problems with update 5.5.134 of iCUE

[Ramog]

I have updated to version 5.5.134 of iCUE, ever since the Fans of my H115i ELITE CAPELLIX randomly switch to 100%.
Sometimes it stops by itself, sometimes I have to switch the fan curve and after switching back it acts like normal again.
I want to stress, that my coolant temp and my CPU never gets hotter during this, happens on idle.
I attach an image that shows the 100% spikes in fanspeed.

Is this a known bug?

Is there a way to roll back to iCUE version 5.4.98?

[c-attack]

No way to roll back to prior 5.xx versions because the files are sever side. It always will download the current version. That leaves riskier methods like Windows backup points or a prior system image — which might be a bit extreme. 
 

There have been several ‘weird fan behavior’ posts in the last couple weeks, but I don’t think any cause has been identified. Are you using coolant temp as the control variable for the fans?

[Ramog]

30 minutes ago, c-attack said:

There have been several ‘weird fan behavior’ posts in the last couple weeks, but I don’t think any cause has been identified. Are you using coolant temp as the control variable for the fans?

No I have  switched to a a bit more onorthodox method . I am controlling it by package temp (I know its abit of a weird decision but worked for me, decoupeling water temp from fan speed has caused my system to be much more silent while still remaining cool enough) and have a alarm set to set the fans to 100% if I ever climb over 45°C. Running on 1000 RPM max most of the time and on idle at 600rpm. Upon writing this I also realize that

This worked out for anything but the most hottest days.

[Ramog]

47 minutes ago, c-attack said:

There have been several ‘weird fan behavior’ posts in the last couple weeks, but I don’t think any cause has been identified. Are you using coolant temp as the control variable for the fans?

My Fancurve looks like that. And I just realized that 100% was a bit of an overestimation its running at the higher slope of 65% (PWM I am guessing?) the weird thing is, that it was not at this temperature at all for minutes yet the fan speed looked like this:

I think it was triggered by a short spike of temp on the package and only got fixed as it spiked again, so my guess is that it somehow didn't notice that the temperature dropped again shortly after the high temperature and only noticed that after it spiked again.

It even being so high was due to my special summer Fancurve I set a week ago, but the weird behavior I described a paragraph is still caused by the update if thats understandable.

[c-attack]

Generally, there are two reasons the fan behavior goes amok.  One is some type of software error in the programming where the configuration curve resets/freezes etc.  The other is when the program is having trouble getting the data -- whether it's fighting with another program or a bug of some type.  The fans go crazy when there is an unexpected value.

 

Not a lot you can do about the first, but the second has a viable test method.  CPU temp is always volatile by nature and and your case and radiator fans don't do anything for the the underside of the CPU where the pins apply the voltage.  Coolant temp is a very slow to change variable and there should be no spikes.  You can use a very low slope curve that should never spike while staying within your fan speed parameters.  At the very least, it's a check against a data fetching error or something else.  

[Ramog]

7 hours ago, c-attack said:

Generally, there are two reasons the fan behavior goes amok.  One is some type of software error in the programming where the configuration curve resets/freezes etc.  The other is when the program is having trouble getting the data -- whether it's fighting with another program or a bug of some type.  The fans go crazy when there is an unexpected value.

 

Not a lot you can do about the first, but the second has a viable test method.  CPU temp is always volatile by nature and and your case and radiator fans don't do anything for the the underside of the CPU where the pins apply the voltage.  Coolant temp is a very slow to change variable and there should be no spikes.  You can use a very low slope curve that should never spike while staying within your fan speed parameters.  At the very least, it's a check against a data fetching error or something else.  

I mean yes, Coolant temp is a very slow changing variable but to put it blunt: I don't really want a workarround I just want to write what I know about the problem so it can be fixed.
I mean yes in my case I have coolant temp but there should be many people that just have a commander pro without water cooling.

[c-attack]

No, most people with a H115i Elite Capellix will be using the coolant temp as the control variable because that is how it was designed.  Very few will be using a Commander Core without the AIO that comes attached to it.  So to be blunt -- you are using it incorrectly and it may be impacting the results you see. It would be logical to try to using it the right way to see if the deviations are connected to what you are doing. CPU temp is not the right control variable for any type of water cooling or case fan control.  If it continues on coolant temp, then there may be a problem with the installation or cooling config file that will not go away on its own. 

[Ramog]

3 hours ago, c-attack said:

No, most people with a H115i Elite Capellix will be using the coolant temp as the control variable because that is how it was designed.  Very few will be using a Commander Core without the AIO that comes attached to it.  So to be blunt -- you are using it incorrectly and it may be impacting the results you see. It would be logical to try to using it the right way to see if the deviations are connected to what you are doing. CPU temp is not the right control variable for any type of water cooling or case fan control.  If it continues on coolant temp, then there may be a problem with the installation or cooling config file that will not go away on its own. 

I thought the commander core is a standalone product as well?
Also I would say that a fancurve not working right doesn't have to do anything with usage error. Like sure the H115i Elite Capellix is a watercooler but the fancontroller part of it is just that. You can tell me all you want that its not the right control variable for a watercooler but technically I am just controlling fans here, that should work regardless of what variable I use (and has worked fine for 4 months before this update).

Like package temp should just be another temperature which I control fans with if I plug in two more fans into the commander pro that are not on the radiator and want to control them by package temperature I should be able to do that right? The commander pro (even if its part of the H115i Elite Capellix) doesn't know that those fans are attached to a radiator. I am not trying to argue that my setup isn't weird but it has been working for quite some time as I said before. And the fans work independently of the watercooler anyways, so making this an issue about "you are just using the watercooler wrong" confuses me.

[c-attack]

The point was to try and decipher a data fetch error (no user solution) from some other type problem with the CUE configuration files that might be user correctable.
 

If you set the sensor for the fan curve to coolant temp and the out of control behavior stops, then that suggests there is a problem with fetching the cpu data, Corsair will have to correct it, and in the meantime you have a quiet and efficient fan control source. 

 

If you change the curve to use coolant temp and the fans are still spiking up and down, then that suggests there is something wrong on config file end and you probably need to do a clean install of CUE, which may correct the original issue as well. 
 

The stand-alone version of the Commander Core is the Commander (Core) XT.  It’s similar to the AIO Commander Core in several ways, but drops the AIO power circuitry to add a usb passthrough and two wire temp sensors for measuring internal air temp.  That is the native control variable for that controller.  The liquid temp is the native control source on the Commander Core.  CPU temp is a terrible control variable for any type of fan control, but rather than try and write out an exhaustive and convincing explanation as to why, the suggestion to use coolant temp was a troubleshooting step so you would have an idea of what to do next — wait it out or clean install.

 

 

[RoninDevil]

Am I right in thinking that what is being said is that on the custom fan section I should be setting all the fans to use ‘coolant temp’ instead of CPU package?

I have the issue at the moment, which you have been giving me advice on, where a bug is not letting me use icue properly. I can’t re order the fans etc, but I also don’t have anything showing coolant temp or the ability to select it as an option.

is my thinking in this correct?

[Ramog]

On 8/27/2023 at 6:58 PM, c-attack said:

The point was to try and decipher a data fetch error (no user solution) from some other type problem with the CUE configuration files that might be user correctable.
 

If you set the sensor for the fan curve to coolant temp and the out of control behavior stops, then that suggests there is a problem with fetching the cpu data, Corsair will have to correct it, and in the meantime you have a quiet and efficient fan control source. 

 

If you change the curve to use coolant temp and the fans are still spiking up and down, then that suggests there is something wrong on config file end and you probably need to do a clean install of CUE, which may correct the original issue as well. 
 

The stand-alone version of the Commander Core is the Commander (Core) XT.  It’s similar to the AIO Commander Core in several ways, but drops the AIO power circuitry to add a usb passthrough and two wire temp sensors for measuring internal air temp.  That is the native control variable for that controller.  The liquid temp is the native control source on the Commander Core.  CPU temp is a terrible control variable for any type of fan control, but rather than try and write out an exhaustive and convincing explanation as to why, the suggestion to use coolant temp was a troubleshooting step so you would have an idea of what to do next — wait it out or clean install.

 

 

Ah okay then there was a big missunderstanding, I managed to make a nice curve that is based on coolant temp and I didn't have any issues for a few days. So either there is a specific problem with fetching the cpu package temp or it doesn't like big tempreture jumps in general (they don't happen in coolant temp since there is quite some thermal mass behind it). But oh well it works for now, I just wanted you guys to know off the bug.

[Ramog]

2 hours ago, RoninDevil said:

Am I right in thinking that what is being said is that on the custom fan section I should be setting all the fans to use ‘coolant temp’ instead of CPU package?

I have the issue at the moment, which you have been giving me advice on, where a bug is not letting me use icue properly. I can’t re order the fans etc, but I also don’t have anything showing coolant temp or the ability to select it as an option.

is my thinking in this correct?

Do you have a corsair watercooler? If not the cooalant temp won't show up at all.

[RoninDevil]

I have a Corsair icue link water cooler. There appears to be a bug preventing some functionality at the moment.

[Ramog]

3 hours ago, RoninDevil said:

I have a Corsair icue link water cooler. There appears to be a bug preventing some functionality at the moment.

Idk coolant temp shows for me, might be a bug. But its unreleated to my problem.

[RoninDevil]

The option to select link h150i rgb: pump, is this the coolant temp option?

[c-attack]

13 hours ago, RoninDevil said:

setting all the fans to use ‘coolant temp’ instead of CPU package?

Correct.  That's how the process really works.  The metal block conducts the heat off the CPU.  That's how it is cooled.  Whether you have a tiny air cooler or a massive external water system, the instant CPU temp for a specific voltage is the same -- for the first few seconds.  It's the cooling method's ability to get rid of that heat over time that differentiates various cooling methods and sizes.  A little 120mm radiator may need to work pretty hard to keep the CPU temp level at 100%.  A 360mm can run moderate speeds and stay cool.  A huge external array can putz along at the minimum and dump all the heat that is added.  

 

Also in this instance, I do believe there is a problem with CUE obtaining CPU data but was trying not to overly influence anyone's observations by stating it as fact.  It's a reasonable guess based on the multiple user reports that surfaced this release.  I have suggested using coolant temp to those affected to see if it mitigates the control issue while waiting for a fix.  It's also possible to have both problems -- the data fetch issue and a corrupt cooling config so you kind of have to explore both options.  

13 hours ago, RoninDevil said:

I can’t re order the fans etc

That is a known issue for the QX-CUE Link Hub on this release.  A fix should be in the next version.

 

5 hours ago, RoninDevil said:

The option to select link h150i rgb: pump, is this the coolant temp option?

Coolant temp also goes by the cooler name, such as "H150i Elite RGB Temp" or generically H1xxi Temp.  Different models use slightly different names, in different places.  For CUE 5 it should say "coolant temp" in the temp array but likely H1xx Temp in the drop down menu for selecting it.  

Edited August 29, 2023 by c-attack

[Neodame]

I just built a Corsair BYO system. In the UEFI Bios load/access there no problems with my Corsair keyboard. I Loaded fresh new Windows 10 pro load, again, no problems with my keyboard. I load first time load of ICue5, SW Ver 5.5.134, then numerous problems recognizing the Corsair K55 Keyboard. It flickers in/out on device display, then I get Orange Warning indicator next to keyboard display in Icue GUI. I get windows audible ping every 10-15 seconds, corresponding to recognizing/clearing keyboard picture in device GUI. This continues without interruption, unless I shut down ICUE in active taskbar. Can I roll back to older driver? Is that an option or are you aware of these new errors for future update?