgrade to either the Tuniq or the Ultra Extreme. I also agree with muddocktor and Bill though about the GeminII, although I understand what you were doing with putting all the heatsinks on "fair grounding". But still, it's the ONLY heatsink that was actually designed to be used with 2 fans. I'm not sure any enthusiast would have bought that particular heatsink and not put 2 fans on it. But anyways, I'm sure it was a tough call, have to draw the line somewhere. I would just assume another degree or two could be cut off the total.

Also too bad it wasn't a quad, but except for the CoolIt products, I would assume that most of the other heatsinks would perform about the same as here, just hotter :p
 
Good review, very thorough but to the point. Just reinforces that I should upgrade to either the Tuniq or the Ultra Extreme. I also agree with muddocktor and Bill though about the GeminII, although I understand what you were doing with putting all the heatsinks on "fair grounding". But still, it's the ONLY heatsink that was actually designed to be used with 2 fans. I'm not sure any enthusiast would have bought that particular heatsink and not put 2 fans on it. But anyways, I'm sure it was a tough call, have to draw the line somewhere. I would just assume another degree or two could be cut off the total.

Also too bad it wasn't a quad, but except for the CoolIt products, I would assume that most of the other heatsinks would perform about the same as here, just hotter :p

Undoubtedly this was discussed. I think next time we will select a standard fan for those units that do not come with a fan, and then go with out of the box for all others. What do you guys think about that? Surely some will complain about that too, but we want to deliver what we think the overall readership will like best.
 
Great review! I was leaning towards the Thermalright 120 for this upcoming build and now this review just sealed the deal for one.

Along with the Thermalright HR-03 GT for my 8800 GT sitting alone in the box.
 
Very well-written article. It was nice to see alot of well-known players up on the block being tested. You covered the usual enthusiast heatsinks (i.e. Ultra Extreme, Tuniq) as well as some I really don't know too much about, like the Noctua and the ZEROtherm. I'm not in the market for a new heatsink, seeing that I'm really pleased with my Ultima-90, but I thought the article was very informative and to the point. Welcome to [H]! :D


On a technical note, when I was reading the article just a few minutes ago I wasn't able to see any of the enlarged pictures. I kept getting the message "Picture Not Available" when the new window (tab) opened. This is with FF 2.0.0.9 (I gave it a quick try with IE but no go there either). Thanks!
 
Neither the Thermalright nor the Coolermaster come with a fan, which affects the value proposition * IF * you don't already have a suitable fan, and need to purchase one. Perhaps an important footnote for some value shoppers.

Otherwise, I really like the use of one fan for testing purposes, to isolate that variable. The price/performance graph is a nice comparison tool.
 
Great article :) I hope the rigorous testing methodology will help you add more in the future and still use the same comparaison basis to see how it is positionned.

i would also like to see some watercooling kits being tested as well... Maybe a round 2 ?

 
What do you guys think about that? Surely some will complain about that too, but we want to deliver what we think the overall readership will like best.

THAT, right there, that THAT, is why [H]Enthusiast has been my home page for well over 10 years. (Well and the fact that CNN blows as a hard news service) . I do not care if you dont listen to me, what the hell do I know anyway, but you listen to US. :cool:
 
Concerning the price/performance ratio, it can certainly vary for heatsinks that don't include a fan. It can be a tad misleading, for example, to mention a $0.03 difference between the Thermalright Ultra 120 and Tuniq Tower 120 if the additional fan wasn't included in the price. If it was, it's not clearly mentioned in this particular section. It's not a huge deal as long as the reader is aware of that factor, but it can definitely mislead a few who don't remember which heatsink has a fan included or not.
 
Personally, I think the 1 fan for all is a good idea. Helps to keep a level playing field for all contestants.

I would say beyond that, if a particular heatsink warrants special testing - ie: the GeminiII w/ 2 fans (or Ninja w/ 1/2/4 fans mounted), or maybe the stock fan on a particular unit is grossly different, then such scenarios could be tested (if time permits) and included to the graph. I'd think that a unit being designed for multiple fans being tested would outweigh the need to re-test a unit due to fan differences though.

ps More HSF testing please - it's been far too long! Given the crappy one slot cooler leading to high temps (good cooler maybe for HTPC people, but not for the overclock crowd so much) and LOUD fan (if maxed) on the 8800GT, I'd love to see a GPU cooler comparo next. Then again, it's been what about 2 years since the last watercooling roundup?

[edit] Back to the current testing though. The GeminiII, with it's seemingly constant FREE -> ~$10 price point (AR!! - HATE stats-now, but I digress) recently, it's got to be one of the best bang for the buck coolers around right now. Sure you have to budget ~$8-20 tops for a couple fans, but I too kind of appreciate the level of flexibility it affords having no fan stock. Certainly the cooler provides some piece of mind for those w/ boards that have passive/heatpipe cooling too.

[e2] @ Below, yes just checked, happening to me too.
 
Is it just me, or do the images not work? (eg the fullview images of the heatsinks show 'picture not available)...Anyone?
 
Why no Zalman?

Just had to stop somewhere....also there were a lot of behind-the-scenes issues with this article. It was supposed to publish in Early August. After the reviewer had the product for two months, he told me he quit and shipped everything back to me. I was lucky to find Marc to put it simply. He did a great job with what he was given to work with and turned it around in less than 30 days.
 
Is it just me, or do the images not work? (eg the fullview images of the heatsinks show 'picture not available)...Anyone?

Yep, you are right, they are borked. The images worked in our preview mode....I will fix it. thanks for the heads up!
 
Excellent article, but only one "issue": is the Thermalright sitting more steady now and not turning like it used to?

I bought one a long time ago, back when everyone complained that no matter how much you tightened it, that it could easily be turned by hand, which mine could.
A 120mm fan on it actually would make that giant hunk turn/spin a bit.

So I went back to Zalman 9700 (which does excellent, btw... shoulda maybe included that in the review too).

Maybe it was the few early batches only, but has that changed?
If not, I cant imagine how it did that good.
Since it wasn't mentioned, I'm wondering what the deal is with this cooler now.

Thx.
 
It's hard to rate the TR-UE as better price/performance when it doesn't come with a fan compared to the Tuniq which not only comes with a fan, but a controller for that fan.

A nice fan runs anywhere from 8 to 20 dollars depending on what brand, what cfm, what dba, and what other features (lighting, controller, 5v resistor mod, etc.) come with said fan.

I also share the same thought as murdocktor though that the heatsinks should have been tested with their default setups wherever possible, and the Gemini II should have been tested with 2 fans as that is how it is intended to be used.

And I also would have liked to have seen a Quad used although I must admit I was a little happy inside to see that you used an E6420 as I have an E6420 and the results were a bit more tailored to what results I would see.
 
Thanks for your feedback, we will tweak the ways we are testing for the next article.
 
Great article here!
Glad to know that I have one of the best air coolers around. Makes me warm and fuzzy inside.
Now if I can just get it to not shuffle around while mounted horizontally :(.
Someone asked if the Ultra 120 Ex can still move/twist easy when mounted. I just got mine ~2 weeks ago and can say yes, it does all too easily :(. Kinda had me worried about contact at first. I'm going to be re-seating it tonight, just to be sure.

It'd be nice to see this on the basic kits like the big waters. They catch my eye from time to time, though I doubt they're much worth it. It'd also be cool to see high end air versus low/mid end water.

All in all it was a good read though. Seems like this is one of those things that you can get away with doing once a year and forget about it until later :p.
 
Thank you to everyone who has taken the time to read the article. I certainly hope you enjoyed reading it as much as I did preparing it for you.

Like any great piece of work there will be revisions, tweaks and adjustments so keep those comments (good & bad) coming. Where else can you be certain to have a hand in the development process of upcoming reviews!


Eulogy: All in all it was a good read though. Seems like this is one of those things that you can get away with doing once a year and forget about it until later .

With so many different cooling products out there and so many on the way you can be sure there will be more articles to come! :D
 
I'm not sure about the 1fan issue, If you were to do an in depth review, you could compare high / low fans, some designs are meant to be used with lower air flow, while others might benefit from the high air flow. Even try some passive setups in a case with good airflow, antec 900 :p? P180, cm stackers etc.

and on zalman, they have lost their touch, their products are overpriced compared to the other choices out there which seem to perform better.
 
I had a thermalright 120 ultra extreme a while back and found it to be a piece of shit, the base was amazingly uneven, i was getting better temps on my quad core with the intel stock hsf. Contacted thermalright about it and got no reply. Was coupled with a scythe 120mm 1600rpm fan.

I dunno if some of the blame lay in my case and mobo, im using a lain li v1000+ and an asus p35 mobo, the socket position on this mobo is really close to the botom mof the case so the hsf had to be mounted facing the bottom of he case and not in the preferred orientation blowing towards the 120mm exhaust.

However ive seen people with it mounted the way i had it and were getting good performance. Ive read numerous complaints about the bases on these hsf's being skewed to one direction and i think its pathetic that people have to lap these things out of the box so they get a good contact. Then again the intel heatspreader isn't exactly innocent as it can be pretty wonky as well.


pc-internals.jpg


You can see there even the artic freezer is a tight fit. 2900's got sold off long ago :(
 
Great job Marc, bet you got a little bit tired of mounting monster heatsinks... I also agree on the conclusions, well except that I disqualified the CooliTs in my mind.

Some words in the intro on the principles of heat transfer through heatpipes would have been nice, since they behave quite differently from massive coolers. Unlike massive Cu/Al coolers that have constant heat transfer coefficients, the heat transfer coefficients of heatpipes will increase with increasing temperature differential (caused by increasing heat dissipation from the CPU) up to a critical point, above which the heat transfer coefficient rapidly decreases. At this point, the liquid coolant inside the heatpipes will not be able to make it from the cold end back to the base end of the heatpipes before it is evaporized, and heat transfer will take place through the metal walls of the heatpipes alone. Needless to say, this will lead to overheating and throttling of the CPU.

I observed "supercritical" heatpipe behaviour when running Prime95 on my Q6600 at 3.5GHz and 1.5V Vcore with a Big Typhoon VX. The core temps did not stabilise and approached throttling temp (100C) in a few minutes. At these settings the max. theoretical power consumption of the Q6600 is around 200W, and considering the max power rating of 130W for the BT VX it clearly performed beyond its call of duty. At 3.4GHz/1.4V it was P95 stable at 85C core temp (around 170W). However, I went with the flow and now have a TRUE waiting to be installed :)

Edit: For reference, the max power of the E6420 at 3.4GHz and 1.4875V will be around 113W (http://web.aanet.com.au/SnooP/psucalc.php)
 
Well, I own a CoolerMaster GeminII w/2 fans. I like it on my Q6600 (B3) @ 3.4GHz, but the installation was a bitch (even in an Antec 900 case)

Anyway, I guess with 2 fans, you guys are saying that it wins right? Right? I mean, you had to run 1 fan to make it fair for the other HSFs.

*hugs GeminII for reassurance*
 
Back
Top