Is my AIO working as it should?

KungFuSpaghetti

KungFuSpaghetti

KungFuSpaghetti

Associate
Joined
7 Apr 2017
Posts
1,762
I recently changed out my case to the new Lian Li o11 dynamic and CPU cooler from my trusty noctua d14 and im seeing some odd stuff.

I replaced the noctua with the coolermaster ML360R aio and idle temps are marginally lower, but during some cinebench test runs the CPU got over 84 degrees!?!? This never even hit 80 degrees on my noctua, highest was 78.

What I have noticed however is that I get a higher score on cb20 with the 360 aio and the cores seem to be boosting higher. Still, can't help but feel these temps are way too high... should I be concerned?
 
It's a 3900x, totally stock, no XFR, no PBO and fresh noctua thermal paste.

Aio fans are mounted case side pushing air or through the rad, I have another 3 intake fans pulling in from the front side (lian li dynamic o11) at the moment and have another x3 fans ordered to pull air in from the bottom.

Yeah fans are 100% mounted in these correct direction too. Beginning to think the pump is broken...

Oh, fan header the pump is connected to is also set to 100% in w/ pump mode and monitoring CPU.
 
Idle seems similar to noctua, as do gaming temps. Maybe I just expected more, but seems weird that a 360 rad sees higher temps. I'll have to do a few more runs and pay attention to all core boost when fully loaded. I think the noctua was 4'075 all core.
 
Doyll, I'm not trying to determine cooling ability in the sense you're insinuating, I just want to know if there's an issue, not a diatribe into why you think aio's are garbage. I've had aio before on my GPU and my experience was a positive one.

I've gone off several reviews before picking this one, as (to me) it looks good and on paper would comfortably out perform my noctua. It does out perform my noctua in cinebench scores, but at 7 to 8 degrees more.
 
Last edited:
I get where you are coming from, at the same time I looked at the benchmark temp and noise roundup(s) of the usual YouTube tech sites and temps on the test rigs (usually Intel) the temps for this cooler were generally lower than other units tested. The reviews half the time were for other aio's but the Ml360r usually makes it into comparison charts and does well.

I'm more than willing to strip it out, send it back and try a different unit... but what. I want a 360 rad, rgb and zero or low maintenance, so custom loop is out.
 
I think there's something up with my cooler either way tbh, as temps shouldn't be as high as they are after a couple of test runs on cinebench. I think I'll return it.

Is there any kit you would recommend? I see that EKWB do one
 
I've got a Lian Li o11 Dynamic Der Bauer one (assuming it's identical to the normal dynamic!) and I still have the D14, it's actually sat on the desk in front of me. It's sadly too big to fit in this case though, on paper it's 5mm too large, but in reality it's only 3mm... but still, it wont fit sadly or I'd be using it still, as it's served me well on x58 4.4ghz Xeon x5650, used it on my 2700x fine and again on this 3900x in my previous case no issue.

It's really odd, as the ML360R is allowing higher boost clocks, gets better results in bench tests for CPU results, but it runs even hotter though, which makes no sense. I've got the ML360R mounted as exhaust and it's on the roof of the case. Airflow is generally good I think, as I have x3 intake fans and my GPU runs a lot cooler in this case, 58 degrees observed playing Civ 6 after hours and hours, where previous case it sat in the mid 60's.

I saw the custom EKWB kits, but they also do for £150ish a closed loop kit like the one I have, but again no flow rates etc... although I cant actually find any info on how flow rate actually makes a difference beyond a certain point, does faster moving liquid have a law of diminishing returns beyond a point?

Really, I should get one of these gravity based coolers, just a shame they are hideous, but appear to out perform even custom loops without so much as breaking a sweat!
 
I'll have to try turning the fans around later, that would result in equal pressure at the moment as I only have x6 120mm fans.

I'm using hwinfo64 to monitor temps, so, today I'm logged into my VDI at work, Civ 6 late(ish) game running, I have my coding software open, discord, steam (plus other bloaty game apps) and the usual keyboard, headset crap etc etc.
Temps are:
Minimum: 30.3 degrees
Maximum: 64.1 degrees
Average: 41.0 degrees

This is fairly similar to my Noctua temps I'd see in my old case... Only real difference is that on Civ 6 I have at any one time, x4 cores on my first CCX holding a minimum in game boost of 4.5ghz and periodically must be going over 4.6ghz. They only ever hit about 4.2ghz before at most playing Civ... I'm so bloody confused with the behaviour of this chip/cooler now... I get that higher thermal headroom allows bigger boost clocks, but that doesnt explain the fact that when pushed the AIO registers higher max temps. Seeing this 3900x hitting 85 degrees gives me pause for alarm!
 
varinder said:
Having the AIO on the side with fans extracting air through the radiator and out the back with an additional 3 fans at the bottom bring air in should make a difference. I would also use at least one fan placed in the top bringing air in over the VRM's...all the air will be extracted via the AIO fans...
Click to expand...

I'm not sure if it will reach that far to mount at the side, as that was my initial thoughts regarding mounting, having the AIO pull air in from the front/side. I've got x3 fans for the bottom on order but probably no point having a fan at the top in this case as there's no exhaust at the back on the Lian Li O11 and my VRM temps have never seen anything above 50 degrees. I'm running a pretty typical setup for this case I think...?

Ross Thomson said:
If you have it completely stock and its hitting 84c either you've not put it on right or there is an issue with the unit. Even your noctua temp seems really high given the total stock settings.

Wouldn't bother with the EK stuff - have a look at the arctic cooler freezer II line of coolers.
Click to expand...

Yup, never overclocked as never saw the point. I've timed my ram and IF and that's it. I thought the temps seemed alarmingly high for the type of kit, so I'll try re-pasting and mounting it again and if it's still hitting those temps I'll send it back to rain forrest for a replacement or get something else. I was going to get the freezer II as you've suggested, but it was out of stock everywhere and OCUK sent my new case earlier than expected lol. The freezer with some rgb fans would look pretty nice with the Bykski fans I've got ordered if I go that route.

FYI, my Noctua was also in a different case, a much smaller one. Fractal Design Define C is pretty cramped...think it's slightly smaller than 400x400x22 lol
 
I've seen that in the coolermaster software it thinks my CPU is idling at 9 degrees and hits 54 under load on cb20, yet hwinfo64 was showing it hitting 79 degrees at the time. Wtf?

@varinder the case I have is the o11 dynamic limited edition one, I guess that's the regular version of the case (it's the chrome mirror glass one). I'd certainly have no issues with the arctic in terms of space, the current rad could be twice as thick and id have space left lol.
 
hmm, okay more food for thought then! I have MSI afterburner open all the time for my GPU, but I only recently (yesterday) put the coolermaster software on, but the temps on that app are in line with what I expect it to be. As the guy above also said, my Noctua temps seem way too high as well. So I'm at a loss, I'm wondering if there's no issue and my mobo is giving out false readings, as the rad I can touch now and it's tepid at best. I can hear and feel the fluid, so the pump is 100% working, how well I've no idea though...

Regarding voltages, I've left everything auto, so should I be running an offset or fixed voltage?
 
Not sure on average clock, but according to hwinfo64 my max effective core clock speed is just 4384mhz on core 0 and core 1. I've had the PC on all day today and had Civ 6 on in background, then had a game of pube g at lunch with my boss, so it's had some mixed abuse. Minimum temp is 29.5, max of 78 (did a couple of CB20 runs before) and an average temp of 40 degrees exactly.

Where do I see my average core clock in hwinfo? (sorry for the dumb questions!)
 
okay nice one @wookiee87 I'll give that a try shortly and see what the difference is. The one thing I will say is that with this AIO my in game clocks are higher, my CB20 results are at worst over 7200 with pretty much all my crap running and over 7400 when I do a clear run. It must be at least doing something right, just that headline figure needs to come down, but every other scenario the temps seem normal :S
 
Okay so tried fixed voltage and it still showed it hitting 1.5v max on hwinfo64, but repeated cinebench runs were 76 degrees max. Tried a negative offset next, temps 79 degrees but performance dropped to 7067...
Gaming temps stay largely unchanged no matter what and stay in the 50s.

I've not tried all core overclock, but don't think my chip is strong enough to hit 4.4 and guessing gaming perf would drop?

I have my ram running cl14, but at 3600. I have 3800 cl16 saved but performance was slightly worse for some reason.
 
Okay I'll try for a lower voltage on fixed, but offset wise I just hit the minus key and went with the first value as a test... but performance dropped a fair bit to almost Intel 10 core like levels... :rolleyes::p
 
okay, so further testing has yielded better results :)

1.28v was too low and CPU scores dropped, so changed to 1.29175 and it pulls a healthy 7300+ on multi and stays at 67 degrees. That seems a little more like it?
 
I'm really pleased with that, so thank you big time, it never even occurred to check fixed voltage or what it was pulling wattage wise.

My mobo doesn't do 1.28v exactly, so it was 1.2750 I tried, then just straight to 1.29175. There may be some more room to drop slightly, but the temps and performance I can live with, the CPU idle is 30 and under now, and that's with the fans down from 70% to 55% :D

Coolermaster software is a little more optimistic, 4 degree idle... lol
 
Im certainly happier with these temps that's for sure! Dropped almost 20w and actually gained performance! Crazy!

7345 in cinebench 20 is repeatable every time even after reboots, it's incredibly consistent. I'd imagine it would do a little more with discord, steam, epic, al my rgb software etc switched off. Going up to 1.3v or above, the results drop back to 7199.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom