I am looking for suggestion for a CPU cooler for an i5 2500K system. I will likely use the onboard GPU, an SSD (at least for the system drive) and I am also considering the Seasonic X-400 Fanless. The mobo will likley be an Asus P8H67-M EVO or something similar.

Therefore the CPU cooler would be the only moving part in the system, assuming that it has a fan. Are there any fanless solutions suggested for this CPU?

I haven't decided on a case yet - does the case really matter assuming it is big enough to hold the CPU cooler? I would like the case to be as small as possible although I would like the option of adding a discrete graphics card in the future. Given the fact that I am unlikley to have lots of cards or hard drives then there really isn't much need for a larger case. Is the NSK 3480 suitable or is there something even smaller? Preferably without a PS if I am buying the fanless PS.

Or are there any mini-ITX mobos that would work for my needs?

I would go with a recent design cooler that is likely optimized for the Sandy Bridge.

There is a lot going on between the CPU assembly and the heat pipes leading away from the base. Different CPU assemblies have different thermal signatures (the clarkdale actually has two chips with different heat signatures on different sides of the CPU assembly).

A heatsink with a base plate designed specifically for an 1155 Sandy Bridge CPU assembly would, you would think, do a better job with it.

Older heatsinks with non-direct contact bases would in my opinion be next in line.

I would stay away from heat sinks with direct contact designs that were originally designed for 775 or 1156 CPUs, even if you can get 1155 mounting kits for them. Their designs or just to likely to be optimized around CPU hot spots that are no longer where they used to be.

Scythe's Mugen 2 is the best cooler SPCR has come across when using their reference fan. It performs really well with low fan speeds and even better with two fans.

I'm using one with two Scythe Slip Stream fans at 675 RPM, they're silent at that speed and the my 6-core AMD CPU stays near room temperature.

For fanless operation I'd definitely use Scythe's Orochi; It's the largest cooler on the market.

Neither of the coolers I mentioned use direct touch pipes.

What will you be using the system for?

How about one of the best? Though if you factor in cost, it does provide the most bang per buck.

Still, I would be more partial to selecting a new cooler, mostly likely designed for a newer CPU, Something like the new Noctua downdraft.

That being said... seems to me even if the Mugen's base is optimized for another CPU (and only Scythe engineers know for use) it has so much excess cooling power, it is still going to get the job done and do it as better than most other coolers.

Thermalright HR-02, Scythe Orochi. Could work. Depends. I'd add a fan to be sure. Doesn't need to run all the times. Just if things get hotter than you'd like.



Depends on what kind of heatsink you're going for. Anything smaller than the NSK 3480 will likely place the PSU over the CPU area, thus greatly reducing the height allowed for the heatsink.



Sure. You seem to need a CPU socket, a couple of RAM slots, 2-3 SATA ports and maybe a PCIe 16x slot for future graphics ... that's covered by like every H67 Mini-ITX board out there.

Are there all that many "mini" H67 boards out there? I haven't seen that many - I believe Zotac has one - the H67ITX-A-E and Asus is set to launch the P8H67-I Deluxe which is supposed to be Asus' first Intel based mini mobo.

Which heatsinks are those that are specifically designed for 1155? When I look at the Recommended Heatsink page it doesn't look like it has been updated for socket 1155. And I thought that the Sandy Bridge CPUs used less power - do you really need a big monster with multiple fans to cool it?

You don't really know, you can only guess. In order to update the reviews for a new socket, they would have to go through a lot of work. And the difference might not be that great. I don't know that anyone has tested it. I think I picked up this subtlety from the Frostytech testing site.

I think that is probably a safe assumption.

Now, I've heard that the 1155 mounting holes are the same as for the 1156, so is anyone actually coming out with a cooler specifically for the 1155 socket? And do manufacturers really optimize so tightly to a specific socket? Would they tell you if they did? Looking at the recommended cooler list, the vast majority support at least two sockets, usually an AMD socket and an intel socket, but often 775, 1156 and 1366. Given that for each of those sockets you have different dual and quad (and hex) core cpus, how could you optimize for specific hot spots? Sounds a little fishy to me.

Power draw and thermal output aren't the same. So better safe than sorry.

No one does specific 1155 coolers. Most of them are universal coolers (775/1155/1156/1366/AM2/AM3).

Of course they wouldn't tell you. Just like they don't tell you everything about the food you buy (and please don't give me any stupid talk over the food... Texas beef producers actually sued Oprah in Texas for saying unkind things about beef... they have a special Texas law that permits them to reach out and sue healthnut Yankees that attack attributes of their agricultural products).

I picked this up from the Frostytech site. They seemed to think that specific direct contact coolers work better with some CPUs and not so well with others depending on the placement of the heat pipes on the base.

Seems to make common sense to me. The engineers who build and test the heatsinks should know. As would anyone else who would test for this.

There are a lot of decisions and tradeoffs the design engineers make. You think Prolimatech engineers get their superior performance by ignoring details like this? They struggle over every tenth of a degree advantage they can get. It's pretty competitive out there. The rational thing for them to do is to determine what CPUs are being used to do reviews, then optimize their design to those CPUs. So do better on the reviews... but not admit to it avoiding loss of sales for use on other CPUs.

I, myself would really like to see someone do some testing on this. The problem is this is the kind of testing that is not as exciting as testing the latest heat sink. And it is an expensive type of testing. You need at least two different testing platforms.... and a range of different cpu coolers to do a proper job of testing for it.

1155 and 1156 CPUs share the same size and shape, so any cooler for 1156 will fit a 1155.

There is no question that is true. The question is the design the the heatsink base and the placement of the heat pipes - and how that affects the efficiency of the transfer of heat from the surface of the CPU unit to the heatpipes. Do different base designs work better, or worse, with different CPU units.

So the issue is not one of compatability, but of differences in heat transfer performance at the base. The highly asymmetrical heat signature of a 2 core Clarkdale is different than that of a 4 core 1155 CPU, or a 3 core AMD, or a 2 core Sandy Bridge, or a 4 core 1155 Ivy bridge. Some of these heat signatures are larger some are smaller. You can design a base that handles all of them reasonably well, or you can figure out which chip is being used by most of the heat sink reviewers and optimize around that chip.

I don't know how many degrees or tenths of a degree you're talking about. I do remember someone doing an A B comparison with Thermalright's 80 pound pressure mounting system. My recollection was they picked up a few degrees just by changing the mounting system.

Then direct-touch pipes are indeed less than optimal because a single surface would spread the heat to all of them.

My understanding is that when they are a good fit to the CPU unit, they are better. But if the placement of the heat pipes misses the hot spots, it impairs their performance.

Of could please keep in mind, I don't pretend to be an expert on this and my sole source of information is the "internets" and we all know how hit and miss that can be.

I really doubt there are socket-specific coolers. The die of the CPU is always in the middle. There's a heatspreader on it. The heatpipes of any cooler run through the middle.

Things that could matter are whether the base of the heatsink is concave or convex (although these issues are generally debated to affect all sockets) and how well the mounting mechanism is designed. The latter in light of the fact that AMD's mounting isn't squared, like Intel's. This leads to a restriction on AMD boards where a lot of coolers can only be mounted parallel to the mainboard.



Intel has one out already and Gigabyte has announced one or two. There's never been all that much choice when it comes to Mini-ITX board. They're still a niche. But each one coming out eventually should cover your needs.

Frostytech thinks many stupid things. Anything they say is invalidated just by looking at this page :
http://www.frostytech.com/top5heatsinks.cfm

TOP 5 Heatsinks by Low Noise* ( *at maximum fan speed)
5th place : Intel Stock Core 2 Duo

Uhm, yeah, sure, whatever.

Direct contact coolers are a very specific type of cooler with specific issues... Pretty much the only quasi-direct contact cooler which does deliver is Corator DS : http://www.hardwarecanucks.com/wp-conte ... r_ds_5.jpg . In case of heatpipe numbers, the problem with 4 heatpipe direct contact coolers is that the two side heatpipes are usually so far out from the center of the cooled area that they don't make any contact with CPU, or if they do then it is very small contact. That is why usually 3 heatpipe direct contact coolers beat the 4 heatpipe versions (case of Xigmatek 1284 vs 1283). And that is why Corator DS is such a great cooler - because they are as close to direct contact, but they still have a single surface area, which can deliver heat to all heatpipes.

We can talk that some coolers are better for specific models (dualcore vs quadcore vs hexacore), but not for socket. Dualcore i3 will have different heat distribution that Quadcore i5 or hexacore i7. But unless you use a very stupid design direct heatpipe contact coolers like the Xigmatek ones, there is absolutely no difference.

That sounds plausible. But I suspect the world is as simple as you see it to be.

And how would you know without testing. Just a wild guess, but you have not done any controlled testing yourself have you?

It is also plausible that most of these tower coolers should perform about the same. They don't. They all look about the same. But they aren't. I suspect things are not as simple as you believe.

There is a lot of intense competition out there over thermal performance. Though much less competition over noise performance.

Frostytech's core competence doesn't revolve around noise. They do seem to be very knowledgeable about thermal matters. Their thermal testing results by and large are consistent with SPCRs and that of other testing sites.

I wouldn't be so quick to call Frostytech stupid... nor so quick to crown yourself so much more knowledgeable than them.... at least when it comes to thermal performance.

Of course I have been wrong before.

1) Look what Frostytech uses for cooler testing. No, it is not a CPU. Frostytech tests how good these coolers cool a piece of copper (38x38mm@125W for AMD and 30x30mm@85&150W for Intel). Sorry, but in my eyes that is not a relevant test. It tests the contact area of a cooler and heat transfer of it - but not how well it cools what it is built for - CPUs. You can't put a 125W thermal load on 38x38mm area and expect it to behave like a normal CPU does. Take a i3 530 for example :
http://www.mobilewhack.com/wp-content/p ... corei5.jpg

Big area is a GPU, small area is the CPU. This is connected to the IHS. Will be the thermal load same at the whole IHS? Of course not. This invalidates their results. It is possible that their results are consistent with other testing sites, but they are not testing CPUs, they are not doing real world tests. It is more about what is under the IHS than about the socket. Core i5 750 will have a completely different thermal signature on IHS than Core i3 530, due the different position of the components inside - while both having same socket.

I don't call Frostytech stupid, i say they say many stupid things. They are also irrelevant in case of CPU cooling. They are good at copper cooling tests tho. If the tests are consistent with CPU cooling tests, it's their luck, but again - they are not testing CPU cooling. They test copper cooling.

2) In case of Core 2's, Xigmatek 1283 (with 3 direct touch heat pipes) was better than Xigmatek 1284 (with 4 heat pipes). Why ? Because 3 heatpipes made more contact with CPU than 4 heat pipes. Look at reviews where they review both of them, 1283 consistently beats 1284. It is not my statement, not my review - there are dozens of them. 1284 can beat 1283 only if all heatpipes make contact - which can be case of Socket 1366 CPU's, but not the smaller ones.

There are 2 things which differentiate the direct touch coolers :
a) surface quality - this is where many of direct touch coolers fail, if the surface is not perfect then you waste lot of potential with thermal compound which needs to be used to have any contact at all.
b) useless heatpipes - 4 heatpipe direct touch coolers work only with bigger contact area, which needs to have even thermal load - and even with i7, you have problems achieving that (look how the CPU looks bellow the IHS). For anything smaller, 2 of 4 heatpipes make less than ideal contact with IHS, making them perform well bellow their potential. That is why direct touch is not a holly grail of CPU cooling - because not all your heatpipes don't make contact.

Long story short, there is just a very, very small percentage of use cases where 4 heatpipes make sense for direct touch coolers, thus limiting their best performance to the performance of 3 heatpipes for majority of use cases, which means they are not good for Socket 1155 either (surface area of IHS is not big enough).

PS: I'm not all knowing, but there are simply some things which makes Frostytech pointless - not using CPU for CPU cooler testing, stating that stock Intel cooler is silent,... There are better sites, better reviews for CPU coolers than Frostytech. You are writing on forums of one of them

I think that there are few coolers that are actually designed based on the architecture of the CPU. Many have brackets that will fit most CPU's as well. I'm sure there are many companies that just slap something together doing very little testing.

i wonder how much orientation plays with direct touch heat sinks with the rectangular die of new intels?

Just pick a top runner CPU cooler on the SPCR recommended list, dial down the fans on it to the lowest possible and reap the reward of never having to worry.
I run the computer in my 1) in my signature on normal range very silently. It's only when one of my storage HDDs spins up that I notice any noise at all.
No noise/heat problems and no worries at all.

Edit: I run that computer in a old case -> CM Stacker. Google it, it doesn't block any noise at all -_-

I would go for Freezer 7 pro rev.2. It's the best selling CPU in neweggs so it cant be wrong. lol

Do you want to go for direct-touch or just the regular CPU cooler?

Bunch of misleading comments in this thread.

Here's what I have: viewtopic.php?f=14&t=63878

I would definitely recommend the Hyper 212 Evo also, for it's low price and good performance. I built a machine for my collegue, also using the i5-2500K and it's very quiet and the temperatures are very low. It can be controlled by the PWM fan controllers on e.g. Gigabytes motherboards.

The Freezer 7 isn't well-suited for low noise cooling because it has very tight fin spacing.
http://www.silentpcreview.com/article285-page6.html



Agree with this totally. Just pick what you can afford and never worry again I personally prefer the Scythe Mugen 3 ) for the same reasons that made the Mugen 2 an editor's choice here at SPCR: Great thermal performance at low airflow, superb stock fan (can be run as low as 300 rpm with good pwm-controlling (!!!)) and then it's fairly cheap. The only problem is the installation. With the Mugen 2, I experienced that it was much easier to have a second person aiding you. This person only needs to hold it so no prior experience is needed for this person.

I would go for Freezer i30. its a great cooler for i5