Overheating/undervolt/setup issues

[ebaum]

I'm new to throttlestop, but not new to PC's and installed it because my 11800h is overheating and throttling. Not to mention, when it hits 70-80c OGH severely reduces performance. Well, from what I've seen thus far. Can someone PLEASE help me configure throttlestop, if possible? I don't feel comfortable changing the PL values etc. without much knowledge. I have been reading up as much as I can, so I can learn.

I've also cleaned the laptop and reapplied thermal.

PC- Omen 16in 165hz, i7-11800h, rtx 3070, 32gb ram, 6 fan USB cooling stand

Hope I posted this in the right place? If not, I'm sorry.
Thanks in advance!

**Update** So... I tried undervolting! Found a stable offset @ -70. Anything above that gives BSOD. Is there anything else I can do? I want to get back to playing overwatch and tarkov ;(

Any help or direction would be GREATLY appreciated!

 

[unclewebb]

The log file shows your laptop is running hot and thermal throttling. Was the log file run before or after you set an undervolt? Did you use the same negative offset values for both the core and the cache?

What thermal paste did you use? Some popular thermal pastes that work well on desktop CPUs with their integrated heat spreaders can quickly fail when applied direct die to a mobile CPU. Many users have reported good results when using Honeywell PTM 7950. Many traditional pastes are not happy long term when CPU temperatures are constantly over 95°C.

At the moment your laptop starts overheating when power consumption is over 80W. Without better cooling, you might want to try reducing the long term PL1 power limit to about 70W or 75W. I would set PL2 to about 80W and I would reduce the turbo time limit to about 8 seconds. This will reduce peak performance. Overall performance might be better if this helps you avoid constant thermal throttling. Not much you can do when a manufacturer puts too much CPU in their laptops but forgets to engineer proper cooling.

 

[ebaum]

I set the offset to 70 on both and ran the test again, so I could be sure. I uploaded the new log. I got a 'power' and 'EDP other' warnings, which I posted SS's of both, but I'm assuming what you told me will resolve this?

reducing the long term PL1 power limit to about 70W or 75W. I would set PL2 to about 80W

-Do I need to clamp either? Or the lock function?

Something I'm confused about is the MMIO and MSR values on PL1 that I just noticed are fluctuating. It went from 90/90-PL1 to 55/109-PL1 (posted SS's). Is this normal? I have not messed with those values yet. I was reading someone say that you have to sync them or TS uses whichever is lowest? Everything else is synced but the 55 MSR on PL1. I'm probably overthinking it haha.

I used Honeywell PTM 7950 and applied it yesterday.

Is it okay to use omen gaming hub for the performance feature or should I just leave it on balanced? I mostly use it for the lights and fans.

Also, should I set speed shift EPP to 0 like everyone is saying or should I leave it at 10?



I really appreciate the help!!! Thanks again!


**Update**Used the settings you suggested, but PL1 is changing from 75 to 90 on the TPL screen, got warning for PL1, temps went up to 95c, but did not seem to throttle since I was still sitting at 4.1 till the end. The temps tho.... Whew. I posted the log.

 

[PHVM_BR]

According to the your log, your system limits power to 80W during the benchmark.

At the beginning of the test, the CPU may reach its maximum clock at full load (4.2GHz on all cores) for some time and then drop slightly (~4.15GHz) because 80W is no longer enough to maintain the maximum clock speed due to the increase in consumption caused by the increase in temperature as the test progresses.

PL1 lights up as it is reached and remains at its limit throughout the test.

Despite the high temperature (95/96°C) your CPU is throttling (slightly) due to the 80W limit of PL1.

Although it would certainly suffer thermal throttling if the power limit was higher.

Perhaps if you get a better undervolt setting you will be able to reduce consumption a little more (thus reducing the temperature a little) and your CPU will be able to complete the test at its maximum.
There's very little left.

Try reducing the core voltage further while maintaining the cache voltage. For example -70mV in the cache and -110mV in the core.

Are you using something to move the air intakes away from the table top?
If not, start using it.
A simple laptop stand is sufficient.

 

[ebaum]

According to the your log, your system limits power to 80W during the benchmark.

At the beginning of the test, the CPU may reach its maximum clock at full load (4.2GHz on all cores) for some time and then drop slightly (~4.15GHz) because 80W is no longer enough to maintain the maximum clock speed due to the increase in consumption caused by the increase in temperature as the test progresses.

PL1 lights up as it is reached and remains at its limit throughout the test.

Despite the high temperature (95/96°C) your CPU is throttling (slightly) due to the 80W limit of PL1.

Although it would certainly suffer thermal throttling if the power limit was higher.

Perhaps if you get a better undervolt setting you will be able to reduce consumption a little more (thus reducing the temperature a little) and your CPU will be able to complete the test at its maximum.
There's very little left.

Try reducing the core voltage further while maintaining the cache voltage. For example -70mV in the cache and -110mV in the core.

Are you using something to move the air intakes away from the table top?
If not, start using it.
A simple laptop stand is sufficient.

Yeah, uncle told me to set PL1 to 70-75W, but I was still having thermal/throttle issues. PL1 by default is 55... Does that make any difference? And there is MMIO values, that I heard override MSR values if the MMIO values are lower. Is this true?

Perhaps if you get a better undervolt setting you will be able to reduce consumption a little more

-Makes sense!

For example -70mV in the cache and -110mV in the core.

-Anything under -80mV in core results in BSOD, but I'll give another go. I was told to always sync cache and core and not to input different values (e.g. -70 core / -70 cache)? Not saying I know what I'm talking about, simply going off what I was told and trying to triple check to avoid damaging anything, haha!

Are you using something to move the air intakes away from the table top?

-Yes, I have a 6 fan USB laptop stand and opened my laptop to redo thermal paste. It's a new laptop with only 6 months worth of miles, but I probably waited too long to reapply, lol.


Appreciate the help and thank you for taking the time to explain all of this to me. I'm really trying to learn as much as I can.

 

[Maxstate]

The log file shows your laptop is running hot and thermal throttling. Was the log file run before or after you set an undervolt? Did you use the same negative offset values for both the core and the cache?

What thermal paste did you use? Some popular thermal pastes that work well on desktop CPUs with their integrated heat spreaders can quickly fail when applied direct die to a mobile CPU. Many users have reported good results when using Honeywell PTM 7950. Many traditional pastes are not happy long term when CPU temperatures are constantly over 95°C.

At the moment your laptop starts overheating when power consumption is over 80W. Without better cooling, you might want to try reducing the long term PL1 power limit to about 70W or 75W. I would set PL2 to about 80W and I would reduce the turbo time limit to about 8 seconds. This will reduce peak performance. Overall performance might be better if this helps you avoid constant thermal throttling. Not much you can do when a manufacturer puts too much CPU in their laptops but forgets to engineer proper cooling.

This isn't related to the thread and apologies for bumping it that way, but I couldn't send you a PM about this. Get an error message.

Just wanted to thank you for your work and legendary support. King.

 

[PHVM_BR]

Yeah, uncle told me to set PL1 to 70-75W, but I was still having thermal/throttle issues. PL1 by default is 55... Does that make any difference? And there is MMIO values, that I heard override MSR values if the MMIO values are lower. Is this true?


-Makes sense!


-Anything under -80mV in core results in BSOD, but I'll give another go. I was told to always sync cache and core and not to input different values (e.g. -70 core / -70 cache)? Not saying I know what I'm talking about, simply going off what I was told and trying to triple check to avoid damaging anything, haha!


-Yes, I have a 6 fan USB laptop stand and opened my laptop to redo thermal paste. It's a new laptop with only 6 months worth of miles, but I probably waited too long to reapply, lol.


Appreciate the help and thank you for taking the time to explain all of this to me. I'm really trying to learn as much as I can.

The MMIO limit is a dynamic limit, which varies with load, temperature and duration according to the manufacturer's specific setting for your specific laptop model.

As soon as it is lower than the MSR limit, it will dictate the limit.

There are two distinct situations in the behavior of the MMIO limit on a laptop.
First in CPU-only loads, where the power limit can remain higher.
Second in CPU + GPU cross loads (games for example), where the CPU power limit needs to be lower for the cooling system to support the consumption of both.

Your system probably limits the CPU to 80W on CPU-only loads for at least a certain period of time (this limit may decrease if the load lasts longer).

In cross loads, the CPU limit is probably 45 or 55W in turbo mode.

Checking the Lock MMIO box can lock the MMIO limit, if your system allows it, ensuring that performance remains stable at the desired level according to the limits you want and that the cooling system supports.

Regarding undervolt, some loads benefit from more undervoltage on the cores (Cinebench for example) and as undervoltage on the cache is the biggest limiting factor, you can try lowering the voltage on the cores further and see if you get any gains.

 

[unclewebb]

uncle told me to set PL1 to 70-75W, but I was still having thermal/throttle issues. PL1 by default is 55...

If 70W is too much for your cooling then you will need to keep lowering your power limits. Most recent laptops have the same problem. Inadequate cooling. At full power they will overheat. You get to choose thermal throttling or you can reduce the turbo power limits and end up with power limit throttling instead. I do not know which type of throttling is going to be better.

I always recommend checking the MMIO Lock box. This eliminates the secondary MMIO turbo power limits. They are not necessary. I find it is easier to only adjust the MSR turbo power limits so you do not have to fiddle with two separate and overlapping sets of turbo power limits.

I was told to always sync cache and core and not to input different values

Do not believe everything you hear on the internet. All those YouTube experts have never asked my opinion. Syncing the core and cache offset voltages might work for some CPUs or some applications but it might not be best for other CPUs. Keep an open mind and do lots of testing. Many users over the years have achieved their best overall results by setting the core to a bigger offset value compared to the cache. Use a consistent benchmark test like Cinebench. When you are living life on the edge of thermal and power limit throttling, it may be difficult to get consistent results to prove if using different voltages is a benefit or not. Look at both performance and temperatures. If you cannot prove that using different voltages is best then I would leave them synced.

Is it okay to use omen gaming hub

It is OK to use ThrottleStop with other similar programs. You just have to be careful that two different programs are not writing different power limit or other values to the same CPU registers. If you want ThrottleStop to be in control of your power limits then you might need to use the ThrottleStop Lock option in its Power Limit Controls section. Once the power limit register is locked, other software like Omen Gaming Hub will be blocked from making any changes to the turbo power limits. You will need to reboot or do a sleep resume cycle to unlock this register. When ThrottleStop is restarted or if it is left running, ThrottleStop will immediately lock the turbo power limits again to the values that you have requested in ThrottleStop.

Just wanted to thank you

You are welcome. You are correct that your post is off topic but I will let it slide this time.
I always like hearing from happy ThrottleStop users. Cheers.

 

[ebaum]

If 70W is too much for your cooling then you will need to keep lowering your power limits. Most recent laptops have the same problem. Inadequate cooling. At full power they will overheat. You get to choose thermal throttling or you can reduce the turbo power limits and end up with power limit throttling instead. I do not know which type of throttling is going to be better.

I always recommend checking the MMIO Lock box. This eliminates the secondary MMIO turbo power limits. They are not necessary. I find it is easier to only adjust the MSR turbo power limits so you do not have to fiddle with two separate and overlapping sets of turbo power limits.


Do not believe everything you hear on the internet. All those YouTube experts have never asked my opinion. Syncing the core and cache offset voltages might work for some CPUs or some applications but it might not be best for other CPUs. Keep an open mind and do lots of testing. Many users over the years have achieved their best overall results by setting the core to a bigger offset value compared to the cache. Use a consistent benchmark test like Cinebench. When you are living life on the edge of thermal and power limit throttling, it may be difficult to get consistent results to prove if using different voltages is a benefit or not. Look at both performance and temperatures. If you cannot prove that using different voltages is best then I would leave them synced.


It is OK to use ThrottleStop with other similar programs. You just have to be careful that two different programs are not writing different power limit or other values to the same CPU registers. If you want ThrottleStop to be in control of your power limits then you might need to use the ThrottleStop Lock option in its Power Limit Controls section. Once the power limit register is locked, other software like Omen Gaming Hub will be blocked from making any changes to the turbo power limits. You will need to reboot or do a sleep resume cycle to unlock this register. When ThrottleStop is restarted or if it is left running, ThrottleStop will immediately lock the turbo power limits again to the values that you have requested in ThrottleStop.


You are welcome. You are correct that your post is off topic but I will let it slide this time.
I always like hearing from happy ThrottleStop users. Cheers.

One more question. Should I mess with iccMAX or power limit 4 values? Or are they irrelevant to my situation?

Thank you! I now know why I've heard about you so much on all the forums! I really appreciate you taking the time to help me!
I finally got it stable and though it sucks to sacrifice performance, it's better than being throttled constantly! Beyond irritated, that I spent all this money on this laptop
only to have to lower stats for it to be 'reliable'.

Anyways,
Thanks again!

 

[unclewebb]

I always recommend setting IccMax to the max for both the core and the cache. I also set Power Limit 4 to a value of 0. There is no legit reason for Intel to have created so many separate throttling schemes. It kept some engineering teams busy but that is about it.

I really appreciate you taking the time to help me!

You are welcome. I always like helping users get the most out of their laptops. It keeps me busy. Most laptop manufacturers like to over promise and under deliver, especially when it comes to cooling. A few simple ThrottleStop tweaks can make a big difference.

 

[ebaum]

I always recommend setting IccMax to the max for both the core and the cache. I also set Power Limit 4 to a value of 0. There is no legit reason for Intel to have created so many separate throttling schemes. It kept some engineering teams busy but that is about it.


You are welcome. I always like helping users get the most out of their laptops. It keeps me busy. Most laptop manufacturers like to over promise and under deliver, especially when it comes to cooling. A few simple ThrottleStop tweaks can make a big difference.

So I did everything and finally got it stable! The only problem now is I'm getting a PL1 error because it's fluctuating from 55W to whatever I have it set to and I can see it doing it in TPL when stress testing. It's most def. omen gaming hub causing this. If I set profile to balanced in OGH, I get 55W in TPL and when I set it to performance in OGH, I get 90W. Is there anything I can do to prevent OGH from interfering with TS values, aside from uninstalling it? I use it for the fans, lights etc.

Making a big difference is an understatement and I'm sure this keeps you busy

Thanks again, boss! o7

 

[unclewebb]

I'm getting a PL1 error

If you set PL1 to 55W in the TPL window then that is going to cause power limit throttling. Boxes lighting up red in Limit Reasons are not errors. Limit Reasons is just telling you the reason why your computer is not running at full speed.

Is there anyway to stop this?

Did you check the Lock options in ThrottleStop. Post an updated screenshot of the TPL window. The power MSR and MMIO power limits should not be changing if you have locked them like I suggested.

By setting PL1 to the default 55W value, your CPU will not be able to run fully loaded at full speed but it will run cooler. I thought that was your goal. A computer that is not constantly overheating and thermal throttling.

Your laptop has inadequate cooling. You have to be willing to compromise. Do you want overheating or do you want power limit throttling?

when stress testing

How about less stress testing and more just playing games. Barely adequate cooling is a problem for almost every recent laptop. Manufacturers are not building laptops so users can run full load stress tests at a reasonable temperature. Save your full load torture tests for desktop computers that have great cooling.