Build Complete 1080Ti AIO liquid cooling project

[Phaaze88]

Gigabyte 1080Ti + NZXT Kraken G12 + Fractal Design Celsius S36

It's been a week since I finished this(June 24), but I must say that it turned out quite nice - for the most part.

The tear-down:

https://imgur.com/3okdykr

View: https://imgur.com/3okdykr

https://imgur.com/VHN41SP

View: https://imgur.com/VHN41SP

^The old thermal pads weren't so fortunate...

That old paste - really DRY and CRUSTY - took some time to get off, but the end result was well worth it:

https://imgur.com/pYnFXLr

View: https://imgur.com/pYnFXLr


What kept this little project from being smooth all the way through was the bloody instruction manuals:
A)The pump block just wouldn't fit in the G12's shroud - I legit started panicking here, thinking I would have to try and return the S36 and be without a gpu for X number of days.
I looked through Fractal's instruction manual for anything - some kinda hint that I may have overlooked - got nothing. An hour or so later, I find an old LTT forum thread confirming with visual proof what I already knew: that the cooler and kit were compatible.
But what was I missing?
A couple google searches later, I find a video taking out 4 certain screws... I was so 😡 after that. That was the worst of it out of the way.

B)Lack of instruction - or perhaps, suggesting the wrong screws to secure the radiator to the fan bracket. The manual showed that I use 6-32 screws with washers - wouldn't secure with the washers, as the screws were a tad too short.
I ended up not using the washers at all - I should've taken a pic of how the rad looked with those screws in there, but it hadn't crossed my mind. It was late, and time to hit the hay.


Wake up refreshed, and soon discovered that I would be able to do a push-pull setup with this. I wasn't originally intending to, but it happened to work out that way after I made another discovery, one which further cemented the idea of doing away with the NF-A14 & NF-F12 IPPC 3000s entirely:
The A14 versions are actually slightly LARGER than their regular brown and Chromax versions. We're talking really subtle here - instead of the specified 140x140mm, it's more like 141x141mm. Switching back and forth between them on the front panel install:
-the A14 IPPCs, as a set of 3, are a little harder to squeeze in the front fan bracket's vertical dimensions compared to the Chromax I already had on hand.
-the above would mean the screw holes on the IPPCs are offset from normal, I guess?
-the 3 Chromax fans fit, but the IPPCs didn't anymore - they did before I had a radiator mounted; goes back to the section about the rad screws I used.

Put it all together, and pressed the power button... powered up and booted without a hitch - NICE!
Went into bios, set up some new fan/pump curves, logged into Windows, started up Afterburner, then played one of my favorite indie racing games for like an hour, and then did a custom run of Superposition:
With my applied OC of +75 core and +450 memory, the darn thing could not hit 40C... granted, that's a reported average, but that's still way better than what I was expecting with the old cooler: right around 60C with fans cranked to 100%.
Also, instead of bouncing between 1990 - 1920mhz, it just sits at 2012...
Fortunately for me, GDDR5X isn't as toasty as GDDR6; I didn't use heatsinks on the VRMs nor memory, so that means chassis airflow is pretty good - I also forgot to mention that I replaced the G12's stock fan with an NF-A9.


It's definitely quieter - not that I was one bothered with noise to begin with. I DO notice the Fractal Blackout fans the most; they're the loudest device in there now. I will be replacing those later with Noctuas 😛

https://imgur.com/ftlSJj9

View: https://imgur.com/ftlSJj9

https://imgur.com/ZfnldNl

View: https://imgur.com/ZfnldNl

https://imgur.com/v4iqYkG

View: https://imgur.com/v4iqYkG

I may not have as much fan/pump control as opposed to AIOs that force you to use their software, but I feel I'm not really missing out there.

I have a better opinion of AIOs now - although, it's only been a week...
Gpu liquid cooling is pretty great. Too bad society has conventionalized cpu cooling over it, when it doesn't benefit from it nearly as much.
-Gpu cooling is direct die out of the gate. That IHS does reduce effectiveness of cooling solutions.
-Gpu dumps less, or none of, it's waste heat in the path of the cpu cooler, thus reducing heat soak.
-For the noise-sensitive folks, the gpu can be run even quieter and cooler than air cooled solutions.

Then again, it is more convenient for cpus, so maybe that's part of the reason.
To me, the only times cpu liquid cooling makes sense:
-overclocking on i7(specific), i9, or X-series cpus
-building M-ITX, or other SFF systems
Aside from those, there's an air cooler out there that fits most situations. Aesthetics are subjective, so I won't list that as a pro nor a con.

Them: "Yeah, I've got a liquid cooled cpu paired with an air cooled gpu that pulls more, or almost twice as much power."
Me: 🤔 Seems legit.


Thanks to everyone who read all the way though this - even those who simply skimmed through.
The purpose was to share my experience putting this together, and I have achieved that.

 

[GarrettL]

I was surprised at how much thermal paste was on the gpu from the factory. I know they are trying to be diligent and let the pressure push out the excess, but wow.

Cool project indeed and nice work Phaaze.

 

[Darkbreeze]

Nice job man. A lot of people would have either botched this or simply given up when the instructions couldn't be spoon fed to them by the manual. Or, come here wanting somebody else to solve it for them. Just goes to show that there are answers and solutions if you put the time into finding them and are willing to put in the work.

 

[Phaaze88]

Thanks!

@GarrettL
I was surprised by that as well. The card still ran darn cool even with the original cooler on.
That paste though, I went through some alcohol on that... also used a small toothpick to help chisel and crack some of the excess paste around the die and tiny transistors(?).
I dared not use that on the die - a lot of rubbing got that off.

@Darkbreeze
I had been planning this for some time and had been reading up on how to do it - keeping some tabs saved for reference. They didn't cover everything, but I'm not lazy with this stuff.


By the way, I forgot to post the following image. This was in the Celsius S36's manual:

https://imgur.com/GiZZ2YQ

View: https://imgur.com/GiZZ2YQ


2 of them seem rather peculiar to me...

 

[Darkbreeze]

There are a lot of indications, if you look, that many of the companies making cooling hardware or cases don't actually have a really good idea how it all actually works. Some of their diagrams are just so wrong, and impossible, that it makes me laugh. Now, I'm no aerodynamics engineer, but I DO happen to know a fellow who WAS one that used to work for Lawrence Livermore labs in California as one, and I occasionally run things by him to get input, or to share a laugh.

 

[Phaaze88]

Let's not forget the marketing knuckleheads that don't have a clue how some of these products actually work, but going so far as to make up crap like 'bio-mimetics'(Zalman)...
Gamers' Nexus burned those Thermalfake reps so hard on that Floe RC cooling, LOL!

 

[Darkbreeze]

Yes, Fooler Master and Thermalfake are definitely not the worst offenders out there, but they ARE the worst offenders of the big name companies. Zalman certainly has it's share of "troubles" as well but I agree, I've seen all kinds of companies do similar dumb crap because they let somebody from marketing have a free hand without having anybody from a part of the company where they might actually know what they are doing have a look at it before it was released or just decided heck with it, what do those guys with engineering degrees know anyhow.

And then you get the Chinglish companies that screw things up simply because you have a bunch of non-English Chinese marketing people together with one or two halfway English speaking/writing contributors and they decide that those are the folks that are going to make the manuals for the products that are going to be sold in English speaking markets. Jebus.

 

[Phaaze88]

🤣
Too true!

 

[rubix_1011]

The Zalman CNPS20x cooler that I reviewed this past week (or published anyway) was interesting - a lot of marketing BS. It's a decent cooler and performs well, but the utter crap that gets put on a website and the side of a box is almost revolting.

"4D steroscopic fins' .....no, they're just shaped slightly different. It doesn't mean they're a special form of anything.

And the inverse/reverse/Converse/Reebok heatpipe orientation. Sorry guys, you just have a baseplate and heatpipes.

I was actually asked not to read and look at any other reviews when I was doing my coverage of the cooler...they didn't want me to be 'influenced' by negative coverage.

I looked at most of them. I have many of the same conclusions they did. It doesn't change anything, the issues are obvious when you look at coolers quite a bit. It's expensive ($100) and cools as well as the be quiet! Dark Rock Pro 4 and Deep Cool Assassin III which cost $10-$15 less.

 

[Darkbreeze]

Which is funny, because MONTHS ago I called out Kit Guru for their review and the flaws in the methodology they used, in which they claimed it outperformed all other air coolers including the D15. Pure BS. Sure, it's a decent cooler.

But is it going to be getting the same performance in two years after heating and cooling has caused the fins to shrink away from the heatpipes because they are not soldered like Noctua's are? Now, I don't know if theirs are or not, I'm just saying, not all of them are and not all of these coolers do as well when retested later down the road after they've been in use for a while which is part of the reason WHY Noctua does that. I think Frostytech did something on that many years ago and determined that there WERE differences in performance on some heatsinks after a period of time, especially those not soldered.

But it was obvious that Kit Guru's review was skewed because it flew in the face of data from some of their OWN previous reviews. Totally different numbers used to represent the performance of other coolers in the Zalman review than what the numbers showed in the reviews for those other, original cooler reviews.

 

[Phaaze88]

Not much to be done about it, really. There's some people that love to gobble that marketing fluff up, and some that'll be drawn to it solely because of it's unique looks.
There's some who won't touch Zalman again because of the crap they've done in the past, and some like myself, who will look for reviews and find that it's really nothing special, and so on...