Thermal throttling in Ring column?

[AOne]

Hello people (and Glen )
It's been a while I was happily using my undervolted laptop, but I bought a new one with 12000h and 3070ti. Of course it has locked offsets in FIVR and runs crazy hot, so I'm trying changing paste and power limiting, but it's the first time I'm seeing THERMAL flashing in yellow in Rings column. Is this the VRAM that overheating too?
Beside PROCHOT there's also HOT red warning under the table. What's the difference between them?
BR

 

[unclewebb]

it has locked offsets in FIVR

Intel was being Intel when they decided to remove undervolting from their 12th Gen H series. They can play the game that they did this for security reasons but if that argument is true, why is undervolting available on the 12900HK mobile CPU?

In theory, PROCHOT in the Limit Reasons window and HOT on the main ThrottleStop screen should light up at the same time. For some unknown reason, this information is available in two separate registers within the CPU so I decided to monitor both of them separately. I believe ThrottleStop monitors this info four times per second in the Limit Reasons window and only once per second on the main screen. That means you are more likely to see PROCHOT light up red in Limit Reasons compared to seeing HOT light up red on the main screen.

During a thermal throttling event, the CPU is rapidly adjusting the CPU speed hundreds of times per second to keep the peak core temperature just a hair under 100°C. When you are just on the edge of throttling, it is possible for Limit Reasons or other monitoring software to miss any brief throttling. You might see a yellow box in Limit Reasons appear without ever seeing a red box first.

If you see PROCHOT 100°C light up red on the main screen, this confirms that the CPU reached the thermal throttling temperature even if this only happened for a millisecond. I would trust this data more than anything else to confirm throttling. The CPU will set this flag within the CPU whether ThrottleStop is running or not. If you start ThrottleStop after a thermal throttling episode, ThrottleStop will immediately show PROCHOT 100°C in red. Some people have noticed this when first booting up. Great thermal design when a laptop cannot even boot up without overheating.

After the CPU cools down a little, you can click on PROCHOT 100°C to reset this flag within the CPU.

Post some screenshots and post a log file after you do some gaming. Some thermal throttling is OK. Some users barely notice any decrease in performance as long as the CPU temperature does not get too out of hand.

 

[AOne]

What's mostly bothering me (beside all looks like a discotheque) is the Thermal under RING. Any hit what it should point to?

 

[unclewebb]

On some CPUs, THERMAL lighting up red under the CORE and RING columns in Limit Reasons happens at the exact same time. It is likely that both temperature sensors are extremely close to each other. Are you seeing anything unusual like only RING thermal throttling lighting up red while the CORE is not showing THERMAL in red? If you see this happening, what does ThrottleStop report for the peak core temperature?

Intel's temperature sensors are only accurate to +/- 5°C. Some slight temperature difference between two separate sensors is nothing to be concerned about.

If you see anything that looks unusual, post a screenshot of ThrottleStop with Limit Reasons open while throttling is in progress.

The real problem is the 12900H is way too much CPU for the majority of laptops. You get to choose thermal throttling or power limit throttling. You might have to move to the North Pole to fully use what this processor is capable of.

 

[AOne]

Usually most of the time the Core only lights up, and the one on Ring is poping rarely. What puzzles me the most is the temps are staying at approximately 50-60 C and suddenly PROCHOT lights up whith no record for such limits reached in the table for max temp. I've set PL1/Pl2 at 80/90. All this happens in a fraction of a second and then all the temps are back at 50-60 C range, which sound impossible to me in physical way. Either the CPU is totally crazy or the readings are wrong. Under load it hits thermals very fast and stays up there for at least until the PL1 kicks in. The throttling cores vary. Some stay in the hight 70s while the other are at 96-100 and they can change places, which is another impossible thing.
When performing a Cinebench 23 test it scores even more then average, although the Limit reasons box is shining like a Christmas tree. I guess my thermal paste is wrongly re-applied by me. I'm waiting for a paste delivery from Amazon and will see if it improves. I need to decide untill Wednesday if I'll keep or send it back even if the second option would cost me 150 euros and possible problem, because the paste was replaced (if they notice). I bought it for a very good price. Now this laptop is missing on the shelves or way more expensive and I would need 400 euro more if want the CPU with unlocked voltages (12950hx), which is quite expensive already, going beyond 3500 euro limit (this one was only 2500). This whole thing is driving me crazy...

 

[unclewebb]

Usually most of the time the Core only lights up, and the one on Ring is poping rarely.

That sounds completely normal. The cores are creating the majority of the heat. Most of the heat being measured at the ring temperature sensor is likely heat being generated by the loaded cores.

All this happens in a fraction of a second

The monitoring table is only monitoring the CPU core temperatures once per second. If you use the More Data option, you might be able to record a slightly higher peak temperature but does that really matter? If Limit Reasons is lighting up red or if the main screen where it shows PROCHOT 95°C on your computer is turning red then your CPU is thermal throttling. Performance is being reduced, maybe only for a few milliseconds at a time but it is definitely being reduced. You need better cooling for maximum performance. Without better cooling, your CPU will be constantly bouncing off the thermal throttle limit. You can try reducing the power limits to prevent this from happening but either way, performance will be reduced.

When you apply a huge amount of energy to a tiny CPU core that is poorly cooled, you should not be surprised when its temperature instantly changes by 30°C or more. The moment that load gets transferred to a different core, there is going to be a drastic change in core temperatures between those two cores. When a CPU is lightly loaded, this load transferring can be happening constantly. It may look crazy or unbelievable but I think it is the reality. Smaller and smaller CPUs are a good thing but this is leaving less surface area to transfer the heat away. Intel and AMD are both hitting the same thermal wall. Similar 12th Gen desktop CPUs with AIO water cooling are struggling to keep core temperatures under 100°C. Most laptops just do not have the cooling capacity necessary for the amount of heat energy these processors can produce. The best thermal paste in the world, even if it is applied perfectly, is not going to solve the problem if the heatsink is inadequate.

 

[AOne]

Thanks a lot. I'll try with two more pastes in the coming days and will report the results. When I'm sure the TIM is not the main culprit, I'll add some Log files.