Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Weird comparison HWiNFO32 and CPUID Hardware Monitor sensors
#1
Your Temperature 2 and Temperature 3 seem to correspond to the two CPUID readings for the Radeon HD 4350.
The CPUID readings are clearly wrong, with a negative minimum. All four of those numbers, two from you and two from CPUID, have max numbers that seem insanely high. The PC would have been on its knees.

PC is new Pavilion HPE-110f desktop. Running beautifully. Every part of case is cool. Can you help me understand those temperature readings?


Attached Files
.gif   ScreenShot005.gif (Size: 29.71 KB / Downloads: 2,130)
.gif   ScreenShot006.gif (Size: 55.22 KB / Downloads: 2,130)
Reply
#2
I don't think there's a relation to IT8720 and GPU values.
Since the sensor chips (like IT8720) are used for general monitoring of various values, they can be connected to various inputs using different methods. Unfortunately there's no generic way to dermine which sensor input is connected to a particular voltage/temperature/fan. Moreover, some sensor inputs are often not connected to any real voltage/temperature/fan and return bogus/invalid values.
Thus every particular motherboard mostly needs special adjustments for sensor values reported (depending on model) - some values need to be multiplied/divided by certain constants and some need to be ignored (if not connected). For this, I need the HWiNFO32 Report + Debug File + BIOS Screenshot reporting real values (or manufacturer's monitoring utility screenshot), then I'm able to adjust this for most motherboards, but not all of them.

cnmoore Wrote:Your Temperature 2 and Temperature 3 seem to correspond to the two CPUID readings for the Radeon HD 4350.
The CPUID readings are clearly wrong, with a negative minimum. All four of those numbers, two from you and two from CPUID, have max numbers that seem insanely high. The PC would have been on its knees.

PC is new Pavilion HPE-110f desktop. Running beautifully. Every part of case is cool. Can you help me understand those temperature readings?
Reply
#3
Martin Wrote:I don't think there's a relation to IT8720 and GPU values.
Since the sensor chips (like IT8720) are used for general monitoring of various values, they can be connected to various inputs using different methods. Unfortunately there's no generic way to dermine which sensor input is connected to a particular voltage/temperature/fan. Moreover, some sensor inputs are often not connected to any real voltage/temperature/fan and return bogus/invalid values.
Thus every particular motherboard mostly needs special adjustments for sensor values reported (depending on model) - some values need to be multiplied/divided by certain constants and some need to be ignored (if not connected). For this, I need the HWiNFO32 Report + Debug File + BIOS Screenshot reporting real values (or manufacturer's monitoring utility screenshot), then I'm able to adjust this for most motherboards, but not all of them.
Unfortunately HP doesn't provide any monitoring utility for anything except performance - no sensor info. And there is no sensor info in the BIOS setup. So I don't know how to provide screenshot.

How do I run debug? Do I just set DebugMode=1 in the HWiNFO32.ini Settings section?

I would run for an hour or so to build up min/max sensor values.
For the Report - Text logfile format? Do you want the whole thing or just the sensors part? I.e. when I create report should I uncheck most of the sections?
Reply
#4
Instructions how to supply debug info can he found here:
http://www.hwinfo.com/forum/Thread-IMPOR...g-a-report
but basically, you're right and after enabling Debug Mode, supply the HWiNFO32.DBG file after you open sensors and close the application.
HTM or TXT report are the best ones, please include all possible items and don't forget to open the Sensors.

I'm not sure if I'll be able to adjust sensor values for the HP machine without seeing real values. For the Foxconn, that might be possible.

cnmoore Wrote:Unfortunately HP doesn't provide any monitoring utility for anything except performance - no sensor info. And there is no sensor info in the BIOS setup. So I don't know how to provide screenshot.

How do I run debug? Do I just set DebugMode=1 in the HWiNFO32.ini Settings section?

I would run for an hour or so to build up min/max sensor values.
For the Report - Text logfile format? Do you want the whole thing or just the sensors part? I.e. when I create report should I uncheck most of the sections?
Reply
#5
Here you are.


Attached Files
.zip   HWiNFO32.zip (Size: 138.23 KB / Downloads: 100)
Reply
#6
Thanks. So this is a HP machine with a Foxconn board..
Can you please download and run the FOX ONE tool from Foxconn site to see if it will report sensor values?

cnmoore Wrote:Here you are.
Reply
#7
This mainboard seems to be a special model manufactured for large OEM. I have checked the data and all voltages returned except +5V and 5VSB seem to be invalid values. I'll mask them off. Fan labels should be probably reversed. I have no idea about temperatures and since there are no reference values available (BIOS, manufacturer's utility) it's not possible for me to adjust them or assign correct labels.
Reply
#8
Edit: I hadn't seen your last reply. I'm thinking the voltages may be correct but the labels are obsolete?
------------------------------------------------------------------------------------------------------

Good to know where the FOXCONN site is. But they do not have ALOE listed as a supported board - presumably HP exclusive? And HP provides essentially no information. This is all there is:
<!-- m --><a class="postlink" href="http://h30434.www3.hp.com/t5/Hardware/Location-of-temperature-sensors-on-Aloe-motherboard/td-p/201719">http://h30434.www3.hp.com/t5/Hardware/L ... d-p/201719</a><!-- m -->
One of the forum visitors there says
Quote:No information on-line to answer this question definitively (I've looked everywhere), but my very good guess is that TMPIN1 is the northbridge chip. My guess is based on temp fluctuations during video playback and heavy browser client use (with Flash, Silverlight, etc.). TMPIN0 appears to be measuring the temp on the motherboard in a location not directly associated with a CPU or bridge chip, as it runs the coolest on my system.
TEMPINI1 is your Temperature 2.

Good news! the weird high readings have disappeared. I think this is because I turned off Logitech keyboard support.

Other than the crazy CPUID GPU readngs, there is now only one significant discrepancy between your readings and those of CPUID: the max CPU fan speed is lower in yours. (Both programs started at same time and running concurrently).


Attached Files
.gif   ScreenShot007.gif (Size: 2.63 KB / Downloads: 2,120)
.gif   ScreenShot008.gif (Size: 2.69 KB / Downloads: 2,120)
Reply
#9
I can't imagine how Logitech keyboard support could influence the values reported.
Since most voltages return ~2.16V I think they are invalid (even +12V returns 4*2.16), thus I have filtered them out.
The different max fan values are possible, since it might happen that one app catched the highest level during it's refresh cycle (just a coincidence).
Based on the data you supplied, I adjusted the values and you can try this build:
http://www.hwinfo.sk/beta/hw32_346_702.zip

cnmoore Wrote:Edit: I hadn't seen your last reply. I'm thinking the voltages may be correct but the labels are obsolete?
------------------------------------------------------------------------------------------------------

Good to know where the FOXCONN site is. But they do not have ALOE listed as a supported board - presumably HP exclusive? And HP provides essentially no information. This is all there is:
<!-- m --><a class="postlink" href="http://h30434.www3.hp.com/t5/Hardware/Location-of-temperature-sensors-on-Aloe-motherboard/td-p/201719">http://h30434.www3.hp.com/t5/Hardware/L ... d-p/201719</a><!-- m -->
One of the forum visitors there says
Quote:No information on-line to answer this question definitively (I've looked everywhere), but my very good guess is that TMPIN1 is the northbridge chip. My guess is based on temp fluctuations during video playback and heavy browser client use (with Flash, Silverlight, etc.). TMPIN0 appears to be measuring the temp on the motherboard in a location not directly associated with a CPU or bridge chip, as it runs the coolest on my system.
TEMPINI1 is your Temperature 2.

Good news! the weird high readings have disappeared. I think this is because I turned off Logitech keyboard support.

Other than the crazy CPUID GPU readngs, there is now only one significant discrepancy between your readings and those of CPUID: the max CPU fan speed is lower in yours. (Both programs started at same time and running concurrently).
Reply
#10
Re Logitech - just coincidence I guess.

New build looks very nice. Thank you! I'll let it run forever and let you know if anything odd develops.

I'm going to ignore CPUID Hardware Monitor. I think they are just guessing or something. PC is a rather new model.

I'm really surprised that HP makes essentially no technical info available, and that CPUID which is generally well regarded has come up with nonsense. So glad I found your excellent utility.
Reply
#11
I suppose this should be in bug reports. I didn't want it to sound like complaints..
The sensors page is not picking up the CPU 0 temperature - you have a 37C 37C 37C flatline..
It is identified as AMD K10+ which isn't correct.
You have the CPU identified (correctly) in the System Summary, as AMD Phenom II X4 925, but show the temperature there as N/A.
I understand that the chip does have one sensor (but not 4 separate ones) and it's not a real core temp but is a sensor on the outside of the chip. Core Temp reads it, see pic.

<!-- m --><a class="postlink" href="http://www.sevenforums.com/662773-post25.html">http://www.sevenforums.com/662773-post25.html</a><!-- m -->
Quote:According to AMD, the maximum sustained operating temperature for a 965 BE is 62C, and that temperature is not the core temperature, it is the the temperature classified as Tcase (And it's completely different from Intel's temp rating) on a spec sheet, and it is the temperature measured on the outside of the core, on the metal part that the heat sink touches.
Also, in the Summary you identify the Chipset as AMD RS880 + SB750.
HP says it is AMD 785G Chipset. <!-- m --><a class="postlink" href="http://h10010.www1.hp.com/wwpc/ca/en/ho/WF06b/12454-12454-3329740-64546-64546-4079700-4107866.html">http://h10010.www1.hp.com/wwpc/ca/en/ho ... 07866.html</a><!-- m -->

And - mentioned before - you don't seem to catch the occasional fan spinups. This could well be because you don't scan the sensors often enough to catch them, and I'm in favor of not scanning too often, so that seems perfectly OK. But I would like to see the CPU 0 temp so I hope you can find it. If not, I can always have Core Temp running, so I am mostly telling you all this because this PC is likely to be a popular model and I imagine you want to get it right. Yours is the only monitor I've seen that has the Radeon GPU readings correct.


Attached Files
.gif   ScreenShot010.gif (Size: 15.42 KB / Downloads: 2,114)
Reply
#12
Thanks for the report.
The K10 sensor issue is a serious one that I haven't noticed yet. I fixed it. You're right that AMD K10 CPUs have just one sensor per CPU (and not as some programs display - one sensor per core).
The name of the sensor is "K10", because this represents the whole family of AMD CPUs (although the correct family name should be Family 10h).
RS880 is the codename of a family of chipsets (including 785G and few others). I currently do not know how to distinguish them. I'll fix this as soon as I have the information.
Please try the following build to see if the K10 sensor reading is fixed:
http://www.hwinfo.sk/beta/hw32_346_704.zip
Reply
#13
Excellent. Thanks!


Attached Files
.gif   ScreenShot011.gif (Size: 8.73 KB / Downloads: 2,110)
Reply
#14
HWiNFO32 recorded a CPU fan spin up! 3947 RPM. So that is fine. It's not missing anything.
CPUID Hardware Monitor saw it as 96428 RPM - haha fan would have flown right off the CPU. Big Grin

Unimportant - but can I somehow change the Sensors screen to say "AMD Phenom II X4 925" instead of "AMD K10+"? Like edit one of the files in the HWiNFO32 folder? (I have a hex editor).
Reply
#15
Smile
Changing the sensor name is not possible for the user. I will think about using a more precise name in the future..
cnmoore Wrote:HWiNFO32 recorded a CPU fan spin up! 3947 RPM. So that is fine. It's not missing anything.
CPUID Hardware Monitor saw it as 96428 RPM - haha fan would have flown right off the CPU. Big Grin

Unimportant - but can I somehow change the Sensors screen to say "AMD Phenom II X4 925" instead of "AMD K10+"? Like edit one of the files in the HWiNFO32 folder? (I have a hex editor).
Reply
#16
cnmoore Wrote:HWiNFO32 recorded a CPU fan spin up! 3947 RPM. So that is fine. It's not missing anything.
Unfortunately that is the only reading I have ever seen where the max values recorded were any higher than
CPU Fan 1610 RPM
Chassis 979 RPM
As I remember, the CPU Fan 3947 disappeared even though the program was not shut down.

Been running continuously. Could you please check and see whether perhaps the max readings are seen but not being recorded as max?
Reply
#17
Min/Max values should be recorded properly, but they are remembered only while the Sensors window is open. When you close it, those values are lost.
How long might such a spin up take? I think the problem could be the sensor refresh cycle (~2 sec). When a value changes for a very short time (much less than the refresh cycle), then it's quite possible it won't be catched.

cnmoore Wrote:
cnmoore Wrote:HWiNFO32 recorded a CPU fan spin up! 3947 RPM. So that is fine. It's not missing anything.
Unfortunately that is the only reading I have ever seen where the max values recorded were any higher than
CPU Fan 1610 RPM
Chassis 979 RPM
As I remember, the CPU Fan 3947 disappeared even though the program was not shut down.

Been running continuously. Could you please check and see whether perhaps the max readings are seen but not being recorded as max?
Reply
#18
Watching the numbers updating, I judge that CPUID Hardware Monitor and HWiNFO32 update at the same rate, looks like about every one second. The current values don't change fast and when they do change it is only by a few RPM. If as you suggest, the peaks don't last long, where is CPUID getting the max numbers?

The two programs have both been running since yesterday.

CPUID HM shows these max values:
FANIN0 2393 RPM
FANIN1 3590 RPM

HWiNFO32 shows
CPU Fan 1610 RPM
Chassis 979 RPM

Even if occasionally a high reading wasn't caught, I believe very few would be missed. The two programs agree well on the fan Min speeds.
Reply
#19
I have a plan. Running in debug mode. If CPUID sees a high number I will wait a few seconds, exit, and if my wild guess is right, the number will be there in your log. Debug is a neat feature.
(Naturally, ever since I started debug mode neither program has seen any fan speed increase...)
Reply
#20
Heh, ok let me know.. You can try to put load on the system to force fan speed increase.
cnmoore Wrote:I have a plan. Running in debug mode. If CPUID sees a high number I will wait a few seconds, exit, and if my wild guess is right, the number will be there in your log. Debug is a neat feature.
(Naturally, ever since I started debug mode neither program has seen any fan speed increase...)
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)