Socket 775 Supercoolers – 23 coolers roundup

[burebista]

Here. It's in Romanian but if Google Translate fails you at least you have a lot of pictures and graphs to look at.
The guy who made it is a friend of mine which made this as a hobby. Most of them are bought from his pocket, some are borrowed from friends and others are samples send by manufacturers.
You have all flavors of testing from fanless to extreme air conditions. Testbed is the same for all of them.
IMHO this is the most complete and comprehensive super-coolers roundup ATM.
Enjoy.

 

[Zehnsucht]

So umm... Where is the actual summary of test results? All I can see is photos of the coolers.

 

[rpsgc]

Page 40 forward
http://lab501.ro/racire/socket-775-supercoolers-23-de-coolere-testate/40

 

[newfellow]

heh, perhaps I finally get to see how much Thermaltake V1 was actually better than Zalman coolers. (just that seen enough biaching by people on Thermaltake neck while I am atm running Zalman which runs +10 degrees higher than my old V1).

Thanks for the review link..

*edit* lol, it's not listed..

 

[burebista]

Sorry man, my friend tried most recent heatsinks but if you're happy with your V1 no need to go for another solution.
What Zalman do you have now? 10X is their best cooler.

 

[newfellow]

Well, since the Thermaltake V1 Fan kinda exploded (Or well lets just say it took it's last spin with a good bunch.. lol). Had to change it to Zalman CNPS9700 LED which is kinda same thing. Got antec 900 here spinning with both side cooling (from CPU I mean).

mounts and material is exactly the same on both coolers. However, Zalman runs Idle at 29-33c while at same setup Thermal take was 26-28c. Just would of love to see what people on web say since I'm used to see 'ZALMAN* RULES style of comments & thermaltake forum full of people saying that their cooling doesn't work, but hell as far I've seen these it's pretty damn much opposite. Gonna try Arctic MX-3 next on this thing + definedly remount another way way out from top. Perhaps 200mm fan can do better job to get air out for this thing while I don't see how as this thing should literally go all the way good out of back + incredible size wings to draw heat out and from top.

X10 yeah did see that one.. Not gonna fall for that one. Although I have to admit and give + to it for mounting 5 Copper pipelines through the bottom.

I'd definedely dump some AS5 or MX-X on between the middle inside that copper heat sink & silver part on back. (Which I am gonna do with my heat sink as well. Can't explain this trick, but I can say it's worked on everything I've tested -5 degrees easily cooler than on stock setups usually. I think this "my trick" has something to do with better heat dispatch. from all the 'pipelines' working together 'really attached' to the actual copper better than original soldered setup as they sell.)

^ If someone actually checks this out or wants to improve their 3rd party heatsinks. I got one advice to you. Just add really 'thin' line in the middle don't try to "sink the heat sink". Inside these parts like 0.01mm is enough in the middle (it'll spread to sides as there isn't room at all) point being it's good if it simply connects all the pipes + the bottom copper ""lag spaces"" inside the closed space. hmm, and it's probably hell a lot more than -5c on true long Linpack stress I've referring to idle.

 

[mastrdrver]

Good grief man! I've never seen so much anal retentiveness in a review in all my life. Looks too like one expensive review if he bought all those.

I've found for my Vendetta 2, I get best results putting the tim in between the 3 pipes. It really needs to be lapped. The aluminum dividers are slightly higher than the pipes causing the V2 to "sit up" a little higher and not get good contact with the pipes. Still, I'm able to run an i7 at 1.275v and not reach 90C under LinX load after several hours with my R4 fan. If anyone is thinking about a i7/i5, I would recommend getting a four pipe over three pipe HDT since on the later, only the middle pipe will lay directly across the die. Where as the former has 2 pipes on the outer edges of the die therefore being able to transmit more heat away from it.

Also, why hasn't anyone made an all copper base holding the pipes on a HDT cooler? I would think that would just make sense.

edit: So, I was looking through the article and clicked on a link that linked to their TIM article. What do you know, they have written an english version of the cooler roundup.

So much easier to read now.

Here.The guy who made it is a friend of mine which made this as a hobby. Most of them are bought from his pocket, some are borrowed from friends and others are samples send by manufacturers. You have all flavors of testing from fanless to extreme air conditions. Testbed is the same for all of them. IMHO this is the most complete and comprehensive super-coolers roundup ATM.
Enjoy.

Could you do me a favor? Ask him: Between the Megahalem and IFX-14, which one would he choose for his system? Thanks

 

[DaMulta]

WTF is that thing?

I bet you could slap a 200tec under it without touching it's cooling power!

 

[burebista]

Could you do me a favor? Ask him: Between the Megahalem and IFX-14, which one would he choose for his system? Thanks

I don't have to ask him because I know. Megahalems without doubts. IFX is doesn't worth the money.

WTF is that thing?

It's mighty Scythe Mugen 2.

 

[Arrakis9]

^^^ beat me to it

WTF is that thing?

I bet you could slap a 200tec under it without touching it's cooling power!

its a scythe cooler, dont remember wich model though

 

[mastrdrver]

^^^ beat me to it

Not really, I thought I had clicked on a link to the english version of the site. Instead, I clicked on the link that google translated to english, poorly and I didn't notice, and I thought it was the english version of the site. :shadedshu

burebista,

Could you tell me what he was saying at the end about the two IFXs he had. Google went to extreme fail at about that page and I'm not sure what he was saying. I'm guessing it was something about a change in the hold down, lapping, and differences in temperatures but couldn't quite put together what he was saying about it.

Also, does he plan on doing a similar review on an i7 setup? I've seen people say that the IFX doesn't really shine until you put it on an i7 because it generates a lot more heat and can separate the field of available coolers.

Last, does he know when the Noctua ND-14 will be available? I'm really interested in this since it is similar to an IFX but with Noctua's better retention bracket.

Thanks again.

 

[DaMulta]

What I walked away with that review was that the 120 extreme is still the king cooler.

 

[mastrdrver]

It will probably be tomorrow before I can sit down and really read it. I just kind of breezed through it earlier. It looks like he gets very thorough talking about fans and spacing of the plates, etc.

He hints a little at a couple websites saying that the IFX is the best cooler and that he never saw that. I'm guessing one of those is Xbitlabs. They seem to show the IFX beating anything that comes its way on i7. With that said, I don't think some realize that they test in a closed case. This is really important to me since my only system resides in my case. Sure testing cpu coolers on benches gives you a good idea of how well it can take care of the heat from the cpu, but tells you nothing about whether it will still be able to do the same in a heated, closed environment. Start testing a heatsink in a closed case, and you will start to heat soak it just from the warmer air. Suddenly, effective cooling surface can play a huge roll in whether a cooler can deal with, not only the heat from the cpu, but also from other internal components. Also, how quickly you can remove the heat from the cooling fins coming from the cpu can play an important roll. If your more effective than your competitor at the same surface area, an option becomes available to be able to reduce the size of the cooler. This way, you will be able to sell your cooler cheaper while being able to cool just as well.

 

[burebista]

Also, does he plan on doing a similar review on an i7 setup?

I dunno about i7 because he's broke ATM after spendind so much on those heatsinks.
But he have an old quick test on i7 with TRUE, TRUE Copper and Megahalems.

Last, does he know when the Noctua ND-14 will be available?

Here I can answer. I just spoke with our Noctua importer and he told me that he'll have it in stock at the end of this month.

Could you tell me what he was saying at the end about the two IFXs he had.

Oh man, you've killed me. I'll try to translate. Hope it will be better than Google Translate.
On page 48:
IFX14 - a study between samples differences
You'll say that I pay too much attention to IFX14 in this review but I do this because he's classified among the best coolers by some reviews sites (one of them well renown) if not THE best cooler.
In this article even it was lapped, modded retention system and a study about fan position impact on temperatures it can't cope with others super-coolers.
It remains only one problem to study to clear up things: a possible performance difference between different samples.
Based on my experience with multiple Thermalright coolers (2xTRUE Black, 2xTRUE and 2xCopper) I can't say that I see a notable difference between samples (1°C which can be easily put in error margin) but we must cover this matter with different samples.

So I've received another IFX-14 for comparison and I'll put against a TRUE modded with IFX-14 retention bracket to have a better idea about performances. Now let's take a look to the second IFX-14 which is Intel BP variant and includes socket 1366 and 775 retention bracket.

pictures

We have to deal with the same convex base specific to Thermalright and like the first IFX heatpipes are welded on base and fins and quality is good (no fins are moving). We notice a different retention mechanism, IFX is well crafted but it have same lack of pressure, you don't feel that you can squeeze it as opposite with Megahalems where you can feel the pressure. It's a pity because that mounting system is well done but unfortunately Thermalright didn't learn nothing from his mistakes.
I was disappointed to see that after such a long time TR is not able to make a good retention mechanism from all points of view. Retention mechanism from IFX works like a charm on TRUE so the pressure was better on TRUE and IFX mounting system does a better job compared with TRUE original mounting system which has a bad efficiency especially after lapping (I've gained ~1.5°C after modding original TRUE retention system).

pictures

For this test I've used HIGH TDP setting (3711mhz@1.488v) and two 3000RPM fans in single and dual configuration as you can see in pictures below. Prime95 Small FFT was running for 10 minutes after the test system remains 5 minutes in desktop with CPU 100% in idle and ambient was ~22.5°C. Second IFX was tested with original mounting system (unmodded) and unlapped, a disadvantage of course.

pictures

Page 49
pictures

As we expected (because new IFX has a convex base and unmodded original retention system with bad pressure) the pattern of TIM layer on base was pretty bad, cooler base has a bad contact with CPU IHS. The pattern of IFX modded was almost perfect, TIM layer was very thin (left picture). The results was pretty surprising for me, I didn't expect them and they're not a good thing for Thermalright and you'll see why.

Results
Graph
IFX-14 MODDED, 2×3000 rpm
Screenshot
IFX-14 MODDED, 3000 rpm
Screenshot
IFX-14, Second sample, STOCK, 2×3000 RPM
Screenshot
IFX-14, Second sample, STOCK, 3000 RPM
Screenshot
TRUE modded 3000 RPM
Screenshot

We can see that second IFX sample (the new one) works slightly better than modded one and keep in mind that it's not lapped or modded (TIM pattern tells all). We can see also that IFX mounting system works perfect on TRUE and behaves like a modded mounting system for TRUE (we gain another ~1.5°C, the gap between TRUE and IFX is larger as in main test where TRUE has original mounting system, the bad one).
We can have 2 clear conclusions:
- there are differences between IFX samples (at least between those two tested by me). If we modding IFX we can gain ~2°C (~1°C from lapping and ~1°C from modding mounting system which is better than stock TRUE which can gain ~1.5°C with modding). Those differences between samples put Thermalright in a bad position (beside convex base and weak mounting system we have now different samples performance) and can show or possible problems on serial manufacture line or from material quality used in manufacturing. Of course I cannot conclude only on those two samples but it's a wild guess.
It's possible to see those samples differences at other brands but I don't tested them so I can't say anything about that.
What I can say is that i have a TRUE from first series and from newest one (LGA1366) and I saw some improvements at new series, especially the fins are better soldered and they don't move but I don't see performance differences between series.
- It isn't possible that IFX-14 to be the best cooler, the difference between IFX and TRUE in bench conditions is very big and no way it can't be covered. As I said IFX architecture shows it's limitations (I'll made a more complex study about the efficiency of that architecture when I'll have all coolers based on this architecture: Noctua ND-14, Asus Triton 88, Thermalright Arrow).

From my experience I can say that a good cooler works fine from the first mount (Zalman 10X or Megahalems for example) and after lapping, modding, remounting, retesting I didn't see a "miracle" in performance.

That's it. Hope you understand better my translation than Google one but bear with my Enghlish though.

 

[mastrdrver]

That's awesome dude, thanks!

BTW, I think I'm going to wait out for the Noctua if it really is coming by the end of the month. I had read over on XS that someone said they were working on a review and it was suppose to come out end of September. Here's hoping for end of October.

 

[InTeL-iNsIdE]

I don't have to ask him because I know. Megahalems without doubts. IFX is doesn't worth the money.

It's mighty Scythe Mugen 2.

its a beast, thats what I'm running, very very effective

 

[burebista]

Yeah, a pain in the a*s to mount it but top performer.

 

[mastrdrver]

Now that I've had times to read it and really look at the graphs, I think his testing proves even further that most of these coolers are not even being pushed by the chip. Most of the two fan configuration graphs prove that the heat sinks are removing the heat faster than the cpu can put it out. This is really evident on the medium TDP with the 1500 rpm fans and he points it out too in the article. Problem is, if all the heat sinks are not being pushed by the cpu with active cooling, how is it possible to discern a winner except the one cooler that is the cheapest? Why spend the extra money on the Meg when the Ninja 2 will do the same job for cheaper?

So, we move on to the High TDP. The expectation is this is where the pack will be separated cause we are really starting to push the processor. Again though, even when moving from one to two 1500 rpm fans, a lot of the heat sinks have almost no response. Moving to the 3000 rpm fan give a good drop in temperatures over the 1500 rpm one because, I believe, the fan type has changed to one that not only moves air, but can create a lot of air pressure. Proof is that when a second fan is added all of the heat sinks respond by dropping 2C. All of them, from the True copper to the HR01+. Yes there is a little bit of a spread between first and last, but all of them were removing just about all that the cpu was putting out. Want the best performer for that fan? Go get a Ninja 2, it will do just as good a job and cost a lot less.

All the coolers he tested were treating the cpu like the fat kid in the park taking the ice cream from the baby. What his tests really prove is that even the hot 65nm C2Q can be cooled by the mildest priced cooler. No reason to run off and get a Meg when you can get a Ninja 2 to do the same when equipped with the same fan. Plus, none of the coolers were made to run in an open air case without a fan. It would have been interesting to see the results provided by a 1000 or 2000 rpm Kaze since given the same rotation as the slower fans he tested they will give more pressure and could possibly make the coolers with the denser fins perform better.

This is also what i7 has brought, it has separated the men from the little girls. My Vendetta 2 was rocking no problem with 1.3v on my Q9400 with the stock 1400 rpm fan. RT never would report more than 55C when running Linpack trying to find stability. If I tried to run the V2 with its fan on my i7 with 1.3v, my processor would go out behind the barn and have its way with the V2 until the processor blew up. According to RT, the temperatures just fire off to 80+ with in seconds and quickly keeps climbing.

While he was really in depth with seeing how flat bases were, force of clamps holding, exhaustive checking of the TIM on the IFX to find a hold down problem, fin density, and others I think he really missed the important point that was right in front of him with the results: The Core 2 chips never really pushed any of the aftermarket coolers out there.

I think this is why it seems that all hell is breaking loose over how more and more people say that the IFX does a much better job on the i7 than the rest of the field. I don't think anyone really realized how little the Core 2s pushed the cooler, then Intel throws us the i7 and can generate the heat to separate the field that the Core 2s couldn't. Its probably also the reason that everyone is in disbelief about the results that Xbitlabs keeps getting with the IFX and thinks they must be doing something to give it an advantage since the cooler doesn't do that with any other chip out there. Not even the chips from AMD seem to be able to generate the heat that the i7 can and uses to separate the current field of aftermarket cpu coolers.

If you want something that is going to do a good job cooling an i7 with HT enabled on air, your going to have to get a cooler that has a large fin surface area. This will allow the heat to spread out over the cooler before it starts to be overcome by the heat of itself because it can't get rid of it fast enough. You can quote me on this, if the Asus Triton 88 and Noctua NH-D14 ever get released, they will perform close to what is being seen from the IFX given the same fan(s).

I'm still going to wait for the Noctua since the hold downs they use for their coolers are a lot better than Thermalright's plus I feel like I should be required to modify a cooler hold down just to get it to perform like it should. I'll spend my money on another cooler that can.

 

[burebista]

Thanks man. I really appreciate your willingness to read all that article through Google Translate and to share your ideas.

One of my desires is to made some tests in closed case. I told him long ago this and maybe someday he'll do a test in closed case.

Thanks again for your thoughts.

 

[mastrdrver]

I can understand why, when cpu cooler tests are done, they are done on a bench or similar. It is just sooo much easier when it comes to swapping out coolers, testing TIM spread, etc. The only thing putting the system in a case will mostly do is raise the ambient temperatures the cooler sees. I think an easier way than using a case is to setup some kind of box he could cover his bench with so he can easily remove it if he needs to do something. It sure would help raise the ambient and give a closer real world idea of what one could expect.

Yea, someone should help Google out as their translation is lacking to put it mildly.

I really did enjoy the read though. I think it would be interesting to see how the coolers stack up when you compare their estimated fin surface area or if certain ones with tight fins respond better to higher pressure fans at the same rpm. It could show something not normally seen.

 

[sneekypeet]

I can understand why, when cpu cooler tests are done, they are done on a bench or similar. It is just sooo much easier when it comes to swapping out coolers, testing TIM spread, etc. The only thing putting the system in a case will mostly do is raise the ambient temperatures the cooler sees. I think an easier way than using a case is to setup some kind of box he could cover his bench with so he can easily remove it if he needs to do something. It sure would help raise the ambient and give a closer real world idea of what one could expect.

Yea, someone should help Google out as their translation is lacking to put it mildly.

I really did enjoy the read though. I think it would be interesting to see how the coolers stack up when you compare their estimated fin surface area or if certain ones with tight fins respond better to higher pressure fans at the same rpm. It could show something not normally seen.

That "boxed" idea is what Tweaktown uses, and most sites use open bench testing as thier testing method with varying ambients from review to review. Cant think of the name of the site at the moment, but they use A/C as a way to keep ambients the same, but at an unrealistic 20*C environment.

 

[overclocking101]

So I'm confused. maybe it's just the translation but i couldnt understand some of it. very thorugh though

 

[burebista]

Man, tell me what you don't understand and I'll try to translate/summarize.

 

[mastrdrver]

So I'm confused. maybe it's just the translation but i couldnt understand some of it. very thorugh though

If your trying to read what details he talked about, I think it is something that isn't much talked about. He not only covered before and after lapping TIM spread, but also looked at clamping pressure. There were a few, like the Zalman, that needed a clamp mod otherwise he got worse temperatures than before lapping.

All HDT coolers were not flat and causes them to not perform as well as they may be able to. I think the Thor Hammer was the exception.

Talked a lot about how Thermalright needs to learn how to make a hold down and flat base. Even covered doing some things with the IFX, include getting a newer version since he had one with a different hold down, to see if he could get it to perform better.

 

[burebista]

Thanks man, you summarize pretty well his tests.