Additional monitoring of Clevo notebooks in HWiNFO32

I have added additional monitoring of 2 temperatures for some Clevo models:
- W860CU, W870CU, W880CU
and need feedback how it works..

Please download this Beta:
www.hwinfo.com/beta/hw32_359_938.zip
and submit your results.

Maybe it's possible to include support for other models as well. Just let me know.
I haven't figured out yet if/how to read additional fan RPM monitoring..

 

Here a new Beta with additional support of: M570RU, M570TU, M860TU, M980NU
and latency issue fix:
www.hwinfo.com/beta/hw32_359_941.zip
Needs testing and feedback.

 

Great, will test it on my W860CU once I'm home.

 

Even more improvements to the latency/stuttering issue with EC reading:
www.hwinfo.com/beta/hw32_359_942.zip

 

Any results? I need feedback

 

Yeah, tested it yesterday a bit, have to test it again. But so far the features are great. EC-Temps show up. I had 55 - 61°C max under stress testing, idling at 44°C, much like the CPU. Vcore of the GPU was correct as well. What stuff do you need to know?

I can make some pics later today / tomorrow. But I especially like the feature that shows how much power the CPU draws (but thats not new I think), it gives you a very acurate picture and shows that my i7 820QM will stop Turbo Boosting as soon as it goes above 45W TDP which it does if 4 or more threads are running (but that was known before as well), I just thought with non-extreme CPUs temps are more decisive than TDP.

BTW, about Tubro Boost. The program always shows all cores beeing boosted, so that if I only use 1 worker in prime 95 it seems that all cores are running at 3.06 GHz, while in reality only one core is running at 3,06 GHz and the other cores are not active. Is there a way to reflect this in HWiNFO32 - maybe fade out the cores that are inactive - or am I just wrong about how Turbo Boost works?

 

Thanks for the feedback. Regarding the EC sensor, I'd need a screenshot (best would be under load and idle), so I can verify how to works and if I can release this.
Yes, the CPU power measurement is available for some time already. It's a real measuring (not guessing). I don't know of any other tool that would be capable of this feature.
About the Turbo Boost.. I have to acknowledge that this is a known issue in HWiNFO32. It happens on some CPUs (and depending on certain Turbo mode preferences) that HWiNFO32 reports all cores boosted when they should not be.. This is caused by the fact that the CPU's internal machine for triggering Turbo is very sensitive and it causes that my clock measurement method triggers the boosting. There are other possible measurement methods, but I wouldn't like to go that way. I'm still working on a solution... Anyway, it doesn't occur on all Nehalem/Westmere family CPUs and depends on certain preferences...

 

I'll post pics when I get home in about 8 hours.

 

Here you go:

LOAD

IDLE

Edit: Well this is weird, I updated the bios to disable HT and ran prime95 again and the result was this: EC Temps are totally messed up...

 

Thanks for the feedback. So Temp1 seems to be CPU (maybe via PECI) and Temp2 seems static, maybe in invalid reading..
However, I have no explanation for the BIOS update.. Please give me a link to the new BIOS image, I'll try to look what has changed there... Maybe that modified BIOS is buggy?

 

I ran a few tests yesterday and everything was fine, got the same values. When I got the strange numbers I had HT turned off. Yesterday I tested with HT on. I'll try it once more with HT off, but I think it was just a random error. Anyway, this is the bios I used. From what I understand it's an early version, so it could be buggy.

 

I ran some tests too and the second value sems stuck at 51 degrees, while the first value seems to resemble the CPU temp... Very impressed with the amount of info in the program, great job...

 

I noticed the fixed 2nd temperature too on some models (might be some fixed threshold value), whereas I think on some others it seems to reflect a real temperature. I'll watch this and eventually remove the 2nd reading in case it doesn't provide real data on any machine...

 

Been trying HW32 v3.60-946 after using several newer betas on my Sager 8760 (Clevo W870CU) with the new 1.00.15S/1.01.15S BIOS running a 920XM/Mobility 5870 combo. I have lost the CPU core power feature on all of these newer versions and miss it terribly. Don't know whether it's the new bios or a change in HW32, but would sure love to see it restored.

​

Edit: Same result in v3.60-973 beta.

​

Nevermind. For reasons I cannot explain, the Intel PCH data is back, including CPU Core power draw, in 3.60-973 beta. Same system BIOS. Same everything as before. That was easy.

 

Well, I was going to try this out, but the links in the first post are all dead...

 

You can get it here: HWiNFO32 Download

 

What is the "Clevo EC" thing shown above?

It shows up in my system as well. except it's consistently 30C higher than the CPU digital thermal sensor.

Which one should I trust to give me my "true" CPU temp.

 

Stands for Embedded Controller that is available on laptops to read sensors (temperature, battery status, etc) built into the mobo or some components and accessed through the ACPI system. The problem is that the identification of which sensors are reading what is for some reason not readily available, so connecting a particular sensor readout to a particular component or mobo section is often a matter of trial and error or even raw speculation. That's the sum total (or more) than I know about EC based temps. Because I don't know what they represent, I ignore them completely, and rely upon the upper set of temps, which are the same temps used by other system and temp utilities. I wish they would disappear to remove the clutter, but Mumak is fond of them.

 

The problem that it's not possible to determine where a particular sensor input is connected to is not an issue of EC, but most sensors. There are only few standards where the monitoring system provides the description of the reading. So for every mainboard model ones needs to know the correct assignments and often needs to adjust the values. EC is even more difficult since it's completely proprietary and the layout is fully in hands of designers - there's no standard and much harder to determine how to read any value from it. But that doesn't mean the values read from EC are not reliable.
I started extensive work on EC reporting (and spent several weeks with it) per request of several NBR users, which wanted to know certain sensor values which can be read from the EC only. So for some NBs (i.e. Alienware) I have been able to report several interesting and very usable values which made lots of ppl happy Now, they can watch several other temperatures and fan speeds (i.e. the true GPU fan speed).
If you are keen on DTS values, I suggest to read a bit more details about it which I posted here: HWiNFO and HWiNFO32 Forum • View topic - CPU Core temperature measuring via DTS

 

Thanks, Mumak. I just want the best information available on CPU and GPU temps (and power use), and HW32 is still the best. Is there any way to determine and report the power used by the GPU? That would be interesting.

 

I don't think that's possible for most GPUs. Only few High-end desktop models feature a special Volterra VRM chip that can monitor the current (Amps) consumption and based on the actual voltage one can calculate the power used.

 

Thanks. +Rep for HW32.

 

Well, if there is anything I can do to help further elucidate the Clevo EC by providing data, etc... I'd be happy to do it. I'm sure I can round up a few more new M570RU-U users around here to do the same.

 

I'm still trying to determine how to read fan speed(s) from EC on Clevo. I have an assumption, but don't know if it's the right one. Could you please use the RW-Everything tool and dump the content of EC? Please make dumps for different fan speeds (which you can probably verify only acoustic). i.e. a dump when you think the fan is off, a dump at low speed (load) and dumps for higher-max fan speeds.

 

Which area in RW-Everything would I need to dump? ACPI?

I can most likely do low speed, mid-speed, and flat out high speed, but my fan never really turns off due to my C2E processor.

 

I need the "EC" dump.

 

Do you need just a regular save or a binary save?

 

Regular (.rw) is ok

 

Ah, in RW it is listed as "Embedded Controller"

Here they are, in RW and BIN format:

Hotfile.com: One click file hosting: EC Data.zip

 

Thanks. There seem to be certain values changing that might be belong to the fan. If that's true, it seems there are 2 fans. What component have you stressed in this test CPU or GPU? If CPU, could you please try to load the GPU so its fan runs at max?

 

Yep, there are two fans. I stressed both in the Max Load file.

 

Could you please try to stress only the CPU and then only GPU, so I can be sure?
I still don't know how can I convert the values into RPM..

 

Here are some EC-Dumps I hope you can work with them.

cu
Arne

 

And here are more EC-Dumps...

CPU fan high; GPU fan low and both highest

I had to cover the air intakes to reach over 80°C, normaly I don't do those things ;-)

bye
Arne

 

Thanks. I'll analyse those dumps..

 

I've got a in which both fans are off... (at least not noticeable)

Could is be, that the shown values of the fans are anti proportional to the real fan rpm. --> FanMaximumRPM - FanReading = AcutualFanRPM. The exception is zero (fan off).

Can anybody confirm this behavior?

 

Yep, that's what I already determined. However, in case the fans are spinning I don't know how to convert those values to RPM yet.

For fan control, you could make an experiment: if the fans are spinning, try to write some values there (00 00 for example to stop). Some EC's fan control works this way..

 

I can overwrite the values with 0000, but their are changed back immidiatly without influence on the actual fan speed.

 

That means this kind of fan control doesn't work there It's either another register or auto control (undocumented)...

 

Is there a programme that allows me to log the ec dumps? Some values are changing so quick that you can't see what they are. Or is it possible to trigger on a certain evnet, like changing of a certain value?

 

I don't know any such program. EC-stuff is quite sensitive and messing with it can case severe problems or damage.