• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

9750h throttlestop undervolt/overclock

Durrendal

New Member
Joined
Jan 22, 2023
Messages
7 (0.01/day)
Hey guys been messing around to try and get the best performance/thermals with throttlestop, i have followed some guides on the forum but i wanted to make sure i wasn't leaving anything behind, my goal isn't so much to have the cooler possible thermals, though that's a nice benefit, but to maximize the performance of the cpu and to stop it from throttling when playing games, to this extend i have repasted both cpu and gpu(2080 max q} and applied the settings below for throttlestop, i also used dragon center to moderately overclock the system, everything seems stable but i was wondering if there was anything else i could do to squeeze a bit more performances.

Thank you so much for your input =)




1674374685214.png
1674374738112.png

1674374756707.png
1674374775928.png
 
I like setting Power Limit 4 to a value of 0 in the TPL window and I set IccMax to the maximum, 255.75, for both the core and the cache in the FIVR window. This can help with some EDP throttling issues.

I usually do not bother with undervolting the Intel GPU. The Intel GPU is not significantly loaded. The Nvidia GPU does all of the heavy lifting. If you do think that undervolting the Intel GPU is necessary then you usually need to also undervolt the iGPU Unslice by a similar or the same amount. Your CPU undervolt looks fine so there is not much else that you can do.

Turn on the Log File option and go play a game for 15 minutes or so. Attach a log file to your next post. Looking at a log will show me if there are any issues during normal use. You can turn on Nvidia GPU monitoring in the Options window so this data will be included in the log file.
 
ok thank you so much for the tips and quick reply =) i will do that as tomorrow and give you the logs, if i want to have another profile for when i am on battery focusing exclusively at maximising battery life at the cost of performance what settings would i change in the current ones for that purpose?

again thank you for your time
 
My old laptop has spent most of its life plugged in. I have not done any extensive battery testing.

Some users like to disable turbo boost when running on battery power. You can also use ThrottleStop to switch to the Windows Power Saver power plan.

1674429299615.png


I am not sure if either of these tricks will make a significant difference. Intel CPUs are pretty efficient as is.

I think the most important thing you can do is get rid of any useless background tasks that prevent your CPU from being idle.
Cores that spend 99% of their idle time in the low power core C7 state will save power.

1674429476890.png
 
here is the log, please tell me if anything is wrong/ needs to be improved, temps while gaming seemed a lot better high 80s low 90s, i believe that is to be expected for a gaming laptop right?

i used your tips for the battery life and it did improve it ever so slightly as i went from 6h to 6h30mn, ill take that =)

thank you for your time,
 

Attachments

It looks like your laptop is running OK. Only one time did your CPU reach 95°C which triggered a tiny amount of throttling for less than a second. This is completely normal for any recent gaming laptop. They all seem to be designed so they run constantly at 90°C or a little beyond.

The one thing I did see in the log file is constant POWER STATUS CHANGE messages every 5 seconds. Is there a feature in Dragon Center that only charges your battery to 70%? Some software like this can cause these constant POWER messages. If you have the option, I would let the battery charge to 100%. I am not sure if that will make any difference. Dragon Center or some other software might still be trying to micro manage your battery charge level regardless.

If your games play smoothly then don't worry about these messages.
 
ok great thank you thats good news, i have the dragon center battery set to charge to 100 percent however that is a separate issue but my battery only charges to 91 percent, im guessing that somehow both those are linked but im not sure how or what to do.

also is there any benefits in raising the pl1 and pl2 or not and what would be the best values according to you?
 
Last edited:
what would be the best values
There is no way for me to know what power limit values would be best for your laptop. Do some testing if you want to find out what works best for your laptop.

I have a desktop computer so I set both power limits to 300 so they never interfere with maximum performance. Lower your power limits if you want to reduce power consumption and heat output. I have heard that some games run quite well with a cooler but slightly slower CPU. Check both Clamp boxes in the Power Limit Controls section when experimenting. This is the best way to enforce both power limits.
 
after playing a few hours with friends of mine, i do get temps around 91/92 average which makes me think i might have done a bad job repasting the cpu, as the gpu hovers around 75, tell me if i am wrong and that temps seems in line for you but with both undervolting and repasting i was hoping to get 85 max whilst gaming for extended periods of time
 
Any temperature under 100°C is a safe operating temperature for an Intel CPU. That is why Intel sets the default thermal throttling temperature to 100°C. They want their CPUs to be safe. 75°C is safe, 85°C is safe and so is 92°C. They are all equally safe.

The previous log file you posted shows the CPU is reaching over 90°C when power consumption is only 25W. What thermal paste did you use? Some popular thermal pastes only last a few weeks before temperatures start to go up and up. If you think you did a bad job then do it again or try using a different thermal paste. Noctua NT-H2 works well for many laptop owners. Some users have reported long term problems with MX-4 when used on mobile CPUs. Did you spread the paste so it covers the entire CPU die? The pea size blob method works well on desktop CPUs that use an integrated heat spreader but this same method might not work so well on mobile CPUs.
 
mx-4 pea size, i suspect i might have not tightened the screws enough but they felt very low quality and i was afraid i wouldn't be able to take them off if i damaged the threads, might have to retry with noctua then as even though i know it is safe i still feel like everything would run smoother if it was a few degrees cooler :x

also its a stupid design where you have to take everything apart to get to the cpu, hate it
 
mx-4 pea size
That was just my lucky guess.

Tiny and low quality laptop heatsink screws is a common problem. Better designed screws would allow more contact pressure between the heatsink and CPU. This would help reduce temperatures.

If you ever get around to fixing this problem, let me know if it makes any difference.
 
will do, thanks for the help, got lucky and bought this laptop new for 900e tax free in japan so cant complain when im back in europe in april ill try all this and keep you posted

again tyvm for all the tips, as long as it manages to run hogwarts legacy without crashing ill be happy until then, if not ill be back here soon enough for some more knowledge :p
 
I like setting Power Limit 4 to a value of 0 in the TPL window and I set IccMax to the maximum, 255.75, for both the core and the cache in the FIVR window. This can help with some EDP throttling issues.
Do you recommend to ALWAYS change PL4 to 0 and IccMax to 255.75? What is the meaning of setting IccMax to the maximum, I mean what does it do?
 
I like setting Power Limit 4 to a value of 0 in the TPL window and I set IccMax to the maximum, 255.75, for both the core and the cache in the FIVR window. This can help with some EDP throttling issues.

I usually do not bother with undervolting the Intel GPU. The Intel GPU is not significantly loaded. The Nvidia GPU does all of the heavy lifting. If you do think that undervolting the Intel GPU is necessary then you usually need to also undervolt the iGPU Unslice by a similar or the same amount. Your CPU undervolt looks fine so there is not much else that you can do.

Turn on the Log File option and go play a game for 15 minutes or so. Attach a log file to your next post. Looking at a log will show me if there are any issues during normal use. You can turn on Nvidia GPU monitoring in the Options window so this data will be included in the log file.
The power setting displayed in above screenshots are perfectly fine for his CPU. Have an identical model (also paired with MSI laptop) - and don't get any throttling whatsoever - with same values. I tweaked it and played with most settings and only triggered EDP throttling - while lowering PL1 & PL2 below 45. Again, those iccMax values and same goes for Power Limit 4 are accurately tweaked for an i7-9750H (MSI followed Intel's power design for this CPU to a T - so it's perfectly tuned with those values).

His throttling issues are actually heat related. He does have BD Prochot enabled. Tho, if his BIOS settings are anything like mine - he probably has another thermal trotting feature activated - as in - CPU TCC offset. As i notice even on other MSI laptops - it's usually set to 5 - which means... if the CPU gets close to 95*C it will trigger thermal throttling quite aggressively. But it works both ways - as in - it won't throttle till the temperatures drops considerably - but try to calibrate the frequencies till it runs consistently at 95*C. Thus, the performance hit is not that impactful - as it is with other throttling features - where the CPU has to cool down for it to run at high frequencies again (same goes for power throttling features). This feature was even promoted for certain Alienware laptops - where some would even recommend to set it to 15 - as in - 85*C being set as the throttle trigger. As it happened in the following demonstration:


Tho, if the game in question is so demanding - that a system running at its MAX frequencies can barely run it at 23 FPS, then... yeah - i guess the impact could be noticeable (if it runs at 16 FPS with TCC Offset triggered). Then again - if that's truly the case... i think it's more reasonable to conclude - that the game in question simply needs more capable hardware (especially if it's a laptop). ^^
 
The power setting displayed in above screenshots are perfectly fine for his CPU. Have an identical model (also paired with MSI laptop) - and don't get any throttling whatsoever - with same values. I tweaked it and played with most settings and only triggered EDP throttling - while lowering PL1 & PL2 below 45. Again, those iccMax values and same goes for Power Limit 4 are accurately tweaked for an i7-9750H (MSI followed Intel's power design for this CPU to a T - so it's perfectly tuned with those values).

His throttling issues are actually heat related. He does have BD Prochot enabled. Tho, if his BIOS settings are anything like mine - he probably has another thermal trotting feature activated - as in - CPU TCC offset. As i notice even on other MSI laptops - it's usually set to 5 - which means... if the CPU gets close to 95*C it will trigger thermal throttling quite aggressively. But it works both ways - as in - it won't throttle till the temperatures drops considerably - but try to calibrate the frequencies till it runs consistently at 95*C. Thus, the performance hit is not that impactful - as it is with other throttling features - where the CPU has to cool down for it to run at high frequencies again (same goes for power throttling features). This feature was even promoted for certain Alienware laptops - where some would even recommend to set it to 15 - as in - 85*C being set as the throttle trigger. As it happened in the following demonstration:


Tho, if the game in question is so demanding - that a system running at its MAX frequencies can barely run it at 23 FPS, then... yeah - i guess the impact could be noticeable (if it runs at 16 FPS with TCC Offset triggered). Then again - if that's truly the case... i think it's more reasonable to conclude - that the game in question simply needs more capable hardware (especially if it's a laptop). ^^
So you say that there's no need to change IccMax to 255.75 and PL4 to 0?
 
So you say that there's no need to change IccMax to 255.75 and PL4 to 0?

For your specific CPU (i7-9750h), nah. Won't make any difference. Here, from Intel's official datasheet:

2023-02-05_004505.png



Above image is cropped (there's more that just those 3 CPU segments) - yet, as can be noticed - even there you have the S-Processor line- which runs at 240 A - so for that given model would make sense.That CPU line also needs 125W (compared to an i7-9750H - which needs only 45W).
 
Last edited by a moderator:
For your specific CPU (i7-9750h), nah. Won't make any difference. Here, from Intel's official datasheet:

View attachment 282324


Above image is cropped (there's more that just those 3 CPU segments) - yet, as can be noticed - even there you have the S-Processor line- which runs at 240 A - so for that given model would make sense.That CPU line also needs 125W (compared to an i7-9750H - which needs only 45W).
I was trying to understand what PL4 and IccMax are, because I can't really understand the meaning behind these settings
 
I was trying to understand what PL4 and IccMax are, because I can't really understand the meaning behind these settings
Those are power limiters for amperage (the strength of the electrical current - which is measured in amperes) - while PL1 and PL2 are power limiters for wattage (which defines the amount of electrical current / where PL1 and PL2 are acting as limiters - limiting the maximum wattage). Both are correlated (work hand in hand), but mainstream wise you only hear about wattage - since that's what's mainly relevant for everyday consumers. That being said, IccMax - shows the Maximum amperage the chip will pull. Tho, it's not a value set in stone - as in - even if you define a specific value in TS (a lower value - that is) - it can still change automatically (if the value proved to be lowered than needed). Yet, if set it to a higher value - like 255.75 amperes (which is basically the maximum) - obviously it can't go any higher than that - so it stays the same (doesn't seem to lower this value automatically - if it proved to be to high).

PL4 is yet another amperage limiter - but this one is more aggressive (locks that given value) - as in - it doesn't change automatically. Thus, if you set it to 50 amperes - even if it proves to be less than required - it won't change. Your system will simply run baldly (handicapped) - while using demanding tasks. If set to 0 - you basically disable this limiter - and your chip will always use the maximum predefined amperage. Anyway, as can be seen in Intel's datasheet (from the screenshot i posted above) - 160 Amperes is the maximum recommended value by Intel for this specific CPU (i7-9750h) - while MSI went with 163 amperes (as can be seen in your screenshot) - supposedly to cover for that slight OC beyond the default values determined by Intel for i7-9750H (for which 160 amperes proved to be more than enough - even while taking into account Turbo Boost). Even an IccMax of 140 amperes can be enough (most of the time) - but hey - since you're determined to push the CPU beyond the default values - you can raise it to 160 amperes (or even 165 - both IccMax & PL 4). If needed - that is. The thing is - the amperage usually increases (automatically) - while increasing voltage - but most (myself included) - use TS for decreasing voltage (under-volt). :)

uncleweeb - usually recommends the maximum - cause that value can cover/work with any given CPU (you don't have to know the maximum values - defined by Intel for any given model - if the amperage is set to the maximum value allowed). Which was particularly the case for Desktops or even older laptop CPUs (more power hungry compared to latest generations) - for many years since TS was released. Yet, 9th gen CPUs meant for laptops - are obviously breaking that pattern in terms of power needs (same can't be said about Desktop CPUs - ofc). Thus, if sett to max value - the power limiter won't get triggered - obviously - tho, the max value (255.75 amperes) is simply not necessary. To check if it's necessary - open TS and click on "Limits (this shows if any limiter was triggered - be it heat or power related)" - clear the shown results (if you never touched/cleared this feature - it will display any trigger picked-up by TS up-to this point). Then run a benchmark tool and/or a demanding app/game - and check it again to see if any limiter got triggered. Tho, as mentioned in my previous post - you might have CPU TCC offset activated as well. If active (and set to 5 by MSI) - even under heavy stress - your CPU should hover around 95*C. If it's not active - it can go higher even above 100*C (that is - if you disable PC PROCHOT too).
 
Last edited by a moderator:
Back
Top