Why you shouldn't water cool your PC

[greenhorn]

Hi, awesome video, but I have a few comments:

How did you measure your "dB" values? Normally this is done with the measuring equipment at a certain distance of the noise source, so how big was that distance approximately?. Next, the noise will usually have several frequencies in it, so it is important to know how you arrived at your dB value. Is it a dBA value by any chance? It makes sence to introduce weightings to each frequency component that take into account how the human hearing works.

Did you ever try using an AiO with the rad itself being anything else than aircooled (water, concrete, whatever)? That would make the rad act as a heat exchanger between two water cooling systems. This approach would also have been interesting in the context of the whole room watercooling project back in LMG's residential house days or for the project using a conventional room heating radiator for watercooling (in the latter case it would have kept all the gunk from the old radiator out of the cooling loop inside of the PC).

[Konsole1337]

Sweet video

They should make a follow up video

Pairing the best on the market Air coolers against custom water rigs

I think it would result in a interesting video tbh fam

[BigDamn]

4 hours ago, Captain Chaos said:

Perhaps because the AIOs already are more expensive and replacing the fans would only increase the price gap?  

 

The Noctua is $100 with 2 Noctua fans. If you take the AIOs and replace the fans (at $25 each), you're looking at around $170 for the H100i and $230 for the H150i (rough calculation based on local prices here in Belgium, so please don't shoot me if I'm $10 off whatever the price is in your local store).  That's not even close to the same price range anymore. 

This is true, however, you missed part of what I said. My point was to see the performance CEILING of each cooler, rather than the better value. I conquer that Noctua coolers ship with better fans than perhaps any cooler on the market, and at a reasonable cost. The reason this test would be useful is for those of us (including myself) who already have high performance fans laying around. It's not really an added cost for me since I already own them.

[DezGalbie]

I run a Ryzen 5 1500X on my PC overclocked to 4.1ghz. Even during the odd time where I do something to peg my CPU at 100% for extended periods (like DVD ripping, video encoding etc) then my Arctic Freezer 33 eSports Edition CPU cooler still keeps temps under 65C.

I got the CPU cooler for something like £30 (much cheaper than even the cheapest AIO I could find at the time). I honestly don't know why anyone would use a more expensive cooling solution for this kind of mid-range hardware unless they are super picky about aesthetics and think AIOs look a lot better.

[RejZoR]

No matter what numbers say here, I'm not buying it. I've had various top of the line coolers in the past and none performed anywhere near as good as even a single radiator AiO. And I don't think i7 920 at 4GHz pumped out more heat than 5820K at 4.5GHz. Not to mention how quiet my NeptonXL AiO with single SilentWings 3 fan is.

 

There is also a matter of radiator placement. With air cooler it's always on top of CPU. Where with AiO you can place it anywhere you want. That was also one of first things I benefited the most when going with AiO in miniATX case where PSU was basically overlapping CPU socket and I had very limited options for CPU coolers while rocking an OC'ed i7 920.

[Bramimond]

This is "what confirmation bias looks like, the thread", pretty much.

Also, wasn't it established a few weeks back that nothing beats hooking up your water cooler to your water tap? \s

 

Personally, I liked to see passive cooling thrown in to the mix. Silent fans are nice and all, but there's something cool about no fans at all.

[Techicolors]

I can kind agree with this video. I used to have a 120mm aio for my gpu not too long ago. Temps were super low and I can overclock the heck out of it. However the pump was pretty loud and the tubing was difficult to accommodate surprisingly enough. 

 

I came across an adapter that apparently able to be connected to the pump header on the aio and then to a fan header on the board. I was able to control the pump speed that way, managed to reduce the speed to a point where it doesn’t make a lot of noise. The bad part tho is the temps climbed like 10 degrees. 

 

Also tried a push pull configuration which lowered temps a further 5 degrees if the fans were cranked at a higher speed than I would like. But that just made the system even louder 

[Inelastic]

11 hours ago, Captain Chaos said:

Perhaps because the AIOs already are more expensive and replacing the fans would only increase the price gap?  

 

The Noctua is $100 with 2 Noctua fans. If you take the AIOs and replace the fans (at $25 each), you're looking at around $170 for the H100i and $230 for the H150i (rough calculation based on local prices here in Belgium, so please don't shoot me if I'm $10 off whatever the price is in your local store).  That's not even close to the same price range anymore. 

Wait, 170 dollars for the H100i?  That's insane! The last time I used an AIO, it was $60 for a double-wide 120mm setup. I think the 240mm equivalent was just over $100. Even the EK A240, a custom water cooling kit, is $160.

[Tristerin]

Me:

 

Go to Aliexpress and build your own loop, custom, for less than $100 shipped

 

AM4 Waterblock - $12 or less (even comes with chinesium TIM)

2x240mm radiator $14-15 depending on the day (comes with standard fittings for 8mm ID tubes, same as the pagodas that will come with your waterblock)

Pump/res combo $23 (or cheaper if you want to go to the $12 pump res combos), I REALLY like my $23 one.

Tubes/zip ties - $ 2-3

50/50 premix radiator fluid - $2 worth

Arctic P12 SP fans - $27 for 5 pack (Newegg, though Aliexpress has CHEAP SP fans I haven't bought those yet so cant comment on their SP'ability)

 

 

 

 

 

[Guest]

Aircool all the things!

 

I ended up with aircooling the PC's I build. Servers do aircooling, that was my reasoning. The watercooled ones where more quiet (I think, didn't measure it, just what I felt at the time). And that "breathing" effect of fans spinning up isn't that present. And nicer to look at - I miss that part. But more work! Mooooore work! Once the PC is build, I'm a lazy man - aircooled: way less work! Tearing apart a custom loop with hard pipes takes so much time! And aircooled is hell cheaper. So I can spend the saved money on more RGB fans - and don't save any 

Wasn't aware the aircoolers would perform so good with overclocked CPU's - will give that a shot!

 

And still - I'm a bit jealous at some of the watercooled PC's out there. Some look so good. Aircoolers automatically come with that transformers look of too many parts. Cool enough for me 

 

[Quadriplegic]

Eh, results seem bit suspicious to me.

 

[an actual squirrel]

Wait, was blender configured for gpu acceleration or not? Because if the noctua wins despite the heat from the 2070 that's right next to it, that's even more impressive than people are imagining.

 

If not, I sorta hope the tests could be repeated with 3dmark or something that uses the gpu as well. And graphics cards with different tdp also, so like a radeon vII all the way down to a 1660 ti.

[bluesky_dolphin]

So, I kinda feel this video was a bit not researched/extensive enough, and I would like to see some more thorough testing please.

 

Never mind the fact I just switched over from Noctua air cooling to water cooling, after 'Everybody' said watercooling is just so cool... Anyway, the upside being that does give me some recent measurements and experiences. Some points that I would like to bring in to the discussion, and I hope LTT researches this further:

 

1. In a way the comparison is unfair, because you are comparing a dual fan cooler with a single fan rad. If you want it to be fair you should have tested the air with only one fan, or the rad with 2 fans, both in push pull for example. My water cooling, a Cooler Master 120 rad runs a lot cooler and quieter with 2 Be Quiet! 120mm fans in push pull, then my Noctua cooler. I'm reaching idle temps of around 12 degrees, and fan noise not even noticable to my ears. The CPU is a FX8350 Black Edition. This in a Fractal Design Define C case. The rad is top mounted, with a total of 4 120mm fans besides the one for the rad. Currently using Argus monitor to fine tune fan speed (and thus cooling/noise levels)

 

Something else I've noticed. I've set my GPU fan to spin silent as possible (WIP) while keeping the temps down. It's a MSI 1060 6GB, and with the normal cooling out of the box it will keep the fans off until 60 degrees or so, this also heats the case air. So by finding a balance with the fans, it helps keep the temps down and thus the fans have to spin less.

 

I do feel the rad setup all in all is quieter and cooler than the air cooled solution I had before, especially when under load. Although all in all I do feel there is room for improvement in the noise level, as it is still at an annoying level sometimes. However I just built this into a new case and new set up and plan to tweak in the coming period.

 

2. I wish more rads would have been tested from different brand, and with different fans than the stock ones that came with it.

 

3. Having owned several fans from several companies now, I'm not that big a fan of Noctua fans for the following reasons:

a. Expensive

b. Ugly with capital U

c. On higher fan spin they can be quite noisy

d. I'm not convinced of their performance as opposed to some other fans (e.g. don't understand the hype)

e. The connecting cables of the fans can be on the short side, making it difficult to use in builds

 

Mounting a Noctua cooler, and all the stuff you get with it are well done though

 

Screenshots:

 

[dfsdfgfkjsefoiqzemnd]

2 hours ago, Inelastic said:

Wait, 170 dollars for the H100i?  That's insane!

 

That would be the price of the H100i + two Noctua fans.  People want to compare the performance with those fans instead of the originals, so it's only fair to add the price of the fans to that of the cooler.   So $120-ish for the cooler (it's 120€ here, and usually when it comes to electronics 1€ = 1$) + $50-ish for the fans = $170 total. 

[bluesky_dolphin]

Curious what CPU cooling paste they used, and if that has an effect? This time I used Arctic Silver 5 for the CPU, and last time used the one that came with the Noctua air cooler. I do feel/wonder if it runs cooler now because of that as well.

 

If  they used the Noctua CPU coolant it might be geared towards air coolers and not to AIO's, depending on the material used connecting to the CPU? Not sure if that would have an effect though

 

 

 

[Tristerin]

7 minutes ago, bluesky_dolphin said:

Curious what CPU cooling paste they used, and if that has an effect? This time I used Arctic Silver 5 for the CPU, and last time used the one that came with the Noctua air cooler. I do feel/wonder if it runs cooler now because of that as well.

 

If  they used the Noctua CPU coolant it might be geared towards air coolers and not to AIO's, depending on the material used connecting to the CPU? Not sure if that would have an effect though

 

 

 

Thermal paste can have an affect - as far as I am aware Noctua TIM is great.  For me, its Thermal Grizzly Kryonaut all the way - dropped all temps by a staggering 5c average when I apply it.

[BigRig]

Just watched this video.  I am a long haul trucker, and I take my gaming rig out on the road with me (I spend a lot more time on the road than I do at home).

Specs:

i7 4790K @ 4.0GHz

Z97 Mobo

16GB Ram

GTX 1070 Founder's Edition

1200W PSU

NZXT S340 case

 

You get the idea.  I use this CPU cooler: https://www.amazon.com/Zalman-Computer-Prevention-Heatsink-CNPS9500AT/dp/B000HKDA0U

 

It is a pretty big/heavy copper air cooler.  I've driven approx. 120k miles over the past 1½ years (along a lot of bumpy roads) and the cooler hasn't bent the mobo or fallen off the CPU.  Thanks for making LTT & the WAN show.

[Inelastic]

3 hours ago, Captain Chaos said:

 

That would be the price of the H100i + two Noctua fans.  People want to compare the performance with those fans instead of the originals, so it's only fair to add the price of the fans to that of the cooler.   So $120-ish for the cooler (it's 120€ here, and usually when it comes to electronics 1€ = 1$) + $50-ish for the fans = $170 total. 

Ok, that makes more sense.

 

PS: Cannonball Run is a great movie.

[zdemigod]

My one question is why the hell did noctua go with that brownish color. I would never use an AiO but i feel bad staring at my black and white pc if it would have a huge brown air cooler. 

[kameshss]

One of my friend's Cooler Master ML240L RGB leaked all over GPU and some parts in motherboard. I came to knew about it after some days. Cooler Master replaced it and they even paid me for the motherboard and the GPU as its because of their product  Bonus point is, I got to get him a better motherboard and GPU. He had basic Z370 and GTX 1060. Now he is rocking Z390 F Gaming and an RTX 2060

[Wizardling]

3 hours ago, zdemigod said:

My one question is why the hell did noctua go with that brownish color. I would never use an AiO but i feel bad staring at my black and white pc if it would have a huge brown air cooler. 

Well, IME unless you're after a directly LED-lit fan, you're probably not going to care about (or possibly even notice) the brown colour when it's under the glare of RGB lighting. And if you don't care about lighting, you're probably not going to have a clear side panel, so you won't see the brown fans anyway

 

But Noctua does sell black fans as an alternative.

[Arika]

9 hours ago, bluesky_dolphin said:

1. In a way the comparison is unfair, -snip everything else-

not unfair at all. all products were "as is" out of the box, you wouldn't buy a 2 fan air cooler and pull one fan off (unless you didn't read the specs and cant fit the other fan because components are in the way)

 

For some reason a lot of people in this thread think it's unfair because of x y z, but like i said, out of the box is what they were testing, you can get drastically different numbers if you change things around, but then that's not the point of the video.

[SpaceGhostC2C]

9 hours ago, bluesky_dolphin said:

I'm reaching idle temps of around 12 degrees, and fan noise not even noticable to my ears. The CPU is a FX8350 Black Edition. This in a Fractal Design Define C case. The rad is top mounted, with a total of 4 120mm fans besides the one for the rad. Currently using Argus monitor to fine tune fan speed (and thus cooling/noise levels)

 

 

Your idle temps are not 12°C. The builtin temp sensor on FX chips only works above a certain threshold (in the 30-40s), and idle temps typically fall below their range. Also, the only way to get 12°C would be with ambient temps at 12 or (more likely) less.

Idle temps aren't a good way to test a cooler's performance. Only draw conclusions from temperatures under high, consistent loads for different coolers.

 

With all that in mind, a thick 120 AIO like the H80 or the Nepton XL will indeed cool any FX chip, and won't even need to get loud at stock speeds. Better or worse than some air coolers? I'd have to test each of them to tell.

Whenever one watches videos like this one, it's good to remember that there are all sort of prices, performances, and price/performance values for both air and water cooling.

[bluesky_dolphin]

1 hour ago, Arika S said:

not unfair at all. all products were "as is" out of the box, you wouldn't buy a 2 fan air cooler and pull one fan off (unless you didn't read the specs and cant fit the other fan because components are in the way)

 

For some reason a lot of people in this thread think it's unfair because of x y z, but like i said, out of the box is what they were testing, you can get drastically different numbers if you change things around, but then that's not the point of the video. 

I think the issue I'm having with that is that the video should have been called "Corsair's (new) AIO vs. Noctua's top of the line air cooler" or some such. Because now with this (clickbait?) title the assumption is made that all air cooling is better than... Put a cheap Cooler Master air cooled fan against a Be Quiet! rad in there and I'm fairly sure the noise levels would be different for example.

 

I hope that is something they will improve upon, it does open the door to some interesting content.

 

Also it is mostly definitely worth to shop around with rads and fans. For example the rad I bought had a single fan that came with it (as opposed to a more expensive option with 2 fans included, same rad as far as I could tell). I already had one 120mm Be Quiet! fan suited for the purpose. I read in reviews that the fans that came with the rad were loud (and they are). So I bought the single fan version at about €35, and one extra fan at about €22. So total price for a recommended rad plus two good fans was about €55. This is cheaper or around the same price a Noctua air cooler (depending on the config). A dual  fan Noctua air cooler would cost about €77 (with 1x 120  mm, 1x 140 mm fan), $66 (with 2x 92mm fans), or €56 (with 1x 120mm fan, however the ones sold with a single fan that I can see are with a thinner cooler).

 

I hope my point is clear though, that there is a lot of room for set ups and price points, and this video is too general to make such a claim.

 

1 hour ago, SpaceGhostC2C said:

Your idle temps are not 12°C. The builtin temp sensor on FX chips only works above a certain threshold (in the 30-40s), and idle temps typically fall below their range. Also, the only way to get 12°C would be with ambient temps at 12 or (more likely) less.

Idle temps aren't a good way to test a cooler's performance. Only draw conclusions from temperatures under high, consistent loads for different coolers.

 

With all that in mind, a thick 120 AIO like the H80 or the Nepton XL will indeed cool any FX chip, and won't even need to get loud at stock speeds. Better or worse than some air coolers? I'd have to test each of them to tell.

Whenever one watches videos like this one, it's good to remember that there are all sort of prices, performances, and price/performance values for both air and water cooling. 

Interesting info, thank you, something to take into consideration.

 

I got the reading in HWmonitor for the 12°C, where it would rise and fall in temperatures, measuring along the way

 

[Guest]

On 4/30/2019 at 4:23 PM, Enderman said:

Air coolers just don't look special at all,

i beg to differ