[SOLVED] Artic liquid freezer II 120mm for i9 9900K

HeartOfAdel

Commendable
Apr 7, 2021
86
14
1,545
Hi to everyone
My ATX case doesn't have enough space to mount a 240mm AIO on top of it, so I'm only left with 120mm ones. As I've seen from reviews Artic liquid freezer II is particularly good for a 120mm aio. So would it cool i9 9900K at stock turbo speeds nicely without exceeding 75-79 degrees at the highest load (for example in AC Odyssey)? I also considered undervolting it a bit. Don't recommend air coolers, they're visually outdated to me.
 
Solution
Front intake usually is the best for it. Does the front of the chassis not support 240/280mm models?
When used as top/rear exhaust, that cooler has to cope with both the cpu and the heat exhausted from the gpu, which most users get models that dump their waste heat inside the chassis.
Because of this, it'd be better to over-provision on size when using it as exhaust.


View: https://www.youtube.com/watch?v=NGHiRrQ2AAo&t=281s

-it was a Blender load
-room ambient was 22C
-one of the 3 coolers was a Corsair H100i Pro RGB
-the H100i Pro RGB still saw over 80C

at stock turbo speeds
This can vary between motherboards. Not all of them follow the Intel-defined specs.

Phaaze88

Titan
Ambassador
Front intake usually is the best for it. Does the front of the chassis not support 240/280mm models?
When used as top/rear exhaust, that cooler has to cope with both the cpu and the heat exhausted from the gpu, which most users get models that dump their waste heat inside the chassis.
Because of this, it'd be better to over-provision on size when using it as exhaust.


View: https://www.youtube.com/watch?v=NGHiRrQ2AAo&t=281s

-it was a Blender load
-room ambient was 22C
-one of the 3 coolers was a Corsair H100i Pro RGB
-the H100i Pro RGB still saw over 80C

at stock turbo speeds
This can vary between motherboards. Not all of them follow the Intel-defined specs.
 
  • Like
Reactions: HeartOfAdel
Solution

HeartOfAdel

Commendable
Apr 7, 2021
86
14
1,545
What is the make/model of your case?
What graphics card will you use?
Do you have any other options?
You will not get the performance you paid for with the 9900K
It's Deepcool Tesseract
Not sure about the GPU, something like 3070/80
I7 8700K is the other option with the same AIO
I clearly would get its full performance. I highly doubt it'd throttle at all (it's around 100 degrees as i know). If you're about OCing then i absolutely don't understand why you have to buy top cpus to just shorten their lifespan with overclocking. You buy a product for the performance it has naturally, not for how much you can oversqueeze out of it.
 

Phaaze88

Titan
Ambassador
Well the chassis explains why 120mm units only, and why one can't be installed at the front - maybe it could, but you'd have to take out a bunch of things first.

Anyways, you'd need to tune both 9900K and the gpu so thermals are easier to manage.
8700K would be even more manageable.
 

HeartOfAdel

Commendable
Apr 7, 2021
86
14
1,545
Well the chassis explains why 120mm units only, and why one can't be installed at the front - maybe it could, but you'd have to take out a bunch of things first.

Anyways, you'd need to tune both 9900K and the gpu so thermals are easier to manage.
8700K would be even more manageable.
I'm currently running a Deepcool Gammaxx l120 v2 on my rear panel as intake with another cpu and that's the way i'd use any AIO, working flawlessly so far. Gpu is also cooled by a side panel fan blowing directly on the card.
I've seen a video of a guy using Artic Liquid Freezer for i9 10900K which is even more power hungry. The temps while stress testing were somewhere in low 80s.
View: https://www.youtube.com/watch?v=uxn2htZKc9s

This guy also oced 10900K with another aio (but i doubt aritic one is worse) at 1.24V and got high 70s-low 80s.
View: https://www.youtube.com/watch?v=M0h6vfRJrbU

After this i wonder if it's really gonna be good with 9900K especially if it's at something like 4.7Ghz 1.2V(maybe even lower if possible)
 
I'm running my i9-9900k on a little H80i v2. I've got the voltage locked at about 1.23v with my OC and LLC barely increases it (think I'm using level 4 LLC).
Performance is great (see benches in sig), however, if I burn the CPU with Prime95 for over 15 mins it will get too hot. Fortunately, that's just in synthetic benchmarks - it rarely gets above 75C in continuous normal operation. Mine's mounted as an exhaust right behind the CPU.

Biggest mistake people make is letting the motherboard dictate the voltage with auto. Most motherboard's use a ham-fisted approach to 'auto' voltage by giving the CPU waaaaay more than it needs.
 
  • Like
Reactions: HeartOfAdel

Phaaze88

Titan
Ambassador
Wait... you already have a 120mm AIO on it?
i9 10900K
This, and the other i9s aren't that bad on power efficiency. It's when users start overclocking, raising/removing the power limits on them that things get iffy.
Power efficiency goes out the window. Gpu still is the biggest power user when active though.
Auto voltage sucks, always has. At the minimum, you should use a negative Vcore offset.

Setup in the first vid, I doubt they could push it too hard; that looks like a B460 series motherboard, which would VRM power throttle if pushed hard enough.
Heck, it probably WAS power throttling during the cpu portion of TimeSpy - I couldn't see clearly, but it did look like it couldn't hold 4.8ghz or higher(depending on how many cores are active at a time).
It should've been able to do that if power and thermals were acceptable.
Assuming I'm right about the VRM power throttling in that video, the cpu was still pushing 90C+... that's not great.

My problem with the second video is that it doesn't represent your own build in any kind of way.
The gpu is custom cooled, and so sending it's waste heat directly to the cpu cooler can be avoided - in a sense.
The other thing is, Aida64: Cpu, Fpu, Cache, all checked is actually a MODERATE load on the cpu, the likes of Intel's own XTU stress test.
The actual heavy cpu load from Aida is FPU ONLY. When you've checked more than that option, you've done it wrong. Cpu only is also not a heavy load.
 
  • Like
Reactions: HeartOfAdel

HeartOfAdel

Commendable
Apr 7, 2021
86
14
1,545
Wait... you already have a 120mm AIO on it?

This, and the other i9s aren't that bad on power efficiency. It's when users start overclocking, raising/removing the power limits on them that things get iffy.
Power efficiency goes out the window. Gpu still is the biggest power user when active though.
Auto voltage sucks, always has. At the minimum, you should use a negative Vcore offset.

Setup in the first vid, I doubt they could push it too hard; that looks like a B460 series motherboard, which would VRM power throttle if pushed hard enough.
Heck, it probably WAS power throttling during the cpu portion of TimeSpy - I couldn't see clearly, but it did look like it couldn't hold 4.8ghz or higher(depending on how many cores are active at a time).
It should've been able to do that if power and thermals were acceptable.
Assuming I'm right about the VRM power throttling in that video, the cpu was still pushing 90C+... that's not great.

My problem with the second video is that it doesn't represent your own build in any kind of way.
The gpu is custom cooled, and so sending it's waste heat directly to the cpu cooler can be avoided - in a sense.
The other thing is, Aida64: Cpu, Fpu, Cache, all checked is actually a MODERATE load on the cpu, the likes of Intel's own XTU stress test.
The actual heavy cpu load from Aida is FPU ONLY. When you've checked more than that option, you've done it wrong. Cpu only is also not a heavy load.
I don't have the cpu yet. I'm rocking a 4th gen i5 with aio.
I appreciate you sharing all that information but I don't understand what you're on about exactly and what your overall point is concerning my question.
 

Phaaze88

Titan
Ambassador
I appreciate you sharing all that information but I don't understand what you're on about exactly and what your overall point is concerning my question.
Mmkay... lemme try again.
You linked a couple videos using a bigger cpu, which I figured was an attempt to show how a 120mm AIO handles them, but I noticed inconsistencies with them.
-The first video, it looked like the cpu VRM power throttled. I can't be too sure - the camera wasn't very steady during the points that mattered. If there's actually was throttling occurring, then the whole thermal testing scenario may as well been invalid.
You don't want the cpu to power throttle, that's lost performance.

-The second video, the cpu there didn't power throttle, but the video author wasn't pushing the cpu as hard as they believed they were. Fpu only is the only option that should've been run when for Aida64.
Checking cpu, fpu, and cache at the same time is a moderate load on the cpu. Here's where I got that info from: https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461-4.html
Often there were cases where user's cooling would 'pass Aida', but in specific applications/games, they found cooling to be worse. That was related to them checking all those options in Aida.
 
  • Like
Reactions: HeartOfAdel

HeartOfAdel

Commendable
Apr 7, 2021
86
14
1,545
Mmkay... lemme try again.
You linked a couple videos using a bigger cpu, which I figured was an attempt to show how a 120mm AIO handles them, but I noticed inconsistencies with them.
-The first video, it looked like the cpu VRM power throttled. I can't be too sure - the camera wasn't very steady during the points that mattered. If there's actually was throttling occurring, then the whole thermal testing scenario may as well been invalid.
You don't want the cpu to power throttle, that's lost performance.

-The second video, the cpu there didn't power throttle, but the video author wasn't pushing the cpu as hard as they believed they were. Fpu only is the only option that should've been run when for Aida64.
Checking cpu, fpu, and cache at the same time is a moderate load on the cpu. Here's where I got that info from: https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461-4.html
Often there were cases where user's cooling would 'pass Aida', but in specific applications/games, they found cooling to be worse. That was related to them checking all those options in Aida.
Alright, i get everything now! Indeed, power throttling isn't a good thing but it's 10900K after all, it's noticeably more power hungry than 9900K, About the second video, yeah, maybe there's stress tests that can load it even more. As i understand even highest end AIOs might struggle with keeping the temperatures below 80 with those tests.
But after all, considering what alceryes wrote above about using 9900K with H80i v2, it seems like Artic Liquid Freezer ii 120mm will do a fine job especially if you undervolt it to healthy balanced voltages.
 

Phaaze88

Titan
Ambassador
But after all, considering what @alceryes wrote above about using 9900K with H80i v2, it seems like Artic Liquid Freezer ii 120mm will do a fine job especially if you undervolt it to healthy balanced voltages.
Yes, it can work. Intel already optimized performance and power efficiency - a little undervolt helps things along.
None of the i9s are that bad, if they're following the Intel defined specs. There's other videos out there of users running small coolers on them.
Things get problematic when:
-motherboards applying enhanced turbo profiles
-overclocking, be it manual or software
-raising/removing power limits
-disabling C-states and Intel Speed Shift/Step
These break that balance of performance and efficiency, and that's when things get dumb.
As long as you don't get into all that, you should be able to manage fine. It MIGHT get uncomfortable(whatever your comfort zone may be) sometimes, but it'll be brief.
 
  • Like
Reactions: HeartOfAdel