karajan

Yes, custom curves cause this issue. It's likely not a harware issue. but I want custom curve for more quite setup :D:

This happens to me as well. only after I wake my computer from sleep mode after a long time. I have to restart the service TWICE to get it back to working.

A little story and a lesson i learned in hard way. This happened to me long time ago. But i thought i just share it :) While i was stress testing to see if my OC was stable, i decided to change the pump speed to see how it affects CPU temps. Just when i applied new speed, system froze... unstable OC, BSOD. It didn't take too long until i noticed my CPU was running at 800Mhz, super slow. Could not boot into windows. even in bios it was struggling to run. Temp was at 100C! Basically pump speed was set to 0... what a mess! So i shutdown my PC, plugged out all power cables, waited like 10 minutes for pump to cool down. (It was really warm!) Then i tried to get into windows, I only had a few minutes before the CPU reached 100C again and shutdown itself. So again after 10 minutes i reduced Vcore voltage as low as it made sense! Finally i could get into windows and set pump speed back to Quiet mode. And there i revived my cooler :3 Not to be paranoid or anything, just make sure your system is stable before changing the pump speed. Because it can be a PITA if you get unlucky... That cooler is still running good though ;) ---- My suggestion to Corsair is to include hard reset button on their coolers because that would be really handy for cases like this.

Fans don't have to run aggressively on water cooled system, it wont damage your CPU if fans are not running according to the curve. In my experience fans did not stop in any scenario, for me they actually run faster when iCUE freezes. The CPU also protects it self from being damaged if it reaches very high temps.

Ok, that was an unfair argument. But if this is really the issue then developers should setup an arrangement and come up with one solution. Nonetheless iCUE temp monitoring stops working entirely. At least that can be solved and will be a good temporary fix. If reading from a sensor fails basically wait 1-2 seconds and try again. Catching exceptions should be easy. At least in C# they are!

"Some sensors can only have one device accessing them at a time." We are in 2019. They are going to make 7nm chips, planing to go to moon and mars... this is just ridiculos! It can be easily solved even if only one device can access sensors at a time. A single service that basically monitors all sensors and other programs should accept this as an standard to only request sensor values from this service. Its a very simple design problem...

@Greybeard Thank you very much, hope it gets fixed soon. After logging for long time (a week or two), here are the errors that caught up. not sure if it helps but i post it in case if it helps. I did run other monitoring software's during this process sometimes. but i believe iCUE should be immune to crashes (that's when you use try-catch statement!) Service_Error.log Exported system info and logs.zip Service_Trace.zip