Need help understanding HWINFO thermal throttle meanings and where they come from

[solitarymassacre]

As you can see in my screenshot, my Thermal Throttling (HTC) and (PROCHOT EXT) both have a thermal throttle "Yes" applied.

From my understanding, Thermal Throttling (HTC) is "HTC stands for Hardware Thermal Control. If it reports "Yes", it means the CPU is overheating." But I am confused because my die temps never went above 100C which is my max temp. Are the die temps at a lower threshold than the TjMax temps?

I also noticed it says my Thermal Limit percent is at 69.5, but my CPU Tctl/Tdie is only at 64.5, could this mean that if my die temps get too hot a Thermal Throttle (HTC) is thrown?

The other one is the Thermal Throttling (PROCHOT EXT), this is "something else" if this is the only thing flagged. But since I also have HTC flagged, this means it could be the die itself?

I am wondering if CPU die alone can issue a PROCHOT but the CPU cores (PROCHOT CPU) be completely fine. Does that make any sense? I am not super familiar with how all this connects and works.


In a GPU benchmark, you can also see my super inconsistent CPU freqs, my guess is due to the DIE temps,


My question is this - does HTC and PROCHOT EXT get flagged if the die gets too hot? Or is something else throttling my hardware?

Thanks!

 

[Martin]

The maximum reached Thermal Limit was 101 C, so that certainly explains why the throttling occurred.

 

[solitarymassacre]

The maximum reached Thermal Limit was 101 C, so that certainly explains why the throttling occurred.

Thanks for the reply!
Just a bit confused, thermal limit of what exactly? The die? Cause the die max temp was 97C

And the only 101 I see is the "Thermal Limit 101.0%" is that the same as centigrade?
I ask cause I found this:

No, this shows how far from the thermal limit the CPU currently is. So for example if the thermal limit is 100 C and the current temperature is 70 C, it will show 70%.

From another post. Does this mean the thermal limit is less than 100C? Which is further confusing considering the AMD website shows 100C as the TjMax. Unless the BIOS has something less?

Thanks again, just trying to understand my laptop more and how the readouts of HWINFO apply to it

 

[solitarymassacre]

So I did a cinebench R23 test, and I hit the same 101 thermal limit but was not flagged for throttling? I am very confused now

 

[Zach]

According to AMD 7945HX specs this is a default 55W TDP with adjustability up to 75W. This means that actual power consumption (PPT) may well exceed 100W and we see it on those screenshots (~120W)
Given that is a mobile CPU installed on laptops I believe its too aggressively tuned unless there is a massive cooler in there that can dissipate all that heat fast enough.
Do you have any controll in BIOS to adjust CPU parameters, like PPT, or max CPU temperature?

BTW, isn't some of those thermal throttling switches (yes/no) related to VRMs as well?

 

[solitarymassacre]

The maximum reached Thermal Limit was 101 C, so that certainly explains why the throttling occurred.

Thanks for the reply!
Just a bit confused, thermal limit of what exactly? The die? Cause the die max temp was 97C

And the only 101 I see is the "Thermal Limit 101.0%" is that the same as centigrade?
I ask cause I found this:

No, this shows how far from the thermal limit the CPU currently is. So for example if the thermal limit is 100 C and the current temperature is 70 C, it will show 70%.

From another post. Does this mean the thermal limit is less than 100C? Which is further confusing considering the AMD website shows 100C as the TjMax. Unless the BIOS has something less?

Thanks again, just trying to understand my laptop more and how the readouts of HWINFO apply to it

 

[solitarymassacre]

According to AMD 7945HX specs this is a default 55W TDP with adjustability up to 75W. This means that actual power consumption (PPT) may well exceed 100W and we see it on those screenshots (~120W)
Given that is a mobile CPU installed on laptops I believe its too aggressively tuned unless there is a massive cooler in there that can dissipate all that heat fast enough.
Do you have any controll in BIOS to adjust CPU parameters, like PPT, or max CPU temperature?

I would have to check what parameters I have access too. I think most of it is pretty locked out.

BTW, isn't some of those thermal throttling switches (yes/no) related to VRMs as well?

This is what I am trying to figure out. From my reading, if "Thermal Throttling (HTC)" it means the CPU is overheating,

HTC stands for Hardware Thermal Control. If it reports "Yes", it means the CPU is overheating.

which makes sense with how hot the die gets, but the core temps seem to be fine. So I am trying to figure out if the die temps are triggering the throttle switch, or the VRMs (as I did a shunt mod and redid the LM and thermal putty, maybe I didn't do it right).

But after running a CPU stress test, I don't get either switches, but the second I run a graphics test, I get both. So I am beginning to think it may be the VRMs

 

[Zach]

But after running a CPU stress test, I don't get either switches, but the second I run a graphics test, I get both. So I am beginning to think it may be the VRMs

GPU?
Discrete or OnBoard?

Look at the "APU STAPM Limit" between your 2 screenshots.
On 1st ss its over limit (115%) and on 2nd its 90%

Difference between CPU Package power and APU STAPM?

I recently bought a Ryzen laptop, and noticed something peculiar in HWINFO. At high power loads, APU STAPM and CPU Package are nearly identical. Meanwhile when the computer is mostly idle, the floor on STAPM is around 2W, while CPU package can drop as low as 200mW. I tried searching online...

www.hwinfo.com

Also on both SS the "PPT Slow Limit" is right on 100% but "PPT Fast Limit" on 1st SS is 101.2% while on 2nd is 99.2%.
I believe that Fast/Slow limits are different CPU package (PPT) power levels allowed for different time durations.
And of course PPT stands for PackagePowerTracking which includes everything on the CPU(APU) package (Cores, IOD-SoC/iGPU)

 

[Martin]

There are several limiters and STAPM can also cause throttling.
The Thermal Limit shown should apply to CPU Die (average).

 

[solitarymassacre]

GPU?
Discrete or OnBoard?

It is discrete. Its a laptop with 7945HX CPU and an RTX 4090 discrete GPU. The test was done on the discrete GPU

Look at the "APU STAPM Limit" between your 2 screenshots.
On 1st ss its over limit (115%) and on 2nd its 90%

Difference between CPU Package power and APU STAPM?

I recently bought a Ryzen laptop, and noticed something peculiar in HWINFO. At high power loads, APU STAPM and CPU Package are nearly identical. Meanwhile when the computer is mostly idle, the floor on STAPM is around 2W, while CPU package can drop as low as 200mW. I tried searching online...

www.hwinfo.com

Also on both SS the "PPT Slow Limit" is right on 100% but "PPT Fast Limit" on 1st SS is 101.2% while on 2nd is 99.2%.
I believe that Fast/Slow limits are different CPU package (PPT) power levels allowed for different time durations.
And of course PPT stands for PackagePowerTracking which includes everything on the CPU(APU) package (Cores, IOD-SoC/iGPU)

I see! So this points to the throttling being from the CPU and not the VRMs is what I am taking away from this?

There are several limiters and STAPM can also cause throttling.
The Thermal Limit shown should apply to CPU Die (average).

So its not the VRMs you would say? That is good. I do have a dark spot on my CPU die and heatsink. I will clean that up with Flitz and see what happens

Thank you both for the plethora of information. This has been a fun learning experience figuring out what all these amazing data points are. I appreciate your time!

 

[Zach]

Thank you both for the plethora of information. This has been a fun learning experience figuring out what all these amazing data points are. I appreciate your time!

Also you can enable "Show ToolTips" from sensors window settings so when you hover the pointer over a sensor will give you a general description.



Example of tooltip
This (external switch) maybe is related to VRMs

 

[solitarymassacre]

Also you can enable "Show ToolTips" from sensors window settings so when you hover the pointer over a sensor will give you a general description.

View attachment 11299

Example of tooltip
This (external switch) maybe is related to VRMs

View attachment 11300

Thanks! I have known about the tooltips, I was just confused by their description as its quite general.

Like "can be" does not mean "is from" ya know?

Plus I read that if you have PROCHOT EXT and that is the only one, then it points to things like VRMs. But if you have PROCHOT EXT plus any others, it is CPU. This is where I got confused and wasn't sure what it was.

I also plotted a log of HWINFO and noticed that PROCHOT EXT and HTC are ALWAYS triggered together, one is never on while the other is not on.

Black line it HTC, red line is EXT. As you can see, they are on top of one another.

Does this mean it is NOT from the VRMs? This is where I am confused

 

[solitarymassacre]

Now I got the two throttles again but CPU die was not high