poppedcorn Posted October 13, 2019 Share Posted October 13, 2019 I have a Ryzen 3900x with the H100i Pro AIO. My idle temperature is sitting at: CPU: 47C ~ H100i: 38C ~ Does this look right? I have the fan profile set to 'quiet' and the fan spins up almost immediately to 2000+ RPM which is super loud. I am concerned whether this is normal idle temp for the combo or H100i in general. Link to comment Share on other sites More sharing options...
poppedcorn Posted October 13, 2019 Author Share Posted October 13, 2019 Also, am I suppose to hear pump? I have pump turned to extreme with 2910 RPM and I don't hear anything. Link to comment Share on other sites More sharing options...
c-attack Posted October 13, 2019 Share Posted October 13, 2019 Hardware and coolant temperatures are always affected by the environmental temperature. 38C is too warm for idle coolant, unless the inside of the case is also 35-38C. What's the room temperature? Compare this any other internal temp data (motherboard sensor, drives, etc). You should be able to hear the pump at 3000 rpm on extreme, although not with the fans blasting like that. It likely is running. If it is not, you would not be writing from that PC. The coolant temp would tick upwards at 1C per second. You shutdown 60 seconds later. If you put two fingers on the pump cover, you should be able to feel the vibrations. The fans are spinning up high because 38C coolant temp is a large increase over the presumed baseline 20-23C. All fan curves have to have a baseline for quiet operation and the defaults assume that above range. In a temperate or tropical climate, that may not be your baseline room temp for half the year and thus a custom curves with a quiet idle are a better choice. However, at the moment make sure you have actually applied the curve to fans and pump. When properly selected, both the fan curve and fan will be yellow if it is applied. Click the preset (quiet/balanced/extreme). Then click the fan/pump you want to apply it to. It's the opposite way from what most people are expecting. So assuming the pump is at a speed other than the quiet 1100 rpm (that might cause this condition), the other possibility is some kind of flow restriction. Cycle the pump down to Quiet 1100, then right back up to Extreme. Repeat every 10-15 seconds for a few rounds. You are trying to create a little pressure to push bubbles or anything else out of the block. Did the H100i Pro temp fluctuate at all? +-1C would be normal in any given interval. However, a 3C drop followed by a 3C rise (or something similar) would signal trouble. Normally you can hear bubbles in the form of static-like clicks and pops. You haven't mentioned that, but again with 2000 rpm fans that is going to be hard to hear. If the pump cycling does not work, you can try tilting the case. The goal is to get the in/out hoses above the block. They are probably on the RAM side of the CPU, so that means tilting the case back and lifting the front end. It can be awkward, so take care. **The fans are going to be annoying and don't appear to be helping. Go to the Performance tab and click the + to make a new curve. Select fixed RPM for now and set them at 1000-1300 or something you can stand. Link to comment Share on other sites More sharing options...
poppedcorn Posted October 13, 2019 Author Share Posted October 13, 2019 First of all, thank you so much for giving me insight. I spent hours trying to figure out why my new build sounds like a jet engine. This will help me tremendously trying to narrow down the issue. So here are the temperature readings from this morning. The room temperature is 20C. Computer Cold Boot: CPU: 36C ~ 38C Coolant: 25C 5Min In: CPU: 38C ~ 40C Coolant: 28C 10Min In: CPU: 42C ~ 44C Coolant: 33C 15Min In: CPU: 42C ~ 45C Coolant: 33C Motherboard: 46C, 61C, 21C, 42C, 51C Drives: 44C, 33C, GPU: 55C The temp seems to be hovering around the last set of values afterwards. I will try to cycle the pumps next. Link to comment Share on other sites More sharing options...
poppedcorn Posted October 13, 2019 Author Share Posted October 13, 2019 (edited) Pump cycling a few times and the coolant temp went from 33C to 31C. After that I have set the fans to be fixed 1,100 RPM and the coolant temp seems to hover around 31C and slowly climbed to 33C. I have this fitted into a small form factor pc. I had to contort the pipes slightly to get it fit. Is it possible that this might cause obstruction thus giving these readings? Edited October 13, 2019 by poppedcorn Link to comment Share on other sites More sharing options...
poppedcorn Posted October 13, 2019 Author Share Posted October 13, 2019 Did 4 cinebench r20 at load testing and got: CPU: 80~82C Coolant: 40C After a minute rest: CPU: 48C trending down Coolant: 39C trending down Link to comment Share on other sites More sharing options...
c-attack Posted October 13, 2019 Share Posted October 13, 2019 So here are the temperature readings from this morning. The room temperature is 20C. Computer Cold Boot: CPU: 36C ~ 38C Coolant: 25C 5Min In: CPU: 38C ~ 40C Coolant: 28C 10Min In: CPU: 42C ~ 44C Coolant: 33C 15Min In: CPU: 42C ~ 45C Coolant: 33C Motherboard: 46C, 61C, 21C, 42C, 51C Drives: 44C, 33C, GPU: 55C OK, not bubbles or other flow blockage. You have a steady state increase in coolant temperature from the moment the system is powered on, which then levels off. This means there is some type of heat management issue with the general case layout/design or the H100i Pro cannot expel its heat for whatever reason. Can you describe the case layout? Where is the H100i Pro located. Is it being used for intake or exhaust? Is there anything that might restrict the airflow across the radiator (dust filter, very thick case paneling, etc)? Typically in a functional system, if your coolant temp goes +6-8C on load or stress test, you can then expect it to drop 3-4C in the first 1-2 minutes when you stop, then the remaining over several minutes. When this does not happen, that usually means the internal case ambient temperature is the high and equal to the coolant temperature. Unfortunately, iCUE and Link before it were not good programs for interpreting motherboard specific sensors. There is no way for us to tell what those values are and which ones are junk. Some of the data is not going to be accurate. It does seem clear your coolant temp is riding +10-12C above ambient. Most users are going to be in the 4-7C range and since I woke from sleep 15 minutes ago, mine is only +2C over ambient. Your GPU is also warm for idle and about +10-12C over what I would expect for a zero fan state -- the same as the other offset. This appears to be were the deficit originates. *If you use another program to assess system temperatures like AIDA or HWiNFO, make sure you quit iCUE and kill the iCUE Service in Task Manager before launching the Portable version of HWiNFO. The two programs will cross swords when accessing the same data and this leads to more garbage values on both ends, which is the opposite of what we're going for. Link to comment Share on other sites More sharing options...
poppedcorn Posted October 13, 2019 Author Share Posted October 13, 2019 Hey thanks for responding. H100i Pro is mounted on the top of the case exhausting air out. The fans are below the radiator blowing towards it and out of the case. I have the https://louqe.com/ S1 case and it does have thin panel which I can try without. The MB and GPU sits below the radiator/fan on the bottom. Seems natural the air from gpu or cpu can rise and affect the radiator temp. I doubted myself about applying thermals correctly so I had it redone and making sure they are tight. The cables are tied up better as well. Here are the temp I have after 5/10 minutes after cold boot: CPU (3900x): 43C ~ 46C Coolant: 30C boot, climbed and hovers around 35C ~ 36C Seems to be about the same. :/ Good to know there are conflicting applications. Measurements are only taking with X1 precision, Ryzen Master, and iCUE. I used HWiNFO on the last take. I'll make sure of that. Is it possible that the case and lack of airflow had made these temps creep up over the average idle temp? Also did some research that some motherboards gives high default voltage 1.4+ vs the 3900x stock 1.2. I haven't messed with any of the motherboard settings though. Link to comment Share on other sites More sharing options...
poppedcorn Posted October 13, 2019 Author Share Posted October 13, 2019 (edited) With a single cinebench r20: CPU: 83C, back to 45C Coolant: 38C, went back down to 36C with balanced settings after a few min. Ran some games for 20min and got: CPU: 60C Coolant: 40C ~ 45C Is 40/45C ok for radiators/coolants? Edited October 14, 2019 by poppedcorn Link to comment Share on other sites More sharing options...
c-attack Posted October 14, 2019 Share Posted October 14, 2019 (edited) Ugh.. this is now the third time I am writing this. The forum keeps glitching on me. The answer gets shorter and shorter each time. The case is not what I expected and it's basically a perforated box. I am not sure how much better it could be for airflow design. However, I am wondering is the radiator in the "top" the only exhaust point, with both sides as intake? Are the two ends solid? Or is there other vented sides with no fans? I don't think there is a cooler issue and this does appear to be a set-up/heat management kind of thing. Regardless, it would be nice to take about 6C off the coolant and pass that along to the CPU as well. Overclocked or not, the 3900X cannot heat up the coolant at idle like this. It doesn't have the wattage. The heat has to come from another source (as ambient air temp, other components) or something has to prevent the radiator from shedding heat (solid top cover, dust filter, etc). The CPU and coolant ratios are as expected, just shifted higher from the start. I don't how much GPU activity is in the R20 test and ultimately that has to be dealt with regardless. However, for testing purposes try a light CPU only test like the Bench from CPU-Z. That is linear in load and light. It should jump about 30C on start, then hold with the only increases coming from increases to the coolant temp. It also takes the GPU out of the equation. **Also, are the side passive intake (no fans)? I am having a hard time grasping the real life dimensions with gear. Edited October 14, 2019 by c-attack Link to comment Share on other sites More sharing options...
poppedcorn Posted October 14, 2019 Author Share Posted October 14, 2019 (edited) Ugh.. this is now the third time I am writing this. The forum keeps glitching on me. The answer gets shorter and shorter each time. The case is not what I expected and it's basically a perforated box. I am not sure how much better it could be for airflow design. However, I am wondering is the radiator in the "top" the only exhaust point, with both sides as intake? Are the two ends solid? Or is there other vented sides with no fans? I don't think there is a cooler issue and this does appear to be a set-up/heat management kind of thing. Regardless, it would be nice to take about 6C off the coolant and pass that along to the CPU as well. Overclocked or not, the 3900X cannot heat up the coolant at idle like this. It doesn't have the wattage. The heat has to come from another source (as ambient air temp, other components) or something has to prevent the radiator from shedding heat (solid top cover, dust filter, etc). The CPU and coolant ratios are as expected, just shifted higher from the start. I don't how much GPU activity is in the R20 test and ultimately that has to be dealt with regardless. However, for testing purposes try a light CPU only test like the Bench from CPU-Z. That is linear in load and light. It should jump about 30C on start, then hold with the only increases coming from increases to the coolant temp. It also takes the GPU out of the equation. **Also, are the side passive intake (no fans)? I am having a hard time grasping the real life dimensions with gear. Top is the exhaust point, the left and right are basically perforated like you said. Gpu on one side and CPU on the other. The bottom of the case is elevated so I assume that's where the intake of fresh air is with some possibility of side. The sides have no fans attached. Based on https://www.techpowerup.com/forums/threads/ryzen-9-3900x-voltage-and-temperature.257295/ I have set the voltage to be 1.25v. I am getting 46C/47C idle on CPU now and coolant is around 37C/38C.. seems just slightly better. I have also taken out the top cover so radiator exhaust is completely unobstructed. In the R20 test, I only tested the multi-core of the CPU, no GPU at all. I wonder though if the GPU 55C heat rises into the fan/radiator contributes to upping the radiant idle temp. Setting the profile of the H100i Pro to quiet, the fan noise is doable at idle, for gaming, it is bearable. The question then is 40C/45C bad for radiator an hour or two a day? What I could also test my system without any panels (left, right, top) and see if the temp improves. But in reality, without any external factors, the radiator should be sitting slightly over ambient right? Edited October 14, 2019 by poppedcorn Link to comment Share on other sites More sharing options...
DevBiker Posted October 14, 2019 Share Posted October 14, 2019 The GPU heat absolutely does rise through the radiator, especially if that's your only exhaust point. Try flipping the fans on the radiator so that it's intake. That will unquestionably tell the tale. Link to comment Share on other sites More sharing options...
c-attack Posted October 14, 2019 Share Posted October 14, 2019 No, the 40-45C on the coolant is not harmful to the physical pieces. It just poses as a penalty to cpu temp. The radiator fans should pull air in from the sides as well. I have a passive, no direct intake fan build in a O11. The coolant doesn’t go like that. I am still pondering alternatives, but for now a +6-8C cpu temp penalty is the only consequence. Definitely take over control of the fans. The three presets won’t be of any use and I recommend everyone make their own regardless. Keep the pump speed at balanced or extreme. Quiet will compound the heat penalty. Link to comment Share on other sites More sharing options...
poppedcorn Posted October 14, 2019 Author Share Posted October 14, 2019 Quite a revelation with these discoveries. I change the GPU profile from 'performance' to 'adaptive' and I am seeing decreased temp: CPU: 42C~ Coolant: 34C~ GPU: 42C~ The GPU originally was 55C on idle now dropped to 42C wow. With all these testing, is this normal behavior based on the limitation of the airflow of the case? Link to comment Share on other sites More sharing options...
DevBiker Posted October 14, 2019 Share Posted October 14, 2019 Different fan profiles for different GPUs will affect the temp, simply by spinning faster at lower temps. That's likely what you did. Those GPU temps are about normal for air cooling. And yes, it may be due to a limitation of airflow. I'm not familiar enough with that case to say for certain but we've seen those issues around here occasionally. Link to comment Share on other sites More sharing options...
c-attack Posted October 14, 2019 Share Posted October 14, 2019 It’s not just games that use the GPU, so setting the power profile to max performance may have been keeping the clocks and voltage up, and the GPU diode temp along with it. However, the temperature of the chip is not what heats the case and you would still only see a minimal power output at the desktop — nowhere near the 250-300W output when gaming and that “waste heat” is what warms the case. Still, no reason not let the GPU clock down and use specific program profiles if you have something that needs it. Link to comment Share on other sites More sharing options...
Recommended Posts