iCUE bugging (CPU AT 100ºC during boot)

[jckodel]

I have a H100i RGB PRO XT with a Threadripper 3970X and I live in a tropical zone (so it's not uncommon the room temperature is > 30ºC).

 

When iCUE is configurated for limited software control or in a default performance setting (quiet, balanced or extreme), everything works just fine, except the fans make too much noise (quiet sets the fans to 2100 RPM, too loud).

 

So, I created a custom profile so the CPU keeps under 70ºC on idle and fans are as lowest as possible (reaching 100% on 85ºC).

 

The problem is: when I restart the computer with this custom profile, the CPU overheat during startup (it seems the pump itself stops, because with the pump on extreme and fans at 0%, the temperature is no higher than 75ºC). From the moment I turn the machine until the moment iCUE boots up, the radiator get extremely hot and CPU temperatures are too high (when my hardware monitor starts, the temperature is near 90ºC - this temperature is not observed even with fans at 0% (but with the pump running, of course)).

 

It seems iCUE has a bug that stops the pump during POST =\

 

Anyone has this issue? Is there any way to reset all fans and pump to 100% when iCUE is shutdown (so, when the system restarts or boots up, all fans and pumps will be 100% until iCUE comes online again).

Edited February 14, 2021 by jckodel

[c-attack]

There is no bug that stops the pump and you cannot manually stop it except by physical cutting power to the unit. What you are describing seems more like a contact issue between cpu and block. Your idle temps are too high, even for an elevated room temp.

 

One way to distinguish the two is a loss of coolant flow (stopped or too slow) will cause the cpu temp to increase in measured, but steady steps. 40....45....50....55....60...70....80...etc over the course of 30 seconds to several minutes depending on the severity. A contact issue will go hot the instant voltage is applied. 50...50...90...50...90...50 as voltage is loaded and unloaded.

 

The XT and most other coolers have an auto save to device feature that puts your custom curve on the internal controller. Stopping pump speed is not an option, but if you’ve used a control variable other than coolant temp (H100i XT temp) it may cause a strange reaction in the fans since the controller can’t read data from other devices without the software.

[jckodel]

The problem is: while iCUE is on, I never can get my CPU over 79ºC (even with SIV CPU burn).

 

If I set the fans to 0, no problem (CPU will stay 55 to 70ºC idle, cooling liquid no more then 40, 42ºC).

 

The ONLY problem is on restart/boot: it gets really hot (and I can't get this kind of temperature when iCUE is online, no matter how much I try).

[c-attack]

Power to the unit is supplied by the SATA cable and thus is constant as long as the PSU delivers it. Any type of defect along that line would be present at other times as well. It’s theoretically possible there is a defect in the motor that is making it difficult to start, but this would not change when the OS or CUE loads. 12v is 12v and that doesn’t change.

 

You can separate the software from the boot state issue by going to the task bar and quitting the cue app and dropping the unit into hardware mode. If there is a problem, it will show there as well.

[Corsair Notepad]

Can you provide screenshots or pictures of the 100C that you are seeing on boot up?

[jckodel]

Here an example (please, see attached images):

 

Just turned the machine on, temperature was very high.

 

After several seconds, Argus Monitors loads then show the temperature decreasing.

 

From the moment the machine booted until the Argus Monitor started (after I type my Windows password), the temperature was > 85.7ºC

 

It is important to say that this temperature is NEVER reached on normal circustances (even when gaming, the temperature doesn't get much higher than 72, 73ºC). Only with CPU stress tools I can get temperatures at most at 90ºC, so is very strange the boot reaching almost the same as a CPU stress test (considering I have 64 CPUs, it is pretty hard to make it all 100%)

 

This does not happen if I use a pre-defined cooling set (quiet, balanced or extreme), but, then again, the noise is too loud.

 

The second screenshot is my custom schema.

 

Based on the cooling liquid temperature:

< 35ºC, Fans 0%

39ºC = 20% (the minimum amount for them to turn on)

41ºC = 60%

42ºC = 100%

 

Now, after some minutes using this schema, CPU temperature is 61ºC, cooling liquid is floating from 40.2 to 40.6ºC.

 

Room temperature is about 26ºC with humidity "yes - all humidity available in the planet" (80%)

 

(BTW, my case is a Redragon Wideload (similar to this one):

 

3 fans at bottom are intake (but only on if the PCI Express region reaches certain temperature), 3 fans at top are exausting (depending on the temperature near the top of the motherboard, near the CPU socket). There is no fans at side (the Corsair H100i is mounted there, taking air from the outside (the Wideload has a second compartment behind the MoBo where the HDs and PSU are, having an filtered opening in the place where the Corsair fans are).

 

All internal fans are off, except if the temperature raises too much (like in gaming).

 

One suggestion: my MoBo have a cooling cycle during startup. First, it turns all the top fans and VGA fan to 100% for some seconds, then it turns CPU_OPT (the fan over the Corsair WC) and all bottom fans at 100% for some seconds, then it resumes to my schema (this is done automatically by the MoBo).

 

It would be nice for the Corsair WC to have this kind of behavior on startup (and, why not, on shutdown). So, no damage can be done in cases like this. A simple hardware power on timer would do and, for the shutdown, when the iCue quits, it sets all fans to 100% (so there is some seconds between programs quitting and Windows turning off the PSU where the fans are at 100%).

Edited February 20, 2021 by jckodel

[c-attack]

If your CPU temp is hitting high marks immediately at boot, fans and case layout don't come into play at all. Even radiator fan speeds are irrelevant for this. As long as liquid is moving, the small amount of heat at that stage will be pushed out of the CPU block chamber/cooling fins. The only two things that should be able to cause this are either: 1) bad CPU to cooling block contact (bad temps all the time and can't run load); or 2) the coolant flow is stopping/slowing for whatever reason.

 

As improbably as it may be, it does appear like it could be the second issue. While this used to be more common when the AIOs were getting their power from the MB fan headers directly, most coolers including the XT get their power from SATA. It is always on with the PSU and always makes the pump run. I don't know of anyway to prevent that, aside from some type of power relay that cuts off SATA power on boot. You want the opposite of that since it also powers HDDs and the like, but people sometimes have other reasons.

 

I more concerned about the coolant and CPU temps after 5 minutes. If the pump doesn't start, it's not really heating the entire cooling loop -- just the the small amount of water in the chamber. So the moment the pump starts again, that is pushed out the CPU would return to normal. You would not still have a coolant temp of 40C (unless the case interior temp is 40C) 5 minutes or even 5 seconds later. I can see a AMD TR running high over coolant temp on idle, so 60C over a 40C coolant temp doesn't necessarily indicate a contact problem. If the coolant temp was 26-30C like expected, then the CPU temp would be 15C less. I think there is something wrong with the cooler's flow response.

 

Contact Corsair Support or use any exchange options you have with the vendor.