How To Choose A CPU Cooler

[JackNaylorPE]

Top mounting is still the most common option though, and radiators placed there usually work best with the fans underneath, blowing upward.

1. This goes against CLC manufacturer's written instructions, the laws of thermodynamics and common sense. I understand the thought process, we all learned in 8th grade Earth Science that hot air rises so lets push air up. Earth Science did not include ceiling fans. Air rises because warm air is a fraction of a % lighter and it does so slowly in the temperature range in question. This force pales in comparison when mechanical force is involved as is evident to anyone who has ever sat below a ceiling fan.

2. Now onto thermodynamics. The effectiveness of a cooling system is directly proportional to Delta T. Example:

Ambient Air = 23C
Coolant Air = 33C
Interior Case Air = 28C

Using Ambient Air as Intake, cooling is proportional to: Air flow x (33C - 23C) or 10Q
Using Case Air as Intake, cooling is proportional to: Air flow x (33C - 28C) or 5Q

It is therefore an inescapable conclusion that using outside ambient air is twice as effective than inside case air in this example. No matter what the temperatures are, ambient air will always be cooler than interior case air. You are talking about buying a CPU cooler in this article ... not a MoBo cooler, not a SSD cooler, not a HD cooler ... why ? because these things do not need cooling and are not impacted in any way in the temperature range in question.

The only item that can be impacted in GFX cards and testing with non-reference cards shows no evidence of that occurring ... well except of you use blower style coolers. When sizing a radiator, the calculation methods "that work" use a factor of 60% of theoretical load. part of that is due to the fact that not everything is at peak load at the same time .... the other part of that is is that everything is radiating heat off its surfaces.

Concerns about graphics card heat being ejected through the top-panel radiator have been addressed in my own builds by using graphics cards that expel most of their heat through a rear-panel expansion slot, as seen in the silver card above. Yet graphics card reviewers frequently recommend dual-fan or triple-fan cards such as the black card above, focusing entirely on the improved graphics temperature-to-noise ratio without any concern for the impact that waste heat has on every component above the graphics card. Because I review both cases and CPU coolers, I consider graphics coolers that blow heat into the case to be defective.

The correct word is "effective". First off, look at GFX cards reviews and blower style coolers always leave their cards running hotter, then the normal style. Second, no card cooler exhaust hot air exclusively in or outside the case. No matter the cooler style, some heat is conveyed internally and some externally ... blower style cooler simply send more out than conventional coolers ... and pay a thermal price for doing so as they can move less air out thru those little tiny openings.

Add a fog machine to your test bench and you will see immediately that the situation you describe simply never happens. Those 3 fans on those GFX cards are bot blowing air up.... with the rad ran set as intakes, they are blowing down. So if you installed your radiator properly, "graphics card heat being ejected through the top-panel radiator" never happens. The fog machine does show what is happening .... all air is exiting thru the grilles or fan on the rear of the case ... none of it is coming back in. You "created" the perceived need for a blower style cooler by improperly setting up the rad fans as exhaust.

3. Today's non-reference GFX cards (well nVidia anyway) have little to gain from water cooling other than noise reduction. Now most folks build there box with a CPU cooler ... and no thought to adding a cooler to anything else. So if there was no concern here, why the sudden concern about what 28C radiator exhaust air will have in your components ? Reviews of the cards like the Gigabyte G1. MSI Gaming etc, show no evidence of thermal throttling even at the highest OCs. What component is it exactly that we are trying to protect here ? We have 6 thermal sensors connected to a digital display in a 5.25" bay and an infrared thermometer. Can't seem to find this issue of components being heated up to a point of concern. Mobo, storage sensors report everything in low to mid 30s

4. The other factor being overlooked is case air turnover. When you move to liquid cooling, you change the case air turnover. With an air cooler, you basically move the air from the CPU to the case interior, leaving the case fans to take it from there. When you add a liquid cooler, you have the exact same condition. Is every air cooler out there leaving every case in a deficient condition ?

People rarely think about one of the most important cooling features of their case, the rear grille. When you take your typical 2 fans blowing in, 1 fan blowing out case setup, and add a 2 x 140mm radiator, you are greatly increasing case air turn over. Now fresh air is coming in at a much greater flow rate so the condition of the air inside the case is much cooler than it was with the air cooler.

Conversely, when ya add that top radiator blowing out .... you have 3 fans blowing out and two blowing in ... and those 2 are restricted by air inlet filters. With well more than twice the air blowing out as in, you now have a) dust coming in thru the rear case grille, b) your hot PSU exhaust is coming in with it and c) all that hot air from your "blower style" GFX card(s) is coming right back into the case. Again, this becomes readily evident when using a fog machine.

Some of our test have even shown these massive coolers outpacing a few of their liquid rivals.

In almost every case I have seen, the better air coolers outperform their CLC rivals, when compared "apples and apples" (noise and cost equalized) it's not even close. CLCs need to make up for the weak pumps and aluminum rads buy using extreme rpm fans which result in excessive noise ratios.

The H100i loses by 3C while being 12 times louder. The X61 wins by 1C wile being more than 8 times louder. And both cost substantially more.

No arguments here against shipping a PC w/o a large air cooler inadequately secured .... but shipping a water cooled system is not without risk either. But in a article entitled "How to Choose a CPU Cooler", it's going to apply more to those building a box as opposed to ordering one as the system assembler has few options. But yes, if handled abusively, the risk of damaging a PC being shipped exists. My son's made it back and forth to college 4 times a year with his PC (2 lb cooler) in the rear deck or on the roof rack of his off-road jeep whose suspension resulted in quite a rough ride.

Custom loop cooling will always outperform air on any level, with properly select components, other than price. CLCs on the other hand, just about all suffer from inherent weaknesses whereby:

a) almost all can't match their comparably priced air coolers
b) almost all are limited by weak pumps and aluminum radiators
c) almost all create excessive noise levels, the ones that don't suffer thermally from weaknesses of a) and b)

 

[kulmnar]

I had a card with a reference cooler (GTX580) for 4 years and I regretted the jet engine noise that it made every time I played an FPS game. Now I have an EVGA GTX980 ti ACX 2.0+ with two fans and I couldn't be happier. The CPU temps have not increased and I have very little noise in my case. I will never buy another reference cooled card again.

 

[JackNaylorPE]

EVGA is a bit tricky here.... The competition's non-reference cards come with a non-reference PCB (improved VRM, chocks, power delivery, thermal pads, heat sinks on chips etc) and a non-reference cooler. Historically, the EVGA SC series comes with a non-reference cooler but a reference PCB.

 

[Crashman]

You're implication is that a case should be made noisier to remove GPU heat, and use a quieter GPU cooler to compensate. But if you blow the heat out the back, the extra noise can be taken care of by a case with fewer, more strategically located vents. And the 3C difference meant something to me in the SBM series 😀

The assumption of that implication are wrong. I have built many open air based GPU systems with great air flow cases and near silent operation even under load. The specific testing I mentioned was with a Fractal Arc mini using a Sapphire 290X Tri-X card. The case fans where three 120 mm fans, 2x front and one bottom for intake and all set at 5 volts. The system was silent for all practical purposes.

As for the 3C, are you referring to 3C making a difference in ambient case temp or specific component? Finally note I mentioned the difference would be 3C IF a blower cooler had a perfect ability to remove heat, which none do. Also that 3C was total case temp rise, it includes heat from CPU, motherboard and so on.

Again I am just saying I think your position that open air coolers are defected is wrong.

3C at the CPU in a large well-ventilated case. In that style case, case temperature scans would probably reveal a warm air spot just below the CPU merging with the one behind the CPU. It's a lot worse in a Mini ITX case, which is why blowers went from a want to a need when I was building those.

 

[Crashman]

Top mounting is still the most common option though, and radiators placed there usually work best with the fans underneath, blowing upward.

1. This goes against CLC manufacturer's written instructions, the laws of thermodynamics and common sense. I understand the thought process, we all learned in 8th grade Earth Science that hot air rises so lets push air up. Earth Science did not include ceiling fans. Air rises because warm air is a fraction of a % lighter and it does so slowly in the temperature range in question. This force pales in comparison when mechanical force is involved as is evident to anyone who has ever sat below a ceiling fan.

We have test data to confirm that bottom-up works better in many cases than top-down. Of course that assumes you're using an effective blower-style GPU cooler 😀

 

[bit_user]

They're making them because you're buying them. And you're buying them because people are telling you to. And people are telling you to buy them because they're quieter. But then you need more case ventilation, and that means you lose most of your noise advantage.

In other words, if you were buying blue cards, they'd be selling blue cards. If you were buying cards with rusty metal covers, they'd sell you cards with rusty metal covers. Engineering has nothing to do with it.

If they were such a superior solution as you claim, you'd think there'd be at least a few. Reviewers would rave about the superior results from these high-end blower cards, and some people would buy them and sing their praises, thus creating demand for more.

It's like when the entire High School Auto Shop

That's a poor analogy, because you're likening a simple metric to the result of complex themodynamics and airflow interplay, with a lot of incidental variables. If you want to convince people of your point, try using hard data - not insults.

I take your downvote as a compliment, it means I've gotten my point across.

Apparently, you're not getting mine. I guess we're done, here.

 

[bit_user]

3°C on the graphics card can be the difference between archiving the maximum boost speed and a reduced clock if these 3°C fall into the threshold.

I actually cared mostly in the System Builder Marathon because it was the difference between 4.4 GHz and 4.5 GHz in a performance competition.

I think the point people are making is that you're just overclocking your CPU for bragging rights. But if you actually cared about game performance, how many games would be CPU-limited on that system?

The purpose of such how-to articles is to provide practical, real-world advice. If you're steering people towards inferior GPU cooling so they can overclock their CPU (which they might not even do, at all) by another 0.1 GHz, that seems kind of missing the point.

 

[jtd871]

Blower-style GPUs have 1 fan, need to move air along the long axis of the card, and then have to exhaust it through stupidly small openings dictated by the IO (DVI must die for this reason alone, at least).

A well-designed open cooler uses at least 2 fans that can force more air (at lower fan speed and thus noise) over the heatsink fins.

Not that open coolers can't be gimped by fins that are oriented the wrong way (along the long axis of the card), shrouds that might look 1337 to marketing types but actually *block* airflow (yes, I'm lookin' at you, Zotac) off the fins, noisy fans, aggressively high rpm fan profiles, poor airflow to the fans, etc.

 

[Crashman]

The purpose of such how-to articles is to provide practical, real-world advice. If you're steering people towards inferior GPU cooling so they can overclock their CPU (which they might not even do, at all) by another 0.1 GHz, that seems kind of missing the point.

I'm steering newbies to the solution that's most likely to work in the most situations. You start out with a mediocre or compact case and suddenly these open vent cards become a major heat contributor to the case. I'm not going to assume they'll go out and buy a bigger or better-ventilated case, I'm going to make the recommendation that works with the most systems. How's that for practicality?

 

[bit_user]

I'm steering newbies to the solution that's most likely to work in the most situations. You start out with a mediocre or compact case and suddenly these open vent cards become a major heat contributor to the case. I'm not going to assume they'll go out and buy a bigger or better-ventilated case, I'm going to make the recommendation that works with the most systems. How's that for practicality?

Okay, this is at least a rational answer.

 

[Crashman]

I'm steering newbies to the solution that's most likely to work in the most situations. You start out with a mediocre or compact case and suddenly these open vent cards become a major heat contributor to the case. I'm not going to assume they'll go out and buy a bigger or better-ventilated case, I'm going to make the recommendation that works with the most systems. How's that for practicality?

Okay, this is at least a rational answer.

Don't highlight it, I occasionally enjoy the friendly banter with enthusiasts who are completely sure that if they're right I must be wrong 😀

 

[madmatt30]

I'm with crashman in that there's a legitimate case for blower style cards.

Going from a twinfan Galax exoc 970 to the blower style Asus 970 turbo knocked 10c off both my GPU & CPU temps under gaming loads in a Silverstone gd06 case which is compact & has limited airflow.

At the expense of noise yes but those results are worth it 100%.

 

[Crashman]

I'm with crashman in that there's a legitimate case for blower style cards.

Going from a twinfan Galax exoc 970 to the blower style Asus 970 turbo knocked 10c off both my GPU & CPU temps under gaming loads in a Silverstone gd06 case which is compact & has limited airflow.

At the expense of noise yes but those results are worth it 100%.

(shhh, situations like yours are why I labeled open GPU cooler designs as defective, fully realizing that a bunch of enthusiasts had already found workarounds that roughly equate to running your PC caseless with a fan in front of it).

 

[f-14]

internally venting hot air into a case that's not mostly made of screen on the front and doesn't have a top vent and a small rear vent 120mm or smaller increases throttling slowing down the gpu and defeating the purpose as well.
i've done a lot of builds for people that want internal venting, when the gtx200's came out internal venting was bad unless you had a mess front panel case like an antec 900 or 1200 and with an artic 7 freezer, the rivaled the noise a vacuum cleaner made behind a door, you couldn't sleep with the thing on even in the next room over and you couldn't watch videos with out having to have closed ear head set cans or turning your 5.1-7.1 speakers up to level 3 which the whole house could hear. (atleast with altec lansing speakers cheap speakers had to be turned up higher than level 3 just to make it so you could hear galadriel talking to frodo in lord of the rings.) that's a huge problem especially for students and apartment dewellers.
any claims of 3Cºis a joke as that requires over 35db of noise just to run idle, when you run a game, were talking hoover vacuum noise levels.
that's defective thinking, in a closed case, that's thermal throttling and big performance hit as well as great irritation when the screen stutters and jumps and skips as frame renders get dumped.
i know i've tried even in a big server tower case. i had to cut more vents into the case, which brought back the noise problem, then i had to install auto carpet and later replace it with floor carpet just to get the noise down to 28db, which was a huge fire hazard near such hot components and a massive dust collector that required cleaning out every 6 months when i noticed thermal throttling.

i moved on from cases at that point and just installed the system into the desk after that

 

[bit_user]

then i had to install auto carpet and later replace it with floor carpet just to get the noise down to 28db, which was a huge fire hazard near such hot components and a massive dust collector that required cleaning out every 6 months when i noticed thermal throttling.

I once installed noise dampening foam, although it was sold specifically for use in PC cases. It smelled bad, for a while, was annoying to do, slightly reduced the space inside the case, and reduced the thermal conductivity of the case. And I'm skeptical it even helped very much. So, I'll never do that, again.

But, to anyone considering this route, I'd recommend Dynamat (the vibration absorbing material sold for use in automobiles), as part of my problem is the product I used didn't seem very effective. I do think a relatively small amount, placed strategically, could cut down on case resonance. But I'd recommend only going this route as a last resort.

 

[TJ Hooker]

@Crashman "(shhh, situations like yours are why I labeled open GPU cooler designs as defective, fully realizing that a bunch of enthusiasts had already found workarounds that roughly equate to running your PC caseless with a fan in front of it)."
... If what you said was only meant to apply to edge cases like people trying to run dual GPUs in SFF cases, why make a sweeping statement labelling open air cooler defective, with no caveats?

 

[Crashman]

@Crashman "(shhh, situations like yours are why I labeled open GPU cooler designs as defective, fully realizing that a bunch of enthusiasts had already found workarounds that roughly equate to running your PC caseless with a fan in front of it)."
... If what you said was only meant to apply to edge cases like people trying to run dual GPUs in SFF cases, why make a sweeping statement labelling open air cooler defective, with no caveats?

(((this is a beginner's article, I'm sticking to basic principles that will be most helpful to people who don't yet have time to learn all the workarounds)))

 

[Japeth Valen]

So everyone has a different mindset for cooling a GPU and CPU. Saying that open air coolers on GPU are "defective" can really heat up the conversation about "Liquid vs. Air Cooling" topic.
Well Air Coolers really has it's own "pros and cons".

For me that has been in the experiment days on computers for years. I personally found that open Open Air Coolers can prevent GPU's from heating up so much. But I'm not saying that Blower Coolers sucks at cooling, it has a different explanation for that.
Like I said open air coolers(the black colored GPU above) "doesn't heat up so much." Because the heat of the GPU is spreading inside the case, that means if your GPU temp is 70c(Celsius) your other components are 55c to 65c. If you are an experienced Builder, you know how to counter this. But still a component that is heating more than it can handle is very dangerous.

Now for the Blower Coolers, it really heats up. But it heats up because the heat is being "held on place". That is inside the GPU case(the silver colored GPU above) that is "more safe on the inside of the case." Now if you are thinking that Blower Coolers are more prone to destruction, IT'S NOT. They are built to be heated up. So if you're thinking of not getting a Closed Case GPU, well think again. Because it is made to last OC(over clock). To last even running 24hrs in 90c. It really can perform.

But this article is for Liquid vs Air Cooling so this is just another comment.

For me I like Liquid coolers more, well that's because I like to push things a bit more to the edge. But that's my opinion, and we all have are own opinion and experience.

But like the author stated above, this article is for those that is new to the system of building rigs/computers. So all things stated will help them decide.