Wrong min and max values on MSI Z490 Tomahawk

va3sd

Member
Hi, everyone!
Who can explain me why I can see such a weird values (even positive where 96 is wrong or 3.080V on cpu...)
And how to fix it?
10850K+Z490 Tomahawk
 

Attachments

  • Screenshot_6.png
    Screenshot_6.png
    28.3 KB · Views: 9

Martin

HWiNFO Author
Staff member
Hmm, perhaps running some other monitoring or RGB control tool that could be causing a conflict with HWiNFO?
 

va3sd

Member
Hmm, perhaps running some other monitoring or RGB control tool that could be causing a conflict with HWiNFO?
Hm, I used Aquacomputer Octo, but since yesterday do not use anything except one flow meter also from aquacomputer. Could it be reason?
 

Martin

HWiNFO Author
Staff member
This really looks like some other software accesing sensors is causing a conflict. Are you sure you don't have anything else running?
 

va3sd

Member
Found issue, but I do not know who is responsible for this bug and who could fix it.
 

Attachments

  • Untitled.png
    Untitled.png
    143.9 KB · Views: 7

Martin

HWiNFO Author
Staff member
Aquasuite. They are accessing the eSIO sensor and don't synchronize with others.
 

SpeedyIV

Well-Known Member
While working on a RAM overclock I discovered that Aquasuite (AQS) adds about 10ns of memory latency. I spent quite a bit of time trying to figure out why my memory latency was always around 65ns when I saw so many people with similar hardware and overclocks with memory latency in the low to mid 50ns range. Turns out it was AQS, which my setup requires to be running. I have all AQS direct sensor polling disabled and use data imported from HWINFO. I have not noticed if AQS also causes an increase in VCore (or at least reported VCore) but will check. Did you get any feedback on the AQS forum about this?
 

va3sd

Member
While working on a RAM overclock I discovered that Aquasuite (AQS) adds about 10ns of memory latency. I spent quite a bit of time trying to figure out why my memory latency was always around 65ns when I saw so many people with similar hardware and overclocks with memory latency in the low to mid 50ns range. Turns out it was AQS, which my setup requires to be running. I have all AQS direct sensor polling disabled and use data imported from HWINFO. I have not noticed if AQS also causes an increase in VCore (or at least reported VCore) but will check. Did you get any feedback on the AQS forum about this?
I already asked them on their forum https://forum.aquacomputer.de/weitere-foren/english-forum/p1460121-new-aquasuite-x/#post1460121
If I understood correctly nobody from Aquacomputer team answered yet, just some other members like me gave me advice do not use this monitoring in Aquasuite and use sensors directly from hwinfo with 'Shared memory support' (what I have already tried and was disappointment because of triggered outputs if no hwinfo instance running.
Probably I am the only one who use outputs with complicated rules which automatically turn off my PC and send me a few notifications.

BTW: when I tuned my memory I also noticed that I have latency 45ns instead of 39ns, but I did not check what is the reason.
 

SpeedyIV

Well-Known Member
I post on the AQS forum now and then. A company representative may or may not respond. They are not as attentive and prompt as Martin is on this forum. I read that thread and understand your point that if HWINFO is not running for whatever reason, AQS will not get any data, which in your case may cause an auto-shutdown alarm to trigger.

I have the Alarm output of my Octo wired to the mobo power switch header (with Aquacomputer #53216 cable). The alarm is configured to send a momentary closure to cause an immediate shutdown if the pump speed passes a minimum value. A secondary alarm sends a shutdown command to Windows. The pump is plugged into an Octo fan port so the speed data is reported directly, even when AQS is not running. The Windows Alarm requires AQS to be running but the wired alarm does not. I would not want my alarms to be dependent on or affected by data being imported from HWINO because too many things have to be working (nothing against HWINFO). Maybe you should re-think how you have your alarms set up, and see if you can configure them so that they are not dependent on imported data or virtual sensors. AQS can do a lot of crazy stuff, but IMHO an alarm circuit should be as basic and autonomous as possible.

I do agree that if Aquacomputer is not polling the eSIO sensor properly, they should correct that. Stay on them and maybe they will agree to do it in the next update. I have gotten them to fix several bugs I found so they do listen. Sometimes it takes them awhile. Creating a ticket should definitely at least make they aware of it. I will keep an eye on your thread on the AQS forum and post any further comments there. This is really not a HWINFO issue. Good luck!
 

Martin

HWiNFO Author
Staff member
Saying they are not properly polling the sensor is not quite correct. They are most likely polling it right, but didn't implement mechanisms to work together with others accessing the sensor. HWiNFO and most other tools (AIDA64, CPU-Z/HWMonitor, SIV and some others) use synchronization techniques (global mutexes) that allow multiple applications to access it. I believe AQS doesn't so it results in collisions when accessing the sensor.
 

SpeedyIV

Well-Known Member
I am not a programmer but I have heard plenty about global mutexes from Ray (author of SIV). I was around when he got into it with Corsair about their refusal to use them. Aquasuite can directly poll a lot of system sensors if enabled to do so. I have all of that disabled since the shared memory import works quite well but I am curious. Do they not use synchronization methods (global mutexes) to monitor any of the sensors they can poll or is it just this one? I thought they were on the list of companies that do things the “right way” but maybe not?

Separately, I also have overview pages set up in AQS that display things like CPU core temps, DIIMM temps, etc. The values come in from HWINFO. Sometimes a few of them stop reporting and I have to click on them (in AQS), then Change Data Source, then reselect the item from HWINFO. Then it will work again. This has happened a few times but it’s not always the same items. I don’t know if something changed in HWINFO or AQS, or the shared memory, or how to tell. When it happens I just reselect them but it’s a pain when a sensor item that dropped out appears in multiple overview pages because I have to reselect them in each place they appear. I have not noticed that this happening correlates to updating either program. Do you know what may be causing the “link” to stop working?
 

Martin

HWiNFO Author
Staff member
I am not a programmer but I have heard plenty about global mutexes from Ray (author of SIV). I was around when he got into it with Corsair about their refusal to use them. Aquasuite can directly poll a lot of system sensors if enabled to do so. I have all of that disabled since the shared memory import works quite well but I am curious. Do they not use synchronization methods (global mutexes) to monitor any of the sensors they can poll or is it just this one? I thought they were on the list of companies that do things the “right way” but maybe not?

I can't tell this exactly as I have no insight into their code. There are several places and different methods to be used for proper synchronization, I assume they just forgot to implement it for accessing the NCT668x eSIO.

Separately, I also have overview pages set up in AQS that display things like CPU core temps, DIIMM temps, etc. The values come in from HWINFO. Sometimes a few of them stop reporting and I have to click on them (in AQS), then Change Data Source, then reselect the item from HWINFO. Then it will work again. This has happened a few times but it’s not always the same items. I don’t know if something changed in HWINFO or AQS, or the shared memory, or how to tell. When it happens I just reselect them but it’s a pain when a sensor item that dropped out appears in multiple overview pages because I have to reselect them in each place they appear. I have not noticed that this happening correlates to updating either program. Do you know what may be causing the “link” to stop working?

No idea, but when it happens, use the Shared Memory Viewer: www.hwinfo.com/beta/HWiNFOSharedMemoryViewer.exe.7z
If it still shows values (you'll need to click Reload), then all values are properly exported by HWiNFO and the problem is in AQS.
 

SpeedyIV

Well-Known Member
Wow. Thank you for making me aware of this tool and providing the link. I will definitely run this the next time values imported from HWINFO via shared memory support stop working in AQS. I suspect the issue is in AQS. If I can confirm that by using this tool maybe I can get them to look into it. I really like AQS and all of the things it can do, but I have encountered a fair number of bugs and my setup is pretty basic. It can be difficult to convince the AQS devs that there is really a bug, but once that happens they are pretty good about fixing them. Thank you again for providing this.
 

Martin

HWiNFO Author
Staff member
But there are ways how to mitigate such issues using synchronization between applications (via global named mutexes). They should contact me if they are willing to implement this.
 
Top