[HWiNFO32 Thread] - Questions, Reports, Suggestions, Discussions

Hi everybody,

since I've seen that there are many people using HWiNFO32 on this forum, there are several questions and reports posted, and I'm also pleased by the community and feedback here, I have decided to open a thread for HWiNFO32 on NBR.
I'm open to discuss things related to HWiNFO32, answer questions, clarify things, solve issues, respond to feature requests, etc.
So please feel free to use this thread !

Note: The HWiNFO32 site does also have a forum, you can also use that one if you think it's better.

Regards,
Martin
[HWiNFO32 Author]

So.. the first issue that has been discussed here was a slow scanning of ATI GPU sensors (startup). I tried to make an improvement, but this needs to be tested if it works properly and faster.
Here's the link to a fresh Beta release:
HWiNFO32 v3.50-720
Please provide feedback..

 

Is there a HWiNFO32 Windows 7 Gadget that displays CPU/GPU Temps?

 

Yes there is. Install the latest version and enable the gadget. Then all you need to do is choose which metrics you want displayed.

 

+ install the Gadget supplied with HWiNFO32 (run the HWiNFO32.gadget file from the package).

 

Awesome, thanks.

 

Thanks, forgot all about that.
That actually might be something else that needs attention, the gadget could be automatically activated if chosen from the startup options (much like everest).

This beta that you supplied works beautifully!
I've measured startup times (from startup up to when the readings are displayed in windows gadget) of HWiNFO32's latest stable version, the beta you linked to and everest :

1. HWiNFO32 Stable : 41.5 Seconds
2. HWiNFO Beta : 10.8 Seconds (!!!)
3. Everest 5.50 Ultimate - 13 Seconds

I am testing stability, but so far it looks great.

Thanks for your attention Martin!

 

Martin. One thing that we discussed before but didn't find a solution for is displaying alienware's fan speeds (is controlling them is wishing for too much?). Could you tell us what is it you need exactly to be able to read those?

 

The problem is that this is probably not done a generic way (using a generic sensor) and I have no information about this.
As many notebooks do, such things are usually controlled via on-board's Embedded Controller and its firmware. Although there's a generic interface used to communicate with such microcontroller (and HWiNFO32 supports this interface), the register layout and particular methods are fully vendor-specific. HWiNFO32 currently supports few notebook vendor's ECs (like IBM, Compal, DELL), but this is a very sensitive area. Besides system health monitoring and control, the EC implements other vital functions (like battery management, etc), so messing with an EC (without proper and exact knowledge) can be very dangerous and cause serious problems or damage to the PC. That's the reason why I don't like to experiment with this much.
I believe the only way here is getting information directly from vendor, however IMO there's a litte chance of success.
There are few things that can be determined about the EC by deciphering ACPI DSDT, but this is also quite difficult and without a guarantee. I tried this for your NB, but without success so far...

 

HWiNFO32 v3.51 available.
Changes:

• Fixed possible crash after having graphs opened for longer time.
• Updated reporting of IDA capability.
• Added reporting of VIA P4M800 memory timings.
• Reduced ATI I2C sensor scan time.
• Added support of Atmel DIMM Temperature sensor.
• Improved sensor monitoring for MSI NF980-G65 (MS-7612).
• Improved reporting of GPU memory clock for AMD Evergreen family.
• Added nVidia GeForce GTX 465.
• Improved sensor monitoring for ASRock 890GX Extreme3 and 890FX Deluxe3.
• Improved sensor monitoring for ASUS NCCH-DL and NCT-D(A).
• Improved monitoring of CPU/PCH (Ibex Peak) power consumption.
• Improved monitoring of nVidia GPU diode temperature.
• Added monitoring of nVidia GPU Fan duty cycle.

 

Hi Mumak,

Great program firstly. Very nicely put together.

Would it be possible that the program could launch straight to the "Sensors" section in a minimised way?

I ask I've got a M1730 and I'd like to include the program in the start up group and have the temps appear in the Logitech LCD at each start without the program screen popping up and without the need to start it manually each time.

Many thanks,
72oo

EDIT: Have found what I needed from mfractal! Cheers

 

I start the program minimized using the following config


Or is it the little screen that shows the initialization that's bothering you ?

 

Rep spread! I must ensure to check the actually preferences page in future!

 

Mumak
Been using HWiNFO32 for a while now & would like to thank you for this great tool.Great work :wink:

 

Martin, how's the IT8512E controller reading going? Monitoring m17x fans would be awesome! BTW - is there anything else meaningful that can be read from the controller ?

 

mfractal was faster with the answer Thanks Michael.

Thanks I'm improving it everyday

Unfortunately not so well yet.
It has an embedded microcontroller (8251 type), but the registers monitoring fans and few voltages are in private (EC) space and don't seem to be accessible from host (CPU) side. It is possible that they could be exposed via EC access methods, but how exactly is vendor-specific. Would be great if the vendor could tell this, but I doubt it... I'll spend more time with it, but it doesn't look good so far.
BTW, does the BIOS or any other vendor tool report those values (fan, voltage, etc) ?

 

I'll ask around, not sure. i'll let you know..

 

That's great to hear.

I've a small (I swear!) request; would it be possible to display 4 lines of text on the LCD settings. I'd like to be able to see CPU1, CPU2, GPU1 and GPU2 temperatures all at once in the one screen.

 

I'm not sure if that would fit with the current font used. Would probably mean to use a quite small font.

 

fwiw, in my last m1730 i used Everest to display the temps in the following format :

First line : CPU 1/2: TEMP1/TEMP2
Second Line: GPU 1/2: TEMP1/TEMP2
Third Line: Fans CPU/GPU1/GPU2 : Number/Number/Number

That sums up the info i personally needed in three lines of 8 size font (Sorry don't remember the font i used)

 

HWiNFO32 v3.51-757 Beta available.
Changes:

• Improved graphical displaying of CPU frequency per core (new bar, color).
• Added reporting of GPU Fan duty cycle for ATI R7xx and Evergreen family.

 

thanks, going to check it out now.

 

Thank you

 

Martin, both GPU fan speeds are stuck at 30% no matter what the fans do. It's been happening in everest also and not just for me.
I believe this is an issue with crossfire systems. Not sure you have much to do about it though....

 

It seems, that in this case the GPU is not controlling its fan, but the IT8512E SIO/EC chip. Unfortunately I don't know how to read those values from this chip, since the EC<->Host interface (and thus a possible access to EC internal registers) seems to be vendor-specific and proprietary..

 

That's exactly what I'd like to see in the LCD.

 

I have added the possibility to choose LCD column too, so that will allow similar behavior. Please try: http://www.hwinfo.com/beta/hw32_351_761.zip
and give feedback.

 

Thanks very much for that, works well. Its get to have a feature added on so quickly!

 

Hello Mumak,

I have an ASUS G71Gx with an Intel T9600.
HWINFO32-V3.51-761 shows
Intel Dynamic Acceleration (IDA) Technology: Supported, Disabled
but Everest 5.50.2163 beta shows
Dynamic Acceleration Technology (IDA) Supported, Enabled
Who's right?

 

Hard to say without further details. Please run HWiNFO32 in Debug Mode and after closing it send me the HWiNFO32.DBG file. Then I can check it.

 

Hey Martin.
Got one more issue or you
I am here with my friends Alienware M17.
I noticed that HWiNFO's CPU temps differ from HWMontor and everest - they are 8-9 degrees higher.
Also, it could not at all read the temps of a dual ATI 3870 mobile crossfired cards.
Here's is a debug file + comparison screenshot

 

Mumak, thanks for the great program!

There's only one problem I've had, and that's been that it incorrectly shows the Clarksfield cpu multipliers/frequencies.

I'm not sure what has to be done, but last time I used HWiNFO32 on an i7-720qm, it reported all cores getting to much higher frequencies simultaneously then they are actually capable of doing.

 

Hi Michael,

The CPU Core Temperature depends on what Tj,max value the particular tool uses. For this CPU model, I use 105 C, HWMonitor probably 100 C so that's the 5 C difference. Do you know which tool is correct and what's the correct Tj,max value for this particular CPU part#? I don't and I guess nobody knows this exactly for every CPU prior to Nehalem (since Nehalem and later report the current Tj,max). Anyway, if you feel a different Tj,max value fits better, you can adjust it by right-clicking on the sensor value.
The whole CPU Core DTS monitoring is a very large topic, but I want to say one thing: the sensor used in pre-Nehalem CPUs is quite inaccurate and the more inaccurate the lower the temperature is (higher distance from Tj,max). I can't publish the exact delta parameters, but some CPUs reported values below a certain threshold gives absolutely unusable values. So the conclusion is that IMO using the DTS to monitor temperatures ~ room temperature doesn't provide real data. This is mostly for pre-Nehalem CPUs, since in the first generations the DTS was designed only to catch the critical temperature (Tjmax, so the distance ~ 0) - my private opinion. Newer (Nehalem and later families) feature/will feature a more precise DTS even at lower temperatures. However, I think that the minimum delta error is ~ 5 C at any temperature.
I just hope more people would realise this and not rely on the DTS values so much.

I'll check the GPU sensor. At least I can report DIMM temperatures and GMT

EDIT: One more thing: Intel published some Tj,max values for pre-Nehalem (Core, Core2 family) CPUs in the past, but it has been determined that not even these numbers seem to be correct. Moreover Tj,max values for certain CPUs vary from part to part and there's no way how to determine the correct value.

This is a known issue to me. It occurs on some Nehalem-family CPUs, because they are extremely sensitive to thread load and will ramp the core clock to Turbo frequency very fast. So sensitive, that even my method for measuring particular core frequency puts it into Turbo mode.
I don't have a solution for this yet, but I'm analysing what could be done.
Btw, Intel has already admitted that the CPU shouldn't be so extremely sensitive to load and raise the clock so fast. There will be an improvement from Intel for this, which could solve this issue as well.

 

That would be great if it's just something Intel can fix!

Thanks for the response.

 

very imformative, thanks a lot!
I did not know the temp reported is a delta from tjmax. I guess those numbers can be taken just as reference and not as precise measurements.

Would be nice to see GPU temps, thanks.
Also, NB temps are not reported in HWiNFO .. forgot to mention it.

 

Please try the following beta:
http://www.hwinfo.com/beta/hw32_351_765.zip
and let me know if the GPU temperature is correctly displayed on that M88 GPU.
Regarding the NB sensor, it might not be properly calibrated since it seems it doesn't provide valid values..

 

This program causes a truly bad latency in my notebook when hard drive temperature monitoring is enabled. Any fix for this instead of disabling it?

 

Is it really caused by HDD temperature monitoring or do you have the HWiNFO32 System Summary window open?
If the 1st one is the cause, then the HDD driver is faulty.

 

I don't have the System Summary window opened so it's the first case (on a side note, the system summary window also causes high latency). Other programs like HWMonitor or Everest doesn't have any latency related issue when monitoring HDD temperatures. It can be solved?

 

The System Summary latency issue is related to the issue with inaccurate clock (Turbo) on some Nehalem systems. I'm analysing other methods...
Does the latency occur when you disable monitoring of all sensors (make sure that the Sensor name is disabled too) except the HDD sensor ? If yes, can you please attach a screenshot of the sensor page?

 

Yes, it happens even when I disable everything except HDD. Here're some screenshots.

1. Everything disabled




2. Everything enabled except HDD




3. Only HDD enabled

 

Thanks for the detailed screenshots. That looks really interesting.
Could you please enable Debug Mode in HWiNFO32, then run the sensors with HDD monitoring enabled for few seconds, close HWiNFO32 and attach the HWiNFO32.DBG file it produced?

 

Sure, here it is, hope it helps.

 

GPU monitoring is working beautifully. Thanks Martin! Regarding NB sensor, it is displayed ok in hwmonitor, would it be possible to display it as well ?

 

Thanks. I did a quick check and everything looks ok from HWiNFO32 side. I just issue a simple call to the SATA driver to get temperature and that returns correct data. So I don't know what else could cause the latency issue other than the SATA driver itself.
I just talked to author of Everest and he said that Everest reads the HDD temperature only 3 times per minute. Since HWiNFO32 does this nearly every 2 seconds, that might be the difference why you see higher latency (in case this takes the HDD driver so much time).
I assume it's the Intel Matrix Storage driver. Can you try to update it?

Glad to hear that
I don't see the NB sensor value in the HWMonitor screenshot you sent. Where is it?

 

It's the first sensor (tz00) under acpi.

 

Ah, that one That doesn't have to be the NB sensor, it might be placed anywhere..
I currently do not report ACPI temperatures. Unfortunately this is not in the short-term plan...

 

Thanks for all your help Martin!

 

Then I guess HWMonitor also reads HDD temperature less times per minute than HWiNFO32.

The driver is updated to the latest version, 9.6.0.1014... Well, it seems there's no fix for this since is a driver's issue.

 

I don't think there's anything else I could do except lowering the HDD temperature scan frequency.

 

It's ok as it is, thanks anyway. I've looking into HWMonitor and it increases latency at least one time every minute. Not as critical as with HWiNFO32, but I can't still watch a movie while I have the program in the background. So I'll try an older driver or I'll wait for the next one.