Sensor issue on Link on H100i

I don't think I'll receive much of a response or support due to having a way old system, but with the newest windows 10 x64 1607 update my system started reporting sensors that weren't there in the Link app, and not only that, impossible things, such as 65000 rpm on non existent fans, and extreme temperatures on the VRM/PCH (Temp 6) of up to 155 C sustained (The limit is 114 C). There's no fan 3, 4 or 5 present, and fan 1 is the mainboard connection to the H100i. Temperatures 7 and 8 don't exist either.

 

Any ideas as to if this could be fixed on a future version or should I just ignore it?

 

I added the Intel (the mainboard is intel) monitoring utility showing the actual values.

 

This is on the latest corsair link software, 4.2.4.25

 

Is there any way to get this feedback to the developers? Any suggestions would be great. I know the important temperatures are the actual liquid temperature and the pump (Since I can get everything else elsewhere), but it is at least worth noting for the developers.

As CL4 uses the CPUID SDK then you best option is to check what HWM reports which I suspect will report the same rubbish values. If this is so then report the bug to the HWM developers and a future CL4 release should have the issue fixed. I guess you could report it to Corsair, but this is unlikely to get the issue fixed. Looking at http://forum.corsair.com/forums/showthread.php?p=864155 you can see this is what I did.

 

As to why the values are reported, what they should be and if it's safe too ignore them I would need more information. See http://forum.corsair.com/forums/showthread.php?p=788225 for how to generate the information I would need.

As CL4 uses the CPUID SDK then you best option is to check what HWM reports which I suspect will report the same rubbish values. If this is so then report the bug to the HWM developers and a future CL4 release should have the issue fixed. I guess you could report it to Corsair, but this is unlikely to get the issue fixed. Looking at http://forum.corsair.com/forums/showthread.php?p=864155 you can see this is what I did.

 

As to why the values are reported, what they should be and if it's safe too ignore them I would need more information. See http://forum.corsair.com/forums/showthread.php?p=788225 for how to generate the information I would need.

 

Thanks for the reply. Interesting, because I usually use HWInfo (They're extremely accurate) and it actually reports the correct values, but I installed HWmonitor and well... it does not report the extra stuff and it does report the correct values for everything except the two fans. The maximum value on the fans is the actual, and the actual pops in every now and then but otherwise goes to 0 rpm.

I was going to get you the information you wanted from the non pro version... but the SDK website redirected me to the Pro version... and once you install that one it blocks all fan readings unless you buy it.

 

I'll just use the intel software for the mainboard and the Corsair software for the actual H100i. Kind of interesting though. I don't want to go on a registry expedition. Thanks again.

Oh crap, I found what the problem was. I got curious and ran system information viewer and the intel drivers instantly said that the memory temperature was frying the board (not true).

 

I was pretty sure that it was HWinfo64 software's kernel driver. I uninstalled it and everything is awesome again, including the H100i sensors and the intel sensors.

.

I guess I won't be using that again. Thanks for the threads red-ray.

Spoke too soon, the system information viewer does the same thing as HWInfo64.

Spoke too soon, the system information viewer does the same thing as HWInfo64.

 

I am none too sure what HWiNFO did. Does SIV run OK? I could comment further is you posted the SIV initial and [status] screen shots.

 

BTW I expect the Intel utility fails to interlock access to the sensors which could cause all of CL4 + HWiNFO + HWM + SIV to report erroneous information. See http://forum.corsair.com/forums/showthread.php?t=141037 for how to check.

I am none too sure what HWiNFO did. Does SIV run OK? I could comment further is you posted the SIV initial and [status] screen shots.

 

BTW I expect the Intel utility fails to interlock access to the sensors which could cause all of CL4 + HWiNFO + HWM + SIV to report erroneous information. See http://forum.corsair.com/forums/showthread.php?t=141037 for how to check.

 

Pretty much, if I run anything other than the Intel Desktop Utilities, the desktop utilities and the programs go to crap. It requires an NVRAM clear (or Bios clear) to get the sensors reporting OK again, and it even reports crap in the UEFI temperature monitoring (not the extra fans, but the PCH goes to 0 C and the memory temperature dissapears).

 

So running the CIV at all pops up an over temperature alert and contaminates the temperature readings. I can live with intel's readings being fine and Corsair Link having the fan sensors reading 0 rpm where none are.

 

I really don't want to clear the NVRAM again, it's pretty late here. I'll try uninstalling the utilities as well to see if that's the issue.

The issue is Intel utility failing to use locks. If you choose to run the Intel utility you should not run any of CL4 + HWiNFO + SIV.

 

What happens in you close down the Intel utility and then run HWiNFO or SIV?

 

Unless you post the SIV screen shots I can't comment further.

The issue is Intel utility failing to use locks. If you choose to run the Intel utility you should not run any of CL4 + HWiNFO + SIV.

 

What happens in you close down the Intel utility and then run HWiNFO or SIV?

 

Unless you post the SIV screen shots I can't comment further.

 

Alright. I left the computer running overnight, and the non existent devices came about, but not the exceedingly crazy values and the over temp alerts were gone (So HWINFO64 was def. a reason for those).

 

Fans 3,4 and 5 don't exist, and temperatures 7 and 8 are non existent (at least on the mainboard).

Well it doesn't like the Intel Desktop Utilities so your hunch was right. Even with the SM drivers installed (They come with the utilities), all is well as long as you don't have the utilities installed and its service running. It's kind of a bummer since the utilities is the only way to get 100% accurate PCH temperature readings (It's around 20C above real temperature otherwise), but it's all good.

 

Intel is no longer making boards or supporting the software due to them no longer making boards (I did some serious googling and found a similar issue on the intel forums). Thank you for your time.

Looking in the SIV save file I can see the SMBus (see Menu->System->SMB Bus) was locked up. This typically happens when utilities fail to use the locks. This is probably the root cause of all the strange effects. To get the SMBus working again you will need to reboot the system and may even need to power cycle it.

 

I can tell the SMBus was locked before you ran SIV as SIV failed to read the DIMM SPD data.

 

Were it my system I would expunge the Intel utility, reboot and then check that the SMBus was working and the DIMM SPD was reported. Until this is resolved strange things will continue to happen.

 

CL4 uses the CPUID SDK so I expect HWM will also report strange values. If so then report this as a HWM bug to the CPUID SDK developers and assuming they fix HWM in time CL4 will be fixed.

I uninstalled intel utilities and rebooted and all apps are well. I'll leave them uninstalled, like I said, they are the ones that need the fix and they won't, they're out of the mainboard business. Thanks though!

OK, what does SIV now report then you press [sPD] then [sMB Bus]?

OK, what does SIV now report then you press [sPD] then [sMB Bus]? Is the SMBus now working?

 

It reports:

 

SMBus Host Status 41, Timeout 999

 

But that's probably because I have an "impossible" overclock on the ram? It usually runs like that when I have it on that specific manual overclock (The RAM).

 

If I place it on an XMP or standard JDEC speed, it goes back to normal.

Ah now it says

[sMB Bus] <- SIV64X - System Information Viewer V5.12 DESKTOP-KCFJEHT::argen

 

# Device Name |Bus-Numb-Fun| Port Min Max RM GB SO CO Vendor-Dev-Sub_OEM-Rev [Y]INUSE Wait 0 of 812 ms Intel DZ77GA-70K

0 P7 SMBus [ 0 - 31 - 3 ] F000 BF 8086-1E22-20378086-04

 

Slave SMBus Device Host Peak Data Hex Data

Device Description Status Time Size Offset 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F

 

|0_10| <excluded> 00 0 0000

... <excluded>

|0_13| <excluded> 00 0 0000

|0_14| <exclusion added> 41 999 0000

...

|0_69| <excluded> 00 0 0000

...

|0_7F| 44 605 0000

 

Total of 1 SMBus and 0 SMBus Devices in 6.295 seconds (0.000 KBps). Slaves Probed 107. [scan Range]0_10 to 0_80. [.. -> 00 : __ -> FF]

 

[ OK ] [ Copy] [Windows] [Machine] [ DIMMs] [sensors] [uSB Bus] [ SPD ] [PMB Bus] [Volumes] [sMB Bus] [PCI Bus] [ALL Dev]

SMBus Host Status 41, Timeout 999

 

The SMBus is still locked up and you should resolve this. I suspect this is unrelated to the "impossible" overclock on the ram.

 

If you reboot the system and then do SIV64X.exe -DBGSMB what is in SIV_DBGOUT.log?

Then there's nothing I can do... It pretty much does this as long as I have an overclock on my system. I already removed the intel SM drivers and their stuff and cleared the Bios...

 

It seems to always do that when the system ram is overclocked.

If overclocking the RAM causes an SMBus lockup then you should remove the overclock.

It might be the Intel Watchdog in the EFI. I'm going to do a RAM test before I disable it, it might be locking the SMbus and I did run into an issue using SPD/JDEC timings and speed, the Motherboard didn't like the sudden switch and required an NVRAM clear and a bios clear.

If overclocking the RAM causes an SMBus lockup then you should remove the overclock.

 

That took a while... I guess the overclock I had was stressing the RAM, so after the 4th pass it failed randomly.

 

I lowered it to standard XMP values and stress tested it for 6 hours and no errors, but the SMbus issue seems to be part of a different issue with the family of mainboards : ​

https://www.hwinfo.com/forum/Thread-Solved-Sensor-Module-and-Intel-DZ77RE-75K-MB

 

Now, the SMbus shows up in HWinfo64 latest and greatest, but the only thing not showing is the SPD info.

 

It's kind of like this: https://www.hwinfo.com/forum/Thread-DIMM-incomplete-info

 

Same blank info in both that program and CPUZ.

 

I'll see if there's something in the old school BIOS mode... but all this happened after the Win 10 upgrade... which I did also switch from regular old BIOS mode to EFI boot (don't know if that has anything to do with it).

 

The mainboard is not giving any indications of anything wrong either, and there's no NVRAM indications that the SMbus is releasing bad data in windows 10.

There's no memory monitoring options to disable or anything of the sorts. I'll try with my old memory and see what happens. Worst case scenario, I have a set of bad memory (possible). Too bad because it is quite new too, less than a month.

I switched to SPD only RAM and, the same SMbus error persists, HOWEVER, the bad devices listed in the Corsair Link software are gone. No extra fans, not a single thing wrong, even with the SMbus unavailability in SIV. I might have a bad stick on that other set? It makes no sense. They passed a 6 hour memtest86 torture test, they didn't even give in to Row Hammer tests.

If you reboot the system and then do SIV64X.exe -DBGSMB what is in SIV_DBGOUT.log?

 

the SMbus issue seems to be part of a different issue with the family of mainboards: ​https://www.hwinfo.com/forum/Thread-Solved-Sensor-Module-and-Intel-DZ77RE-75K-MB

 

Looking at the HWiNFO thread then you need to add an SMBus exclusion for 0x60. To do this do Menu->Tools->Configure->SIV SMBus Setup, select Slave 0_60 and then reboot the system.

 

Use the above command to trace the SIV start-up.

 

Hm... must be another device. It still doesn't work, but everything seems to work. I quit. I mean... seriously quit. I wish I had time to troubleshoot it via debugging but it's just not worth it. I can live without SPD, I don't use it, and the only things showing up are 2 or 3 temperatures that don't exist in the Link software.

 

I hope anyone that sees the thread online gets some sort of relief that it is happening on that board (or not, it's quite old).

 

Thank you very much for the attempt and patience.