P-core temperature of the i9-13980HX processor

[Bogus5]

I bought a few months ago an ASUS laptop with an i9-13980HX processor, in which the processor works strangely, that is, the temperatures of the P cores under load can vary by 15-17°C.
I asked a question about this problem to Intel, but the employees of this company could not give a clear answer, which is completely unremarkable.
I asked a question on the same topic to the Taiwanese ASUS service and they went even further, writing that these are normal states for this processor.
In my opinion, however, the issue is about the bad implementation of liquid metal on the processor, what do you guys think ?

 

[Lewzke]

Modern processors have good power management under idle, so a slight variation of multithreaded load can create power differences on the cores, i think it is normal to have 15C variations accross the cores, especially under games and real world applications. When using synthetic benchmarks and full load, maybe the difference is 5-10C.

"Intel® Turbo Boost Max Technology 3.0 identifies the best performing core(s) on a processor and provides increased performance on those cores through increasing frequency as needed by taking advantage of power and thermal headroom. Intel® Turbo Boost Max Technology 3.0 frequency is the clock frequency of the CPU when running in this mode."
So there are selected cores where the clock is a lot higher, then the temp should folow the laws of physics, it's normal.

 

[Bogus5]

Not very normal, in this processor the first two P-core (P-core 0 and P-core 1), have a multiplier of ×56, so in principle they should work most efficiently, but despite this, P-core 1 has the lowest temperatures of all P-core under load.
Its neighboring P-core 2, on the other hand, has the highest temperatures under load of all P-type cores.
For example, P-core 1 has 80°C, and P-core 2 has 97°C.
Neither of these temperatures threatens the safety of the processor, but the high temperature of the P-core 2 causes premature throttling of the entire chip.

 

[bug]

Not very normal, in this processor the first two P-core (P-core 0 and P-core 1), have a multiplier of ×56, so in principle they should work most efficiently, but despite this, P-core 1 has the lowest temperatures of all P-core under load.
Its neighboring P-core 2, on the other hand, has the highest temperatures under load of all P-type cores.
For example, P-core 1 has 80°C, and P-core 2 has 97°C.
Neither of these temperatures threatens the safety of the processor, but the high temperature of the P-core 2 causes premature throttling of the entire chip.

That's silicon lottery at work right there. Some chips are more homogeneous/predictable, some aren't.

 

[Bogus5]

That's silicon lottery at work right there. Some chips are more homogeneous/predictable, some aren't.

This is the most sensible answer on this subject that I have ever heard or read from anyone.

Indeed, the inhomogeneous structure of the silicon wafer can cause such anomalies, but where is Intel's quality control in all this ?

 

[ThrashZone]

Hi,
Yeah I'd say you're doing pretty good within that amount
Most are lucky if they were within 20c hehe

ASUS bios should show a SP rating for the chip and cooling what does it show for your system ?

 

[Bogus5]

Asus' BIOS only shows a single CPU temperature without breaking it down by individual cores, and it must be said that the fan curves are chosen well enough that the CPU temperature in the BIOS does not exceed 60°C.
After logging into Windows 11, the CPU temperature already depends on the active power plan and at rest in the Silent plan is about 50°C, and in the Turbo plan about 34°C.
So basically the norm.
The fun starts only under heavy load of all CPU cores.

 

[ThrashZone]

Hi,
I'm sure 8 p-core 16 e-cores hehe

At first though the chip had more p-cores hell a normal 8 core 16 thread chip would be within 10c of each other max. hell I get that with 10 cores and 20 threads.

 

[Bogus5]

It turns out that the phenomenon of excessive temperature differences between individual processor cores is quite common in the ,12th, 13th and 14th generation series, and these processors are among the quite difficult to handle under heavy use.
Among other things, this is why programs like ThrottleStop come in handy, but if it were possible to lower the clocking of only the hottest cores, this would be an ideal compromise.

 

[ThrashZone]

It turns out that the phenomenon of excessive temperature differences between individual processor cores is quite common in the ,12th, 13th and 14th generation series, and these processors are among the quite difficult to handle under heavy use.
Among other things, this is why programs like ThrottleStop come in handy, but if it were possible to lower the clocking of only the hottest cores, this would be an ideal compromise.

Hi,
Yep that's always been an issue those series just magnify it more than other series

Just be thankful you were smart enough to get a HX chip and not H alone hehe at least you can play with power limits.

You using hwinfo64 for core temps ?

 

[Bogus5]

Yes, I use HWiNFO to monitor the hardware, as it does quite well, although Kevin claimed that HWiNFO has some friction with ThrottleStop, but I haven't noticed any irregularities.

 

[unclewebb]

Kevin claimed that HWiNFO has some friction with ThrottleStop

The only problem is HWiNFO clears the Limit Reasons data out of the processor as it gathers this information. When HWiNFO is running, it can prevent other programs like ThrottleStop from reporting the Limit Reasons data correctly. You will still see red boxes in ThrottleStop if throttling is in progress. The yellow boxes that indicate that throttling was previously happening will be automatically cleared by HWiNFO. That is the only conflict that I know about.

You using hwinfo64 for core temps ?

ThrottleStop usually does a better job at reporting the maximum core temperature when a CPU core is fully loaded. ThrottleStop runs at a higher priority so it gets preferred access to the CPU temperature data. If you are running both programs together during a stress test and ThrottleStop reports a higher peak temperature, you can trust that the value ThrottleStop is reporting is correct. A full load Cinebench run would be a good test when comparing peak temperatures.

 

[bug]

This is the most sensible answer on this subject that I have ever heard or read from anyone.

Indeed, the inhomogeneous structure of the silicon wafer can cause such anomalies, but where is Intel's quality control in all this ?

Quality control is a balancing game. It's easy to set a maximum of, say, 5C max between all cores. But then you'd be throwing an awful lot of CPUs away. And sell the rest for much, much more. This is what is commonly referred to as "node maturity". As time goes by, manufacturing improves and so does the variance in the silicone. The "problem" is, by the time that happens, customers already demand goodies manufactured using the newer nodes.

 

[unclewebb]

Intel rates their core temperature sensors to be accurate to +/- 5°C. That means you could have two cores running at the exact same temperature with one core reporting 5°C too high and the one beside it reporting 5°C too low. A difference of 10°C between two equally loaded cores might not mean anything. The temperature sensors Intel uses are definitely not as accurate as most people assume.

the temperatures of the P cores under load can vary by 15-17°C

That is beyond the accuracy of the sensors.

 

[Bogus5]

That is beyond the accuracy of the sensors.

Look, I know this, but both Intel and Asus pretend that these are normal processor states.
I pressed Intel's technical department, but they are so unprofessional that I felt sorry for them.
Asus' technicians, on the other hand, are Chinese, but they pretend to be Greek, and it's also quite comical.
The nature of my work does not require loading the CPU for long periods of time, as it is photo editing, so I am unlikely to make a revolution in Taiwan because of Asus' dodges, but let this be a warning to others.

 

[ThrashZone]

The only problem is HWiNFO clears the Limit Reasons data out of the processor as it gathers this information. When HWiNFO is running, it can prevent other programs like ThrottleStop from reporting the Limit Reasons data correctly. You will still see red boxes in ThrottleStop if throttling is in progress. The yellow boxes that indicate that throttling was previously happening will be automatically cleared by HWiNFO. That is the only conflict that I know about.


ThrottleStop usually does a better job at reporting the maximum core temperature when a CPU core is fully loaded. ThrottleStop runs at a higher priority so it gets preferred access to the CPU temperature data. If you are running both programs together during a stress test and ThrottleStop reports a higher peak temperature, you can trust that the value ThrottleStop is reporting is correct. A full load Cinebench run would be a good test when comparing peak temperatures.

Hi,
You show all 24 core temps hehe

 

[bug]

Look, I know this, but both Intel and Asus pretend that these are normal processor states.
I pressed Intel's technical department, but they are so unprofessional that I felt sorry for them.
Asus' technicians, on the other hand, are Chinese, but they pretend to be Greek, and it's also quite comical.
The nature of my work does not require loading the CPU for long periods of time, as it is photo editing, so I am unlikely to make a revolution in Taiwan because of Asus' dodges, but let this be a warning to others.

View attachment 340311

I don't see dodges in there. Maybe you're just looking for trouble where there isn't any?

 

[Bogus5]

I don't see dodges in there. Maybe you're just looking for trouble where there isn't any?

According to you, this is how the processor should work, that the temperature difference between the power cores should be 17°C ?
Or maybe you can answer me why Intel could not answer that such operation of the processor is within their specifications ?
Asus, on the other hand, writes back that there is nothing wrong, that is, how to treat it, like a grim joke or other misfortune ?

You buy a powerful processor for work and get a disabled chip that is constantly throttled under load.
This is the norm according to you, if so, in what country are such standards ?

I suggested to the technicians of Asus, to check the correct application of liquid metal on the CPU, and they, instead of issuing a disposition to check the laptop write off some foolishness.
This is how you see normalcy ?

 

[bug]

According to you, this is how the processor should work, that the temperature difference between the power cores should be 17°C ?
Or maybe you can answer me why Intel could not answer that such operation of the processor is within their specifications ?
Asus, on the other hand, writes back that there is nothing wrong, that is, how to treat it, like a grim joke or other misfortune ?

You buy a powerful processor for work and get a disabled chip that is constantly throttled under load.
This is the norm according to you, if so, in what country are such standards ?

I suggested to the technicians of Asus, to check the correct application of liquid metal on the CPU, and they, instead of issuing a disposition to check the laptop write off some foolishness.
This is how you see normalcy ?

Yes, according to me (and seemingly everybody else), this is normal. Then again, I'm an engineer, I understand a bit about materials, fabs, variation, tolerances...

To put it bluntly, having the same temperature on all cores was not advertised anywhere, ever. It's just something you decided should happen, on your own.

 

[Bogus5]

Yes, according to me (and seemingly everybody else), this is normal. Then again, I'm an engineer, I understand a bit about materials, fabs, variation, tolerances...

To put it bluntly, having the same temperature on all cores was not advertised anywhere, ever. It's just something you decided should happen, on your own.

17°C is not within Intel's specification tolerances, and I don't know where you are bringing up such revelations from.
If you have access to documentation, other than me, on the processor in question, please provide a link to it so that everyone can see the news.

 

[bug]

17°C is not within Intel's specification tolerances, and I don't know where you are bringing up such revelations from.
If you have access to documentation, other than me, on the processor in question, please provide a link to it so that everyone can see the news.

Where did you find a link saying 17C is not within Intel's specs? What specs?

 

[Bogus5]

Where did you find a link saying 17C is not within Intel's specs? What specs?

I'm not the one who says that 17°C difference is the norm, but you, however, if you say that it is the norm, then prove your words.
Otherwise it's just more words without proof.

 

[bug]

I'm not the one who says that 17°C difference is the norm,

You're the one saying it's not the norm, that where this all started. So go ahead, prove it. (Mind you, you haven't even provided proof your reading are that far apart, but let's not get hung up on details.)

but you, however, if you say that it is the norm, then prove your words.
Otherwise it's just more words without proof.

Asus's engineers told you that. I just said between your allegations and their statement, I put my faith in the latter.

 

[Bogus5]

I'm sorry, but there's no point in continuing this conversation.

Re: Asus ROG G614 Stuttering

Anyway, liquid thermal pastes based on silicone have a maximum thermal conductivity of 14 W/mK, so they are not suitable for efficient processors.

rog-forum.asus.com

Something about trust.