Randomly disappearing Temp/Fan readings

romanovskis

Member
I am using HWiNFO64 5.12-2740 to manage fan on HP 2530p. First of all, big thanks for this functionality, it is a lifesaver. I want to report something, that happens seemingly randomly, but almost everyday. Sometimes after reboot or after sleep resume, ExtTemp shows 255.9C and fan rpms 0, sometimes both disappear. It could be somehow related to RightMark CPU frequency mgmt utility, but I am not  100% sure, because it happens without it too. I just noticed that after reboot, hwinfo first shows normal values for Temps/FanRpms, but looses it after several seconds after start of RMClock.

As I run it in sensors-only mode, it does not generate reports in html/txt, only debug logs. Please see gdrive link for 8 debug logs below. https://drive.google.com/folderview?id=0B_JhIAmG1wOvR2FYbFdfVUFLOEE&usp=sharing

1st time, I enabled debugging, restarted HWiNFO and it came up without fan info
2nd time, I quit hwi, sleept/resumed and starte hwi, this time both temp and fan info was OK
3rd time, rebooted laptop, hwi started without both MB/Ext temps and fan
4th time, quit hwi, start hwi - temps came back, no fan rpms
5th time, sleep/resume (not quitting hwi) - temps became 255.9, no fan
6th debug - quit hwi, start hwi - both temps and fan OK
7th debug - quit hwi, start hwi - tempsOK, fan missing again. (That happened when FAN was down - 0 rpm, set by hwinfo during 6th debug run. Could it be the case that if fan rpm=0, that it is not found?)
8th debug - without any sleep/resume, just restarting hwi, both temps and fanrpms are back (this time fan was at 2200rpm)

runs 4 to 8 are done without RMClock.
 
Can you reproduce the problem also when not using custom fan settings (all fan settings at default) ?
 
Best would be to "Reset Preferences" and then power off the machine and switch on.
Don't forget to backup your settings before in case you want to keep your existing settings.
 
I uninstalled and reinstalled 5.12 (stable x64). I uploaded 9th debug file - hwinfo started with temps and rpm, then I performed sleep/resume and after resume hwinfo showed 255C/0rpm
 
Well, that looks to me as a bug in the BIOS, which after suspend/resume doesn't properly re-initialize the SMSC fan controller.
 
Martin said:
Well, that looks to me as a bug in the BIOS, which after suspend/resume doesn't properly re-initialize the SMSC fan controller.

why then, quitting hwinfo and starting it again, cures the problem?
 
Hm, so if you restart HWiNFO then it's back to normal? Does that happen also when you're not using custom fan control?
I'm asking because if you're using custom fan control, then after suspend/resume HWiNFO does reinitialize fan control tables which are overwritten by the BIOS during resume.
 
Martin said:
Hm, so if you restart HWiNFO then it's back to normal? Does that happen also when you're not using custom fan control?
I'm asking because if you're using custom fan control, then after suspend/resume HWiNFO does reinitialize fan control tables which are overwritten by the BIOS during resume.

How do I get rid of custom fan profile? I uninstalled it, cleanded up registry. Still, after installing 5.12 my custom fan setting is there

EDIT: After couple of reboots, custom fan profile is gone. I do not understand how.
 
btw, what are all 4 temp readings that fan control can respond to?

In the report I see CPU core 1,2 temps, and under SMSC EMC2103-2 I see Ext1, Ext2, and Motherboard temp.
But in fan profile there is Ext1, Ext2, Push1 and Int. Is Push or Int related to CPU cores somehow? I'd prefer fan to react to Core Max T.
 
After clearing HWiNFO preferences you need to make a reboot to reset everything back to default.
This fan controller can react only to temperatures measured by the SMSC chip, not external values like Core N (DTS).
Ext1/2 are temperature sources that the SMSC measures - one of them should be the CPU temperature, but I'm not sure which one for your model (I haven't worked with those HPs for pretty long time). You will need to check which of those is rising when the CPU is under load.
Int is the internal SMSC chip temperature and since it's placed on the mainboard it can be used to measure the mainboard temperature.
 
what happens after clean reinstall, is EMC2103-2 is not reported missing anymore upon start. Still, sometimes, after couple of minutes of observation, EMC2103-2 reported temps are all turn 255.9C. I also found that if temp reading are ok, and fan rpm is zero (on cold wakeup), sometimes HWI does not see (or does not show) fan until it is starts rotating.
 
When any fan is not spinning, HWiNFO doesn't show it. This is because it's not possible to tell whether a fan that returns 0 RPM is just not spinning or not present (a hardware limitation). This is to avoid having lots of 0 RPM fans in the list which in fact might not be present at all.

I'm still not sure what happens there and why the chip looses some information like temperatures and fans. From the dumps you provided it looks like the measurement logic in the chip doesn't provide valid data (as if they would be out of spec or similar). In one of those dumps, the chip didn't respond to any requests so it wasn't found at all...
This might also be caused by a collision with some other system monitoring tool. Are you maybe running other such tools in parallel ?
 
Martin said:
When any fan is not spinning, HWiNFO doesn't show it. This is because it's not possible to tell whether a fan that returns 0 RPM is just not spinning or not present (a hardware limitation). This is to avoid having lots of 0 RPM fans in the list which in fact might not be present at all.

I'm still not sure what happens there and why the chip looses some information like temperatures and fans. From the dumps you provided it looks like the measurement logic in the chip doesn't provide valid data (as if they would be out of spec or similar). In one of those dumps, the chip didn't respond to any requests so it wasn't found at all...
This might also be caused by a collision with some other system monitoring tool. Are you maybe running other such tools in parallel ?

let me test it for couple of days with /without RMClock. and with/without Efmer Tthrottle.
RMClock manages CPU voltages/frequencies
Tthrottle throttles selected apps when certain CPU temps is exceeded.

I will come back after couple of days. Thanks a lot for your help
 
I appreciate that ancient EMC controller on an ancient laptop might not be the most exciting thing to debug. I wonder, if it possible to make a build that would check if this EMC controller is missing or returns 255.9C(0xFF?) for Ext1, Ext2, etc and laptop model is 2530p, then it would silently restart itself (i.e. restart hwinfo)?

In the alert section, you already have "start another program". Would it be possible to add there "restart hwi", or "quit hwi"?
(If hwi auto quits, then I might try restarting it using TaskScheduler)
 
I don't like this idea, as I believe that problem should be solved in a different way. Understand why it happens and how to fix it on the fly.
But for that I still need to know if it also happens without custom fan control (so that I can be sure it's not a problem of that mode).
 
OK, let me know if you have test results when running without those additional tools. I want to make sure first we can confirm or exclude such a conflict.
 
Back
Top