H100i Platinum Not Performing to Expectations

Weeks of searching the internet for fixes seems to have proved fruitless.

 

My CPU at idle ranges from 40-55, with the base at 40 but then spiking to 55 or sometimes 60 at random points. I've attributed this spiking to the way BIOS is OC'ing my Ryzen processor so I believe I can fix that just with some tweaking, but still a minimum of 40c when the ambient temp is 27 is a bit high for an AIO innit?

 

CPU clock never goes above 4.3Ghz and voltage is set to 1.35. During gaming it sits at about 47-54c depending on the game, and spikes to 60-65 randomly, which again I'd attribute to my BIOS playing jump rope with my core. Cinebench gets about 65-67 at 5 passes, and I haven't had the guts to run prime95 for more than 3 minutes but the temps there are about the same. During these high workloads the CPU will sometimes spike to 74C, which is my main concern.

 

AIO Liquid temps stay at around 32-34C, and both the tubes and the radiator seem lukewarm at best. Kinda odd since I've been hearing people say how the tubes should be hot to the touch...

 

I've reseated the cooler, wiped and applied some Thermal Grizzly Kryonaut, made sure the CPU_FAN header for the pump is at max, and tried tapping and tilting the AIO, albeit not by much (last thing I want is for this thing to leak).

 

So any suggestions as to what's going on? Is the pump busted (it's regging 2000+ rpm in iCUE), is my NZXT H500 hotboxing it? Is the radiator clogged? Help!!!

 

(also thank you)

I think you are looking at pretty common Ryzen behavior.

 

Idle temp - Little air cooler, AIO, or massive external cooling array... it won't change much. Your idle CPU temp is going to be voltage and CPU behavior/power settings. There is barely any heat to dissipate so the cooler doesn't play much part, except as the point of conductivity for the CPU. Right now my 8700K is running through 3x360mm rads all by itself with no GPU load. Coolant temp is 27.7C. The package temp is at 41C and moving around. Core temps pinging through the 30s. This is pretty normal for any current CPU. The only way to get idle temps down is to turn on every single power saving feature available, from BIOS to OS and tranquilize your CPU. That has clear side effects and not something I would recommend for most users, especially when idle CPU temp has no bearing on its longevity or health.

 

Coolant temp is about watts in vs watts out. We tend to think of the actual temperature of an object as the defining measure of heat, but it is not in this case. A pin head at 450F is not the same as a Pizza Oven at the same temp. With Ryzen's small die they are very power efficient in terms of watts. You don't have that much heat to dissipate, but nevertheless the limitation is on the CPU side. You will hit a voltage/temperature limit long before you can surpass the capacity of the cooler. This is true for Intel CPUs as well, but there is quite a bit of difference between a 65W TDP 3600X and a 10900K pulling almost 300W from the socket. Your coolant rise appears to meet the watts from your CPU and I don't see any concerns there. Coolant temp isn't going to matter much for you. You can let the fans ride easy. Remember, +1C coolant temp rise = +1C CPU temp rise. No need to blow the house down to reduce coolant by 2C.

 

I can't give you exact voltage and overclock point for the 3700X and I don't follow those CPU as closely. That information is surely out there, but understand no matter what you strap onto the CPU, it will still be voltage and its resulting CPU temp that stops the show. The heat must pass through the CPU from the pins to the cooler. It can't go around. I would tweak cooler settings to be pleasant and then work on the BIOS settings. Intel or AMD, tuning the voltage will yield more benefit than any cooling option on the table.

First of all thank you for the response, that was lightning fast. I'm probably gonna just ride this out until I get a case with a bit better airflow (this is my first build and I wasn't aware of the poor thermals on the H500.) Also apparently this case is optimized for negative pressure so maybe down the line I'll buy something that doesn't require positive pressure inducing intake fans (Noctua NH-D15?) While I appreciate the advice high my Package run at 50c just on one firefox tab isn't exactly super optimal imo

 

EDIT : I tried lowering the clocks to the default 3.6Ghz instead of 4.2 as well as lowered the voltage to 1.3, and the temps were at 69 max in Cinebench. Wtf?

 

EDIT : Tried resetting bios to default + xmp profile for my RAM. Cinebench scores are def better, but temps are largely the same. On boot temps fluctuate wildly, even up to 70, but then seem to stabilize 40-50 like usual. Still this seems really iffy. My room is HOT rn.

Edited May 29, 2020 by AndRaxius

Even if you underclock, the issue is the CPU/MB is applying heavier voltage at times you are not expecting, like at idle. One of the trends of the last several years is to tweak both CPU and MB behavior out of the box to be very snappy. This was an effort/response to trying to win all those motherboard shootouts by 2 benchmark points or whatever. At the default settings, you may not have an actual idle state. The voltage may be loading moderately all the time, for every folder you open, program running, etc. The Intel equivalent is C-states, some underlying P-states through Speeshift, and the older power management features. It is quite possible the board is set up to not use any of these features that most users have been accustomed to in the past. Even with them on, the days of the idle clock frequency sitting flat like a pancake at 1 GHz are gone.

 

I don't think you should or need to underclock to get temps in line. However, there may be some basic AMD settings in the BIOS that address the behavior. Unfortunately, I haven't had an AMD in 15 years and I don't know what those are. However, for what it's worth I have the same base coolant temp, the same idle temp, and our game temps are similar as well with my 8700K. I don't think you are far off the mark and I am not sure what CPU you had before, but there might be a period of adjustment to the new behavior.

All AMD Ryzen 3000 series CPU's bounce frequencies up and down and thus temperature. Remember if you are using software to measure you CPU temp your system is not really idle and the software is looking at the temps and everything else it is monitoring. My 3900X 'idles' between 34-44C, according to the LED on my motherboard. I set my fan curves to stay flat until the CPU gets over 55C. My water temp stays pretty flat on the AIO except under load where it may rise a few degrees, if that, while gaming. The AIO keeps temps under 70C at full load for benching, unless I run P95 where of course it goes higher as water temp increase, but stays under 85C.

 

My point is the temps bounce around at 'idle' and light load. An air cooler is not going to change that and neither would the best custom loop. The CPU just does it, every single one of them. If you want the temp to sit steady then clock it at a set speed and voltage and turn off any monitoring or any other software, basically what the BIOS does, and the temps will sit idle.

Room temp is around 29c and my cpu package is sawtoothing between 50-60 on idle, during gaming is now sawtoothing 55-65. Furnace room moment.

I seem to have """""""fixed""""""" it for the time being. Locked CPU clock to 36.50 and voltage to 1.25. Temps are definitely lower than before and SO much less spiking. Still about 42 on idle (i can't get the core to underclock/undervolt at idle for whatever reason which i'm not sure how to fix), but hey it's better than what I was getting before. Maybe I'll invest in a NH-D15 and maybe a better cooled case (i've heard the Lancool II is solid)