LINK H150i RGB, atrocious cooling performance

[Imagine_]

I got a Link H150i RGB cooler around Christmas time and have had issues with it. It struggles to cool my Ryzen 9 7900x even when power is limited to 65W. At idle my coolant temp hovers around the high 40s low 50s mark. Today my cooler went into overheating protection from a game that only used around 40% CPU utilization. My CPU temperature is idling around the high 65-70C mark. Before anyone asks, yes I have reseated and repasted my cpu. First using the inbox paste, then a fresh application of Arctic MX-4, and then finally a fresh application of Arctic MX-6 which I currently have. I have the AIO in a pull configuration, with the pump lower than the top of the rad. I can't believe my experience is this poor with the cooler after my H150i elite Cappelix lasted me 4 years with zero issues. Screenshots and Specs are listed below. All screenshots from just running Chrome and iCue. Any help would be great.

Specs:

Cpu: Ryzen 9 7900x (set to 65w PBO), GPU: Msi 4070 Ti Suprim X, Motherboard: Asus X670-E Prime, Ram: 32GB(2*16) Corsair Vengence RGB DDR5-6000MT/s, 9 QX120 Fans, Link H150i RGB AIO cpu cooler. The case is a Lian Li O11 Dynamic

Edited February 21 by Imagine_

[c-attack]

AMD CPUs tend to idle on the high side as they don't really idle much at all.  You'll need to take a look at the Vcore when on the desktop, but it likely is not sitting down and it usually is doing something.  

 

The clear abnormal sign from above is the 51C coolant temp.  That is typically out of bounds for all but the worse environmental conditions in the middle of Summer.  A typical idle coolant temp is something around 4-7C above your room temperature or at a minimum equal to the internal case temp.  That's as low as it can be.  It's not unusual to see moderate load CPU games return some of the highest coolant temp changes because the GPU is raising the internal case temp all along.  So rather than CPU waste heat, the large increase in coolant temp is environmental. 

 

Even so, 51C is high and you'll need to find some other internal temp point for comparison.  The typical "motherboard temp" sensor is usually in the middle of the board and should be close.  Other things like RAM temp and m.2 drive temp will ride higher than actual environment temp, but do help put things into context.  If your RAM is 70C+ and so is the m.2 drive, you know things are very warm.   Your QX fan sensors are on the rear side of the fan.  From the above, it looks like you have some intake air temps in the low to mid 30s and exit air temps in the 40s.  That is warm and most certainly part of the issue.  If you blow 45C air into a radiator, the water inside will be 45C with zero additional added heat from the CPU.  

 

The other way to check actual AIO performance is to eliminate the GPU heat for a safety check.  A malfunctioning AIO will fail a basic stress test.  Use something mild like the test from the Bench Tab in CPU-Z.  It's linear and moderate so it makes for a very easy to see process.  CPU temp should go to Temp X as soon as you start and hold steady.  It will creep up about +1C every 60 seconds or so during the first 5-6 minutes of the test and then level out.  That is the coolant temp increasing at +1 intervals and the CPU temp will go up in accordance.  An AIO that shoots up 15C in a matter of minutes is not working as it should.  

 

There are some other questions to ask about radiator placement, air flow direction, and if there are barriers to the radiators functionality like dust filters, tight fitting glass panels, etc.  However, those typically won't cause you to flunk the CPU-Z test but are a problem for normal mixed GPU/CPU load use.  

Edited February 22 by c-attack

[Imagine_]

I ran the CPU-Z test for ten minutes, and my coolant temps held steady at 55C, actually, they slowly decreased. My CPU also didn't exceed 70ish C on all cores. I also ran a Cinebench render and the same scenario, 70ish C on all cores and coolant temp hovering around 54 and slowly dropping. Could this be some issue with thermal probes? My AIO is in a pull config with three intakes on the rad, three intakes on the floor of the case, and three as exhaust on the top of the case. Any suggestions?

CPU-Z figures

Cinebench Figures

[c-attack]

From the QX data above, can you tell me where each QX # fans is located?  Normally I’d guess the mid 40s QX temps are the exhaust fans in the case, but I think it’s going to be the other way around. If your three 45C+ QX fans are the front radiator fans, then that is the exhaust air temp off the radiator. That would confirm the coolant temp is probably ~50C and that is a problem. It should not be anywhere close to that. One way to try and undercut QX fan location is with a speed test. Create a custom fan curve and set it fixed PWM 100%.  Then assign it to one QX fan at a time and see which one fires up. You also likely can do this with RGB and the same process of creating some particular lighting effect and applying to one fan at a time to identify the QX number. Where this gets more complicated is the CUE Link system lets you re-arrange the order of components and I don’t know if that will complicate things. I had some strange experiences trying to my QX to be identified as 1-4 in proper physical order when I first got them. 
 

The next time you cold boot the PC after it has been off for a few hours, quickly open cue and check the H150i Temp. When you first power on, it should essentially be the room temp and it will warm up 4-5C over the first 10 minutes of sitting on the desktop. I am trying to find a way to differentiate between an internal physical problem and an environmental one. An AIO that was powered down all night but then is at 40-50C in 60 seconds has an internal issue. That means replacement. If it starts off at 27C but continuously creeps up to 50C, then it’s likely environmental and there are things we do to lower that penalty. 
 

Coolant is minimum possible cpu temp with zero volts at the cpu. As long as the coolant is 50C, you’ll never see cpu temps lower than 57-60C even when the PC is truly at idle. Then when it’s loaded, you go up a normal +30C and that’s when you start hitting temp safety limits. I’m not sure it really ran the cpu-z test at all. Modern cpus are fairly smart and simply stop working. Something is off with the score and the peak core temps. 

Edited February 22 by c-attack

[HN1]

I have a computer with 4000D case and H100i. Another one with 5000D and H150i with extra fans and the first one gets lower temperatures (about 5º lower). Maybe it's performance is not as expected

My first H150i broke after 12h btw

Edited February 22 by HN1