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

Help me figure out best temp settings for i9 13980hx

Cyneptic

New Member
Joined
Aug 19, 2023
Messages
5 (0.01/day)
Hey guys, I have recently purchased a laptop with i9-13980hx (g16 rog with 4080 incase that matters), as one would suspect this beast of a cpu is really hot, and by really hot im talking spikes to 100c hot.

funny thing is in games like cyberpunk im getting consistent 80-90ish temps, but in much less demanding games such as phasmophobia/dead by daylight / world of warcraft, im getting spikes upto 100, sometimes even 101-102

I find it super hard not to worry about these temps as this is my first laptop and i'm used to temps being sub-70, I would love your comment on this and whether or not constant 95-96 or spikes to 100 (sometimes even 101-102) is something to worry about or not considering this is going to be my main machine for atleast a couple of years and i kind of spent ~14 hours a day on it.

that aside (I appreciate any comments regarding the question above) I wanted to ask for help regarding undervolting/tpl/prochot and anything else there is that can help me out in managing my temps, I have messed with throttlestop for about a week now but anything besides straight up setting prochot to 75 or something doesn't seem to keep the temp off. I asked on reddit whether or not setting prochot to 75 would damage my system long term and they suggested I ask here for help in my specific case.

so here goes, as you suspect i9 13980hx is probably an overkill for almost any game out there currently, with a prochot on 75C im getting consistent 80-120 fps on cyberpunk max quality. so I'm willing to sacrifice some performance in order to get better temps, I don't really know what temps im aiming for but my goal is for it to be somewhere that it wouldn't damage my system by being there constantly (+12/14 hours a day) for atleast a couple of years. obviously some performance will be sacrificed in order to achieve such temps if that is possible, but i'd like it to be as minimal as possible, so if you would be so kind as to guide me on figuring out what the best settings are for my case i'd appreciate it.

cheers.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
8,239 (1.34/day)
Intel has been setting the thermal throttling temperature of almost all of their Core i CPUs produced during the last 15 years to 100°C. Intel must be confident that their CPUs can run reliably at high temperatures without an excessive number of warranty claims. As long as my computer is running reliably, I rarely think twice about what temperature it is running at.

Lowering the thermal throttling temperature below 100°C will reduce maximum performance. If you are happy with that, your CPU does not care what temperature it runs at. You will not hurt it either way.

Post some ThrottleStop screenshots if you need some help. Include the FIVR and TPL windows. I cannot guess what settings your BIOS has locked or left unlocked. Asus tends to lock their new laptops down more than most manufacturers do. Avoid updating the BIOS if your laptop is not locked.
 

Cyneptic

New Member
Joined
Aug 19, 2023
Messages
5 (0.01/day)
Intel has been setting the thermal throttling temperature of almost all of their Core i CPUs produced during the last 15 years to 100°C. Intel must be confident that their CPUs can run reliably at high temperatures without an excessive number of warranty claims. As long as my computer is running reliably, I rarely think twice about what temperature it is running at.

Lowering the thermal throttling temperature below 100°C will reduce maximum performance. If you are happy with that, your CPU does not care what temperature it runs at. You will not hurt it either way.

Post some ThrottleStop screenshots if you need some help. Include the FIVR and TPL windows. I cannot guess what settings your BIOS has locked or left unlocked. Asus tends to lock their new laptops down more than most manufacturers do. Avoid updating the BIOS if your laptop is not locked.
Thank you for your response, I do not mind sacrificing some performance, someone on reddit mentioned that even though cpu itself might not be damaged its possible that nearby components get hurt over time from such high temps?

these are some screen shots from my throttlestop, i have done some massive limits for now until I get a response here, but the default turbo boost ratios and tpl are:
P-Cores = 1-2: 56, rest: 52
E-Cores = 40
PL1 = 175
PL2 = 175
Thermal Velocity Boost was ON and I turned it OFF thinking it had something to do with thermals?
1692464076177.png

1692464088382.png

1692464148912.png

1692464130703.png


I suspect these are too extreme, but it was mainly just to check out different settings and their effects, and ofcourse the fact that I do not know what temp to aim at to stay on the safe side considering my +14h a day of constant use
 

Cyneptic

New Member
Joined
Aug 19, 2023
Messages
5 (0.01/day)
Intel has been setting the thermal throttling temperature of almost all of their Core i CPUs produced during the last 15 years to 100°C. Intel must be confident that their CPUs can run reliably at high temperatures without an excessive number of warranty claims. As long as my computer is running reliably, I rarely think twice about what temperature it is running at.

Lowering the thermal throttling temperature below 100°C will reduce maximum performance. If you are happy with that, your CPU does not care what temperature it runs at. You will not hurt it either way.

Post some ThrottleStop screenshots if you need some help. Include the FIVR and TPL windows. I cannot guess what settings your BIOS has locked or left unlocked. Asus tends to lock their new laptops down more than most manufacturers do. Avoid updating the BIOS if your laptop is not locked.
So far ive lowered the boost ratios for P_Cores from 56 and 52 to 36 and E_Cores to 10 instead of 40, also set a tpl of pl1 90 and pl2 100. these are quite extreme I believe however as I don't know what temp to aim for I guess I'm just playing around with it.

however I did notice something, it appears that the E_Cores are for some reason prioritized instead of P_Cores when it comes to power, when I lower the tpl if the E_Cores still have the 40 boost ratio and P_Cores have something like 30, my P_Cores go to 1GHz and give the power to E_Cores instead, and as I lower the E_Core ratios I get higher P_Core ratios...

I have no idea what temp to aim for and/or what settings have the most effect on temps and how I can lower the temps while maximizing performance as much as possible. i'll be waiting for your response.

thanks again.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
8,239 (1.34/day)
what temp to aim for
You could try aiming for the Intel recommended maximum temperature which is 100°C. Our goals are different. I have a desktop computer so I always prefer maximum performance.

The original purpose of ThrottleStop was to fix throttling problems so users could get maximum performance out of their mobile CPUs. You are doing the opposite. Don't tell anyone that ThrottleStop is crippling your CPU or it might get a bad reputation.

Thermal Velocity Boost was ON and I turned it OFF
When Thermal Velocity Boost is checked in the FIVR window, this tells the CPU to slow down 100 MHz when the CPU reaches approximately 70°C. What Intel calls Boost is actually a throttling feature. Clearing the Thermal Velocity Boost box tells the CPU not to TVB throttle.

i'll be waiting for your response
I have zero hands on experience with Intel 13th Gen technology. I am always a little surprised that ThrottleStop works as well as it does on these CPUs. You will have to decide what ThrottleStop settings give you the best performance vs temperature compromise that you are looking to achieve. Hard for me to make any suggestions if I have never used a 13980HX or anything similar to your CPU.
 

Cyneptic

New Member
Joined
Aug 19, 2023
Messages
5 (0.01/day)
You could try aiming for the Intel recommended maximum temperature which is 100°C. Our goals are different. I have a desktop computer so I always prefer maximum performance.

The original purpose of ThrottleStop was to fix throttling problems so users could get maximum performance out of their mobile CPUs. You are doing the opposite. Don't tell anyone that ThrottleStop is crippling your CPU or it might get a bad reputation.


When Thermal Velocity Boost is checked in the FIVR window, this tells the CPU to slow down 100 MHz when the CPU reaches approximately 70°C. What Intel calls Boost is actually a throttling feature. Clearing the Thermal Velocity Boost box tells the CPU not to TVB throttle.


I have zero hands on experience with Intel 13th Gen technology. I am always a little surprised that ThrottleStop works as well as it does on these CPUs. You will have to decide what ThrottleStop settings give you the best performance vs temperature compromise that you are looking to achieve. Hard for me to make any suggestions if I have never used a 13980HX or anything similar to your CPU.
I understand, as a guideline should I be aiming to lower the power limit in order to get better temps or lower the boost ratio? i'm wondering which one do you think will have the more impact.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
8,239 (1.34/day)
Without access to any similar hardware, there is no way I can answer your question. Keep playing around and use whatever settings work best for you.
 

PowerThrottle

New Member
Joined
Aug 13, 2023
Messages
19 (0.03/day)
By the way, since the topic of temperature has already come up, I also have some slight confusion regarding the adjustments on this CPU when it comes to Undervoltage settings. @unclewebb, is there anything specific I should know when adjusting CPU Cache and Core settings on these CPUs (13th Gen)? Should Cache and Core be adjusted in the same way? Should the Offset Voltage values for both Cache types be the same? Additionally, what would you do if you were in my shoes? All Core: 38x. I was able to lower both Cache types to -90. I haven't experimented with CPU Core settings. It's working at -90. I brought the system down to -110, but I encountered some minor blue screen issues. In the -110 attempt, I had only adjusted CPU Core, CPU E, and P Cache settings.

1692632733189.png

1692632754584.png
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
8,239 (1.34/day)
is there anything specific I should know
I have not done any hands on testing of ThrottleStop on a 13th Gen or 12th Gen CPU.

As far as I know, the P core and P cache offset voltages do not need to be set equal to each other. On some previous gen CPUs, you might get slightly better results by setting the P core to a bigger offset compared to the P cache.

Many people leave all of the other voltages including the E cache at an offset of +0.0000.

Post your results if you learn anything new about what voltages these CPUs like.
 

PowerThrottle

New Member
Joined
Aug 13, 2023
Messages
19 (0.03/day)
I have not done any hands on testing of ThrottleStop on a 13th Gen or 12th Gen CPU.

As far as I know, the P core and P cache offset voltages do not need to be set equal to each other. On some previous gen CPUs, you might get slightly better results by setting the P core to a bigger offset compared to the P cache.

Many people leave all of the other voltages including the E cache at an offset of +0.0000.

Post your results if you learn anything new about what voltages these CPUs like.
Hello again. I've mostly completed the relevant tests. The CPU E/P Cache doesn't bring much improvement, but they respond well to undervolting. When I went from -85 to -125 (with the same Core Offset Voltage), I only achieved a reduction of 1.3W. However, since these stress tests didn't heavily stress the Cache and the power consumption hardly changed, I reverted the Cache values to their stable state. But things are different for the Core. There's a 10W decrease between -170 Core and -120 in an 8-thread test. The Core seems more crucial. I'll continue the tests a bit more. By the way, I haven't encountered any BSODs.

By the way, I might have also understood the RING or CORE EDP issue. As I mentioned, I believe it's related to DRAM RAPL or Dynamic Boost. The first possibility seems more likely.
 

Cyneptic

New Member
Joined
Aug 19, 2023
Messages
5 (0.01/day)
Hello again. I've mostly completed the relevant tests. The CPU E/P Cache doesn't bring much improvement, but they respond well to undervolting. When I went from -85 to -125 (with the same Core Offset Voltage), I only achieved a reduction of 1.3W. However, since these stress tests didn't heavily stress the Cache and the power consumption hardly changed, I reverted the Cache values to their stable state. But things are different for the Core. There's a 10W decrease between -170 Core and -120 in an 8-thread test. The Core seems more crucial. I'll continue the tests a bit more. By the way, I haven't encountered any BSODs.

By the way, I might have also understood the RING or CORE EDP issue. As I mentioned, I believe it's related to DRAM RAPL or Dynamic Boost. The first possibility seems more likely.
while you are testing things out, i'd appreciate it if you could share your findings, what kind of temps are you aiming for and what settings you find to be the best. thanks
 

PowerThrottle

New Member
Joined
Aug 13, 2023
Messages
19 (0.03/day)
Our best way of tuning these CPUs is as follows, in order. Of course, while doing these, make sure to resolve any temperature and throttle issues to unleash maximum performance.

*Disable downclocking in C states and, if possible, try controlling Package C State to values like C1.

*Undoubtedly, you'll need to undervolt to prevent overheating. Additionally, setting the clock speed to around 4 GHz might be beneficial. I've noticed that increasing the W value primarily affects the E-Core, but there could be variability.

*Setting Timer Resolution to 0.5ms might be helpful, although it's not necessary and might decrease stability, but it can alleviate minor stuttering issues.

For now, that should be sufficient.
 

Darthsanzodius

New Member
Joined
Sep 23, 2023
Messages
3 (0.01/day)
Hi Cyneptic Strix G18 user here and also having the same sentiments you're having. For the past 2 months I've been playing games not more than 4-5 hrs noticed that my temps were not goin >85C. But last week I finished Atlas Fallen in 3 days around 8hrs/day suddenly noticed after that I'm getting as high as 92C. So I decided to try Throttlestop. What I did was just ticked of disable turbo when gaming and noticed that I'm getting 65C on my Cpu Temps and My Fps are way better from before. Am I doing right? Thanks
 

Attachments

  • Screenshot_20230923_231205_Photo Editor.jpg
    Screenshot_20230923_231205_Photo Editor.jpg
    471.7 KB · Views: 400

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
8,239 (1.34/day)
Selecting Disable Turbo will make your CPU run significantly slower. Less CPU MHz will allow the CPU to consume less power and produce less heat so your laptop will run cooler. Perhaps this allows your GPU to run better. It is impossible to say why a slower CPU gives better FPS. You would need to log data for both the CPU and Nvidia GPU while gaming. You can check the Nvidia GPU box in the ThrottleStop Options window to add some Nvidia GPU data to the ThrottleStop log file. You can also use GPU-Z to do some GPU logging.
 

KirkH420

New Member
Joined
Dec 29, 2021
Messages
9 (0.01/day)
This CPU's TJmax is 100c but that doesn't mean that that is where you want it running at, and before you go saying that this CPU "runs hot", you should consider changing your Thermal Interface Material. Sure ASUS claims that they put liquid metal on their CPUs, but I've seen photos where customers have that much/most of their LM has seeped out from in between the cpu and cooler. It's why they have to create a special foam barrier to prevent that seepage from becoming a dead laptop. Besides, why would you only want it on the CPU when the GPU could be cooler if we used a better TIM?

I do not recommend Liquid Metal for any laptop scenario, unless you plan to never move it. It does actually seep out. It's too risky when there are alternative TIMs out there which can perform nearly as good as LM. Enter Honeywell PTM7950. It's a "Phase Changing" thermal interface material that changes from a solid to a liquid when it reaches operating temperatures. You can find an 80mmX80mm square of it on Amazon for under $10 (will be enough for 4 or 5 repastes of the CPU/GPU combo). Also you can replace the Termal Pads (if your device even uses any. ASUS often uses general Thermal Paste on the VRAM and VRMs).

What do I mean when I say your brand new laptop needs to be disassembled (cooler removed from GPU/CPU) and repasted?
• It's common knowledge that PC manufacturers generally SUCK at applying thermal interface materials.
• OEMs are going to use the cheapest products that they can source at bulk discount costs to them (not high performance).

*Bonus* You can also purchase a VERY functional laptop cooling pad called the IETS GT600. It's gonna run you $100 but it's worth every cent (if your laptop draws it's cool air from the bottom vents like most gaming laptops do). This device creates a seal around the bottom of your laptop (using memory foam) and then it uses a turbo fan to jam air up into the chassis.

Having done these things, you will probably be able to OVERCLOCK with Intel XTU rather than undervolt it.
Of course, you can actually do both and achieve amazing results, but my point is, you don't have to buy this crap logic that your CPU is doomed to thermal throttle. Because it isn't. My i9-13950HX doesn't go over 80c now (proof in photos), and that's when I'm benchmarking.

*P.S. I set the record for TimeSpy Extreme today with this thing! I still need to install the faster DDR5 kit that I bought, so tests will improve
( )

MSI Vector GP68HX w/ i9-13950HX and RTX4080@175w (No Overclock / No Undervolt, but XTU is installed and ready).
 
Joined
Sep 11, 2023
Messages
57 (0.10/day)
I do not recommend Liquid Metal for any laptop scenario, unless you plan to never move it. It does actually seep out. It's too risky when there are alternative TIMs out there which can perform nearly as good as LM. Enter Honeywell PTM7950. It's a "Phase Changing" thermal interface material that changes from a solid to a liquid when it reaches operating temperatures. You can find an 80mmX80mm square of it on Amazon for under $10 (will be enough for 4 or 5 repastes of the CPU/GPU combo). Also you can replace the Termal Pads (if your device even uses any. ASUS often uses general Thermal Paste on the VRAM and VRMs).
In case of notebook stuff PTM 7950 is so helpful not because of that rigamarole that walks around internet but mostly because of one thing. Due to its consistency and some usefull physical properties it can fix some defects of chips' curvature. In another words, when it is applied it levels out surface, and then when cooling system assembled together, there should be no any of "pushing out" known effects. Of course if it is genuine PTM 9750.

Liquid metal (LM) is excellent stuff too. The most innovative one comes from Coollaboratory https://www.coollaboratory.com/product-category/thermal-compound/ in my opinion. Their LM formula includes 5 or even more components that are mostly metallic. Final composition is not to be disclosed somewhere. I heard that first releases were too aggresive to some sort of surface coatings but now they made it more universal and safe.

LM is helpful when it has "right" recipe and it applied correct. If you notebook model is not prepared to LM from the factory you will have to do that either. Remember, most of that preparation is not about to be done by yourself. So this way you have to do a nickel plating for the areas where LM is to be contacted with cooling system's radiators. Then some extra steps to protect chips, die and other parts from LM in bad scenarios are to be done.

The other important thing to know if you are about to apply LM in your notebook or PC is that surface of the CPU die should be as ideally flat at least close to ideal state, a curvature is not allowed no way. If you have any, one of a possible solution will be "die lapping" technique first! Otherwise, if you have curved surface with microcaverns on the chip die, when LM is applied you will probably have an oxygen or another gas accumulated there soon. Once that happens, you will notice significant temperature rise because "microbubbles" of gas push out LM, and gas itself has very poor thermal conductivity.

The main idea of LM is to take away maximum heat from CPU die with a minimum possible surface area. That is why the real thermal conductivity of LM could be more than 80 W/mK.

Some CPUs are not much good for LM to use. They have two or more dies on the chip that could have different height sometimes.

In the end LM could give you extra 5-10 degrees less in notebook but only when it is all done absolutely correct.

Thus, the combination of PTM 7950 and Laird Tputty 607 are cheaper and much more safer for end customer. And seems it allways will be.

You can also purchase a VERY functional laptop cooling pad called the IETS GT600.
There is one good man in the internet that constantly looking for best cooling pad, and by this moment he tested several promising items.
This is his YouTube palylist of related videos with tests

So far IETS is the one that truly well engineered and performs better then more of the others.
 
Last edited:
Top