A second look at Arctic’s Liquid Freezer III using our updated cooler testing methodology

[Admin]

This article explains Tom’s Hardware’s new methods for testing and benchmarking CPU coolers, and features Arctic’s Liquid Freezer III 360.

A second look at Arctic’s Liquid Freezer III using our updated cooler testing methodology : Read more

 

[Cameron R]

Maybe run spell check before you post? What the heck?

 

[rluker5]

I like the controlled, documented, close to real world testing addition of including the GPU use in a closed case.
That's the typical heavy use case for my CPU coolers - gaming in a closed case. Adding the real world variables of case interior cooling while a large heat generator is cooking inside of it could add significant inaccuracy if a bunch of standardization is not implemented. That difficulty could be why we don't see much realistic cooler testing like this.

Part of the documentation could also be a pic of that pretty test setup with the fans not spinning. It provides some good information like case fan choice layout and airflow, and that the AIO is acting as an exhaust. That picture is worth a lot of words. The setup is very clean and well laid out, but also looks very mainstream.

All good choices for approximating user cooler performance.

 

[icelancersr]

Yup, those are the numbers that I am seeing in my gaming sessions with arc3. Spot on

 

[RubberMallet]

This 360 AIO is topping out at 257W cooled before thermal throttling.

The LFIII 420 AIO that I own, when pulling 260W average (no throttling) in CB R23, tops at 75C at similar ambient. If I increased the multipliers and allowed it to thermal throttle, it would be north of 300W.

I have a hard time believing that the 360 is so much less effective. I and others were hyper critical of previous review testing, and I still feel like Albert Thomas isn't the guy who should be testing coolers based on his defense of the previous test method.

That article was the reason I stopped reading TH and this article will not change my opinion.

This review is really just an indirect advertizement for the Tryx Panorama, which no one previously has heard of.

 

[Albert.Thomas]

The LFIII 420 AIO that I own, when pulling 260W average (no throttling) in CB R23, tops at 75C at similar ambient. If I increased the multipliers and allowed it to thermal throttle, it would be north of 300W.

Unless you're using the same conditions I am, results are not going to precisely the same. Are you using a similar style case, with similar style fans, set to the same fan speeds, with the same motherboard, and with otherwise "out of the box" settings? Do you also have a comparable relative humidity? And are you using the same CPU?

That article was the reason I stopped reading TH and this article will not change my opinion.

This review is really just an indirect advertizement for the Tryx Panorama, which no one previously has heard of.

I'm sorry that you feel that way.

This article is simply an explanation of the new testing methodology I've adapted.

"No one" has heard of the Tryx Panorama because it was just announced at Computex, but Steve from Gamer's Nexus took a look at it, see here:

View: https://www.youtube.com/watch?v=Td0XORmSJp8


While I find the Tryx Panorama extremely interesting due to its one of the kind display, it isn't a cooler I'd personally use for reasons I'll explain in the upcoming review which will include comparisons with other coolers that are also able to pass my maximum strength tests without throttling.

 

[MagicSwordKing]

Okay, this article made me hopeful and then I read it and utterly lost faith in Tom's Hardware for cooler reviews.

First, there's a lot of honestly irrelevant posturing about how your benchmarking methodology is better than unstated competitors. If you're gonna bring up bad methodology, you need to make sure your methodology is actually as solid as you think it is. There is still one, massive, inexplicable omission that tells me all I need to know about this: you don't test coolers on both AMD and Intel systems.

They have different thermal properties, they have different mounting mechanisms, and different behavior under load. What's more, it's especially indefensible in the case of a cooler such as the Liquid Freezer 3 which offers bespoke mounting hardware for offset mounting on AMD platforms in addition to the contact frame for Intel. Every single credible review of these coolers have noted the poor relative performance of the Contact Frame which has some engineering flaws that keep the cooler from reaching the same sort of performance that it can with AMD hardware.

If you want to review coolers for CPUs, the one thing you absolutely must do is test for both major CPU variations to provide some semblance of a complete picture. This review is still worthless for a solid percentage of the buying public, in fact, it's actively misleading.

 

[Albert.Thomas]

There is still one, massive, inexplicable omission that tells me all I need to know about this: you don't test coolers on both AMD and Intel systems.

In the past I used to test CPU coolers on both AMD's Ryzen 7700X and Intel's i7-13700K, but in general most coolers performed similarly well or poorly on both systems. For example, my Phantom Spirit 120 Review featured both CPUs:

https://www.tomshardware.com/reviews/thermalright-phantom-spirit-120-review

I do plan on testing coolers on Ryzen 7700X again in the near future, but I'll be using it for testing in a SFF setup - I'm leaning toward's Cooler Master's NR200.

 

[MagicSwordKing]

In the past I used to test CPU coolers on both AMD's Ryzen 7700X and Intel's i7-13700K, but in general most coolers performed similarly well or poorly on both systems. For example, my Phantom Spirit 120 Review featured both CPUs:

https://www.tomshardware.com/reviews/thermalright-phantom-spirit-120-review

I do plan on testing coolers on Ryzen 7700X again in the near future, but I'll be using it for testing in a SFF setup - I'm leaning toward's Cooler Master's NR200.

You still don't get it, you need to test on both platforms for ALL coolers, precisely because while there are not OFTEN differences, there sometimes are rather dramatic ones as is the case for the Liquid Freezer 3. Seriously, go look up any other reviewer's take on the Liquid Freezer 3 where they test it on AMD. It changes the calculus. This matters. This is not a corner you can cut, and you are cutting it. Pointing out the corners OTHERS cut does no one any good when you yourself are cutting one of the most significant ones you can.

Reviews of things like this aren't jazz, it's not all about the notes you don't play, you can't just skip doing the bits of the job you don't feel like doing for whatever reason. You need a representative sample which means evaluating their performance on AMD and Intel. If for whatever reason you cannot do this you can't pretend like your reviews have robust testing methodology.

 

[User of Computers]

You still don't get it, you need to test on both platforms for ALL coolers, precisely because while there are not OFTEN differences, there sometimes are rather dramatic ones as is the case for the Liquid Freezer 3. Seriously, go look up any other reviewer's take on the Liquid Freezer 3 where they test it on AMD. It changes the calculus. This matters. This is not a corner you can cut, and you are cutting it. Pointing out the corners OTHERS cut does no one any good when you yourself are cutting one of the most significant ones you can.

Reviews of things like this aren't jazz, it's not all about the notes you don't play, you can't just skip doing the bits of the job you don't feel like doing for whatever reason. You need a representative sample which means evaluating their performance on AMD and Intel. If for whatever reason you cannot do this you can't pretend like your reviews have robust testing methodology.

If you think you can do it better, then do it yourself.

 

[Albert.Thomas]

You still don't get it, you need to test on both platforms for ALL coolers, precisely because while there are not OFTEN differences, there sometimes are rather dramatic ones as is the case for the Liquid Freezer 3. Seriously, go look up any other reviewer's take on the Liquid Freezer 3 where they test it on AMD.

If that's true, I should be able to observe it when I test the 240 or 280mm LF3 on the SFF system I'm building.

 

[thestryker]

Introducing a simulated set of gaming testing is a welcome addition and I'm unsurprised by the results. In fact I'd be surprised if any air cooler can keep up with an AIO in these circumstances. From what little experience I have with the same systems running both an air cooler will only keep up when there's an airflow problem with the case that it helps with (or an airflow focused case like the Torrent).

On the topic of AMD testing I do understand what a pain it can be. To run full suites on both would certainly require two identical setups one for each Intel and AMD. The vast majority of the time it's not particularly important, but for any cooler that does have bespoke mounting it becomes a big deal. I think it'd be a good addition to run a baseline with a mid range/high end air cooler and AIO on AMD that could compare to any coolers that are tested that do have bespoke mounting.

There are holes in every review methodology, but I generally prefer the way the testing here is done. The problem here is of course that the results aren't directly applicable to AMD.

 

[MagicSwordKing]

If that's true, I should be able to observe it when I test the 240 or 280mm LF3 on the SFF system I'm building.

It is, but... that would also not be a valid comparison. An SFF build will have wildly different thermal characteristics. You'd need to then compare that data against an Intel SFF build. This is basic statistical validity stuff, you need to make sure ALL the variables are reasonably similar, ideally identical but that would be a rather heavy lift for one person. At the very least you need a similar case with similar ventilation with an AMD CPU to test. I get not wanting to gut your whole machine every time you test something new, I would describe that as a reasonable corner to cut so long as you aren't claiming to have maximal validity.

I get that you're a freelancer but if Tom's Hardware is going to have you doing this stuff they really need to equip you for it. It's a big enough company, it should not be a huge ask and if it is, honestly y'all have much larger problems. Seriously, I made an account here just to bring up this point, if it helps, show this to your editors and have them run it up the chain. More outlets should have good data, you're not far off from getting there, you just need to have valid comparisons so as to not get caught by outliers like this.

 

[Albert.Thomas]

It is, but... that would also not be a valid comparison. An SFF build will have wildly different thermal characteristics. It is, but... that would also not be a valid comparison. An SFF build will have wildly different thermal characteristics. You'd need to then compare that data against an Intel SFF build.

An SFF build would be slightly more thermally challenging due to the compact size, but I'm pretty sure that results showing how different products perform in it would scale similarly in a larger case - at least for the best products and strongest coolers. The lower strength products might struggle a bit more.

While I may test on different AMD and Intel platforms in future reviews, I simply don't have the time to test a cooler on four separate systems.

 

[Amdlova]

Methodology lol...
The IHS it's not the same on the cpus.
Even the amount of thermal paste can change the results.
The material of coolers base, fins and water can average.
The place where the tests are taken never will be the same as in your location.
Stop complaining about everything and appreciate the work and time to do those test.

 

[35below0]

Good article. Tells me what i need to know about the cooler and the test setup.

Don't know why people complain.

 

[ViperThrall]

Yikes, people need to chill. I use my 7950x to video edit and render with the Arctic Freezer III 360. As many reviewers have stated this cooler does not do well on Intel 13th and 14th gen. processors. Part of the problem is due to Intel cranking stupid voltages and TDP. The other part is the wonky mounting solution Intel “invented” for their latest design. My 16 cores run at over 90% on all cores and the highest temp reaches about 81C. The III is still relatively quiet with fans spinning up to about 88% percent. For $97 this unit is a steal. All three connections are separated out to the motherboard with PWM settings set within the bios. RGB looks good also;}

 

[heine svendsen]

You are not right about dry climate being better for cooling.
Humid air (at the same temperature as dry air) is heavier, and therefore can absorb more heat from the radiator.

 

[Albert.Thomas]

You are not right about dry climate being better for cooling.
Humid air (at the same temperature as dry air) is heavier, and therefore can absorb more heat from the radiator.

TIL, thanks.