RTX-3060 Laptop GPU thermally throttling without reason

[VortexSupernova]

Fellas,
I've been attempting to optimise my damn i9 for almost a year now (bought late september 2022). I understand that Acer is not a great gaming laptop company, but I've come to find out that no company has great gaming laptops, and this one was the best available in my area. Unfortunately, 4 months later this same laptop with the 12900H and a 3070 instead of a 3060 came out, so I am very pissed from that.

Basically, this laptop has jack shit heat dissipation. I solved that... ish, with the IETS GT500 (it works very damn well). The only thing this laptop does right is the GPU never hitting 60 degrees. But that's also a problem: GPU throttles for absolutely no reason, and BD PROCHOT isn't on.
My PROCHOT is set to 84 degrees because I'm paranoid and because I want to try get heat levels lower than that at 4.3GHz+.
I've unlocked the undervolting in my laptop, then had an EDP other issue which I solved (I think?) with ICC Max slider set to 255.75A. I've not found a good undervolt yet, but essentially what I want to do is increase my PL1 and PL2 to get more clock speeds, and lower the temperatures. Lowering temperatures is the first problem.
Without any touches on ThrottleStop (e.g. removing the config file) my CPU rapidly oscillates from 55-92 degrees very easily (92 is the PROCHOT when removing the config file), even on idle. Gaming is basically a constant thermal throttle at that point.
The IETS GT500 has significantly reduced ambient internal temperatures, my NVMe SSD is no longer burning up and the idle temperatures are brought down. However, I've heard that undervolting is the best way to bring temperatures down (besides underclocking). If I run around a speed shift of 39 max, at around EPP 84, I get no throttling... or almost none. The GPU (RTX 3060) still thermally throttles for weird reasons that remain unclear to me, and I'd also like to deal with that. I can run a rough 220-240MHz boost clock with 1000MHz+ overclock on VRAM and it's stable, and the temperature still won't exceed 60, but it throttles anyways.

This is my gaming profile in FIVR. Note that the realtime values do not reflect the actual settings, I'm not currently using the gaming profile.
I've searched far and wide and not many forums exist about the i9-11900H. Honestly, I'm very pissed I missed out on the 12900H. That would've been a lot less temperature-fricked.
Does anyone know any settings to best optimise my CPU temperature? Will keeping ICC Max at 255.75 negate the undervolt?

Thanks in advance. Happy to provide additional info; again, BD PROCHOT is disabled for this profile, Disable Turbo is unchecked, clock mod is set to 100 (this actually does work on my CPU, oddly enough, despite what I've been told, which I used to use to dial back speeds for gameplay consistency).
TPL window is this:

[Edit]: My CPU Cache offset undervolt doesn't seem to be applying either, which is weird. Any ideas? Core Isolation is off, Intel VTX and VTD forcibly disabled through BIOS.

 

[VortexSupernova]

I've been using TS for over a half year now, and I've noticed my GPU's reasons for throttling are *pathetic*. Thermally throttling? Bullsh*t. It's 50 degrees, maxing at around 56-57 at worst. It never has EVER hit 60. I've overclocked it by roughly 250MHz GPU and 1000MHz+ on VRAM, and it still won't go over 60 degrees. It won't throttle constantly, but it will throttle at some time or another. I don't know how I'm meant to fix this. Is there something I'm missing? I've set the GPU throttle in throttlestop to be 76 degrees, which is nowhere near the max of 56-57.
Any ideas?

 

[unclewebb]

Not sure what you are adjusting. ThrottleStop has zero control of Nvidia GPUs. It cannot be used to the change the throttling temperature of a Nvidia GPU.

Nvidia GPU throttling in laptops is often times power related. Post some screenshots of ThrottleStop including the FIVR and TPL windows.

 

[VortexSupernova]

Not sure what you are adjusting. ThrottleStop has zero control of Nvidia GPUs. It cannot be used to the change the throttling temperature of a Nvidia GPU.

Nvidia GPU throttling in laptops is often times power related. Post some screenshots of ThrottleStop including the FIVR and TPL windows.

I made a post with my FIVR and TPL, but lemme reupload here
TPL:

FIVR: (also, for some reason my iGPU Unslice doesn't exist, it used to before I unlocked my CPU's undervolt; similarly, the CPU Cache offset won't apply)

I've run a log on a TS_Bench on 4-thread, I got PL2 warnings in the limit reasons for Core and GPU, and "thermal" on both, along with the red PROCHOT. However, the log says nothing about that. In fact, the log says I never reached my PROCHOT (I set it to 84 because this i9 is aggressive like nothing else in terms of temperature).

 

[MentalAcetylide]

I'm not sure how laptops are currently as mine is 10 yrs old now, but they tend to be higher maintenance(more frequent cleaning/dust removal from the insides) if you want to keep thermals reasonable. Even then, they can still run hot. Not that I ever plan on it, but if I ever had to get another laptop, I would certainly be investing in some kind of cooling pad for underneath it. Hope you get it sorted out!

 

[VortexSupernova]

I'm not sure how laptops are currently as mine is 10 yrs old now, but they tend to be higher maintenance(more frequent cleaning/dust removal from the insides) if you want to keep thermals reasonable. Even then, they can still run hot. Not that I ever plan on it, but if I ever had to get another laptop, I would certainly be investing in some kind of cooling pad for underneath it. Hope you get it sorted out!

My man, my GPU is below 60 degrees regardless of the load placed on it. My cooling pad is the best one out there too, the IETS GT500. I have no thermal problems with the GPU. BD PROCHOT isn't on. Something else is happening and I'd love to know what it is.

 

[unclewebb]

I set it to 84

Show me a screenshot of what you set to 84.

I am not sure what CPU model you have so post a screenshot of that too. The Options window would be useful and so would a log file and a Limit Reasons screenshot when anything is lighting up red.

Delete the ThrottleStop.INI configuration file. This file can get corrupted when VBS is enabled.

 

[MentalAcetylide]

My man, my GPU is below 60 degrees regardless of the load placed on it. My cooling pad is the best one out there too, the IETS GT500. I have no thermal problems with the GPU. BD PROCHOT isn't on. Something else is happening and I'd love to know what it is.

I would be curious to know, too. 60 degrees celsius is excellent. When I'm rendering with both RTX 3090 + RTX A6000 together for a couple of hours, the workstation card hits around 80 C and 1-2 of the VRMs on the 3090 will hit between 60-66 C with the rest of the VRM temps falling somewhere between 40-54 C.

 

[A&P211]

What model laptop so you have?

 

[demirael]

I've been using TS for over a half year now, and I've noticed my GPU's reasons for throttling are *pathetic*. Thermally throttling? Bullsh*t. It's 50 degrees, maxing at around 56-57 at worst. It never has EVER hit 60. I've overclocked it by roughly 250MHz GPU and 1000MHz+ on VRAM, and it still won't go over 60 degrees. It won't throttle constantly, but it will throttle at some time or another. I don't know how I'm meant to fix this. Is there something I'm missing? I've set the GPU throttle in throttlestop to be 76 degrees, which is nowhere near the max of 56-57.
Any ideas?

I have a feeling you're confused. ThrottleStop, FIVR and such only control the CPU (including the iGPU). For any discrete GPU you need different software. Install something like GPU-Z, that'll at least show the GPU's reason for throttling if any.

 

[unclewebb]

Here is a convenient link to the other post you made.

Acer Nitro 5 i9-11900H undervolting

Fellas, I've been attempting to optimise my damn i9 for almost a year now (bought late september 2022). I understand that Acer is not a great gaming laptop company, but I've come to find out that no company has great gaming laptops, and this one was the best available in my area. Unfortunately...

www.techpowerup.com

After you delete the ThrottleStop.INI file, try adjusting the Cache voltage and see if that works. Your undervolt might not work correctly until both the core and cache offset voltage values are showing up correctly in the FIVR monitoring table.

If the cache voltage adjustment is still not working then post a screenshot of Windows Features. You might still have something enabled in there that is virtualization related that is preventing ThrottleStop from accessing your voltage control register.

I am still using Windows 10. Compare how your feature list compares to my list. These three definitely need to be disabled.
Virtual Machine Platform
Windows Sandbox
Windows Subsystem for Linux

I would leave PROCHOT Offset at its default setting. At maximum performance, the 11900H processors will run as hot as the sun. That is just the way it is. Most laptops with this processor are going to power limit throttle or get too hot and thermal throttle. You will have to slow your CPU down significantly to avoid this. Whatever size heatsink that is used, it is hardly ever enough for maximum performance. Allowing users to overclock the 11900H was wishful thinking by both Intel and Acer.

 

[VortexSupernova]

Show me a screenshot of what you set to 84.

I am not sure what CPU model you have so post a screenshot of that too. The Options window would be useful and so would a log file and a Limit Reasons screenshot when anything is lighting up red.

Delete the ThrottleStop.INI configuration file. This file can get corrupted when VBS is enabled.

VBS, VTD/X are all disabled anyways. But I'll try your method, it might actually be because of VBS since I just tried retrieving my config after I rebooted with undervolts now accessible.

I have PROCHOT at 84 degrees because I want to try make my clocking frequency skim off it without just ramming head-on into it. As in, I want it to maximise roughly around there instead of try run headlong into the 90s. It's a precaution for lifespan's sake. If I can get it running a good 4.4-4.6 or 4.7GHz when gaming and running heavy processes and only skim that limit, I won't need to do anything more. That's literally all I want. I used to use the alarm system as a heavy down-clock, but now I use it to downclock to like 3GHz so it's not as extreme.

As for those checkboxes, two of the three you stated are disabled, the third isn't existent (windows sandbox).

I would be curious to know, too. 60 degrees celsius is excellent. When I'm rendering with both RTX 3090 + RTX A6000 together for a couple of hours, the workstation card hits around 80 C and 1-2 of the VRMs on the 3090 will hit between 60-66 C with the rest of the VRM temps falling somewhere between 40-54 C.

Sounds like we have a similar problem.

I have a feeling you're confused. ThrottleStop, FIVR and such only control the CPU (including the iGPU). For any discrete GPU you need different software. Install something like GPU-Z, that'll at least show the GPU's reason for throttling if any.

I understand this. My CPU is meant to pull a lot of power, but my GPU doesn't seem to be. If I recall correctly, when my GPU is under load, it shows a power-related performance cap. Which makes very little sense, considering my CPU is limited to basically half its maximum power draw (iirc it can pull over 110W, it's only allowed 65, 70 at worst on PL2, PL1 is 47 [just above standard 45W] ). I thought



[edit]: I deleted my config file and redid it all the same as it was before, cache is now working, thanks very much. I've been blue screened 3 times now, and I believe it's due to me setting my 1 and 2-core boosts to 54 (I wanted to see if it would work, but as soon as I applied it the last time, I froze and got the clock watchdog timeout thingo). I won't mess with that anymore, because I'd like to find the ideal undervolt. I'm running -40 and -25 core and cache when on power, -55 and -25 when using my gaming profile.

 

[unclewebb]

Limiting all of the turbo ratios to a fixed number like 40 might be a way to control heat. I find it is easier to get an undervolt stable when you limit the maximum turbo ratios. Less light load performance but maybe more consistent performance with less power limit or thermal throttling. Adjust the turbo values so your computer does not get so hot.

when my GPU is under load, it shows a power-related performance cap

What software are you using that shows this. I think GPU-Z does a good job showing Nvidia GPU limits. The ThrottleStop Limit Reasons window only shows throttling info for the Intel GPU which is not very useful for most people.

cache is now working

Thanks for confirming that.

The 11900H is basically overclocked by Intel. When running at full speed, it needs all of the voltage and power you can give it. That means there is not a lot of room for undervolting. Slowing the CPU down by lowering the turbo limits might allow you to undervolt some more.

 

[VortexSupernova]

Limiting all of the turbo ratios to a fixed number like 40 might be a way to control heat. I find it is easier to get an undervolt stable when you limit the maximum turbo ratios. Less light load performance but maybe more consistent performance with less power limit or thermal throttling. Adjust the turbo values so your computer does not get so hot.


What software are you using that shows this. I think GPU-Z does a good job showing Nvidia GPU limits. The ThrottleStop Limit Reasons window only shows throttling info for the Intel GPU which is not very useful for most people.


Thanks for confirming that.

The 11900H is basically overclocked by Intel. When running at full speed, it needs all of the voltage and power you can give it. That means there is not a lot of room for undervolting. Slowing the CPU down by lowering the turbo limits might allow you to undervolt some more.

GPU-Z is what is monitoring the GPU, I've done numerous recordings and it's mostly pwr, sometimes vrel or a mixture of two. It's never temperature. But throttlestop says my CPU and GPU thermally throttle even when I haven't gone into thermal throttle range. Idk what's going on there tbh, the GPU thermal warning is what's worrying. Should I undervolt the iGPU? To make it stop throttling, that is.

What do you mean by "i9 is overclocked by intel"? I notice that the max limit oscillates from 48 to 49 for some reason, I haven't found the proper reason for it.
All I know is that I can run relatively good temperatures (I still throttle) at max 41 (when gaming), but I want to push my limits higher. Like, maybe use 44 at 8-core.


I'll check my undervolt this avo when I play a game, the game uses AI and is very intensive on CPU, so I will test out what kind of undervolts I can run.

Another thing, whenever running off my battery, I get constant EDP other warnings, they flash red, in all 3 columns. I had this problem the first time I undervolted and setting ICCMax to 200 or so would fix it, but why is this happening on battery? I've had this for quite a while now.


[Edit]: I don't have iGPU Unslice still... I used to have it before I unlocked FIVR with the vars method... I have CPU core, cache, Intel GPU, and System Agent, but that's it. I know I used to have iGPU Unslice, but now that I've unlocked FIVR (since like 1-2 weeks ago) it's just... poof. Gone. Reduced to atoms.


[Edit 2]: I've looked quite literally everywhere but nobody can explain what the cache ratios do. I know how the sleep settings work, but I don't know exactly what the ratios do in reference to CPU performance. Could you explain it?

What model laptop so you have?

Acer Nitro 5, AN515-57. NVMe SSD used to be a prick and a half but my cooling pad has eliminated that issue.

 

[unclewebb]

Should I undervolt the iGPU?

I never do or recommend doing that. If you have a Nvidia GPU, there is nothing to be gained by undervolting the Intel GPU.

i9 is overclocked by intel

When I was young, you could overclock an Intel CPU by 50% or more. Now Intel ships their high end CPUs so they are running on the edge of stability. The default voltage is cranked up which creates a lot of heat. There is very little room for any further overclocking because even at default settings, you are already at the thermal limits of what the cooling can manage. Your CPU includes a feature so you can overclock it another 500 MHz. Trying to use that feature would create a ridiculous amount of heat in your laptop. It is difficult for any manufacturer to create a heatsink big enough to manage the heat an 11900H can produce. The full potential of this powerful CPU is limited by the cooling.

Intel is doing the same thing with their desktop 13900K and 13900KS. It is difficult to keep these CPUs under 100°C during a stress test even when using top end AIO coolers. Intel is running their top end processors flat out so they can compete with AMD.

why is this happening on battery?

Some manufacturers set much more restrictive power or current limits which are used as soon as you switch to battery power. EDP is current throttling. This is designed to protect your battery. You might need something the size of a car or motorcycle battery to fully power your CPU and Nvidia GPU. Users do not want that so some throttling is used to help keep the battery from being damaged.

I used to have iGPU Unslice

You used to have this when VBS was enabled. The ThrottleStop FIVR window does not work correctly when VBS is enabled. VBS prevents ThrottleStop from examining the CPU to see what voltages are available. Some CPUs do not have a iGPU Unslice setting. I am not sure why. I only know that this is normal so don't worry about it.

cache ratio

The cache ratio controls how fast information can be transferred between the CPU and the on chip cache memory. Slowing the cache down can improve stability. If you slow the cache down too much, some apps or games will slow down significantly. If you want to do some testing, use ThrottleStop to set the Cache Ratio Max to 8. Run Cinebench and see if your computer slows down. A fast cache is good but 100 MHz or 200 MHz slower than max makes very little difference in performance for most apps and games.

 

[A&P211]

GPU-Z is what is monitoring the GPU, I've done numerous recordings and it's mostly pwr, sometimes vrel or a mixture of two. It's never temperature. But throttlestop says my CPU and GPU thermally throttle even when I haven't gone into thermal throttle range. Idk what's going on there tbh, the GPU thermal warning is what's worrying. Should I undervolt the iGPU? To make it stop throttling, that is.

What do you mean by "i9 is overclocked by intel"? I notice that the max limit oscillates from 48 to 49 for some reason, I haven't found the proper reason for it.
All I know is that I can run relatively good temperatures (I still throttle) at max 41 (when gaming), but I want to push my limits higher. Like, maybe use 44 at 8-core.


I'll check my undervolt this avo when I play a game, the game uses AI and is very intensive on CPU, so I will test out what kind of undervolts I can run.

Another thing, whenever running off my battery, I get constant EDP other warnings, they flash red, in all 3 columns. I had this problem the first time I undervolted and setting ICCMax to 200 or so would fix it, but why is this happening on battery? I've had this for quite a while now.


[Edit]: I don't have iGPU Unslice still... I used to have it before I unlocked FIVR with the vars method... I have CPU core, cache, Intel GPU, and System Agent, but that's it. I know I used to have iGPU Unslice, but now that I've unlocked FIVR (since like 1-2 weeks ago) it's just... poof. Gone. Reduced to atoms.


[Edit 2]: I've looked quite literally everywhere but nobody can explain what the cache ratios do. I know how the sleep settings work, but I don't know exactly what the ratios do in reference to CPU performance. Could you explain it?


Acer Nitro 5, AN515-57. NVMe SSD used to be a prick and a half but my cooling pad has eliminated that issue.

Maximum Graphics Power

• Up to 95 W

There is the main problem, very bad cooling. It's a low powered 3060. A normal 3060 is rated at a 115w. With the nvidia powerboost, or whatever it's called, the max is 140w.

 

[VortexSupernova]

I never do or recommend doing that. If you have a Nvidia GPU, there is nothing to be gained by undervolting the Intel GPU.


When I was young, you could overclock an Intel CPU by 50% or more. Now Intel ships their high end CPUs so they are running on the edge of stability. The default voltage is cranked up which creates a lot of heat. There is very little room for any further overclocking because even at default settings, you are already at the thermal limits of what the cooling can manage. Your CPU includes a feature so you can overclock it another 500 MHz. Trying to use that feature would create a ridiculous amount of heat in your laptop. It is difficult for any manufacturer to create a heatsink big enough to manage the heat an 11900H can produce. The full potential of this powerful CPU is limited by the cooling.

Intel is doing the same thing with their desktop 13900K and 13900KS. It is difficult to keep these CPUs under 100°C during a stress test even when using top end AIO coolers. Intel is running their top end processors flat out so they can compete with AMD.


Some manufacturers set much more restrictive power or current limits which are used as soon as you switch to battery power. EDP is current throttling. This is designed to protect your battery. You might need something the size of a car or motorcycle battery to fully power your CPU and Nvidia GPU. Users do not want that so some throttling is used to help keep the battery from being damaged.


You used to have this when VBS was enabled. The ThrottleStop FIVR window does not work correctly when VBS is enabled. VBS prevents ThrottleStop from examining the CPU to see what voltages are available. Some CPUs do not have a iGPU Unslice setting. I am not sure why. I only know that this is normal so don't worry about it.


The cache ratio controls how fast information can be transferred between the CPU and the on chip cache memory. Slowing the cache down can improve stability. If you slow the cache down too much, some apps or games will slow down significantly. If you want to do some testing, use ThrottleStop to set the Cache Ratio Max to 8. Run Cinebench and see if your computer slows down. A fast cache is good but 100 MHz or 200 MHz slower than max makes very little difference in performance for most apps and games.

View attachment 297444

I'm a bit new to this place so I don't know how to quote individual text, but as for the cache ratio: I'll try see what happens.
My CPU can run 47 on single and double thread stress tests, doesn't seem to want to hit 48, might be power related but I don't think so. It hits around 60, 70 at worst. The heat only gets out of hand at 8 thread and higher. 4 threads will see some throttling momentarily as it skims 84C, but 6,8,etc will cause too much heat and it will just dive into the 84C threshold and I'll even see my temps fluctuate between 85 and 50 as the power plan changes. I might need to change some turbo clocks when more cores are active, since the 1 and 2 core boosts are fine, but more is too much.
What is my min max cache ratio btw? Default puts it at 8-39, but I've heard a cache has a max that is some fixed constant away from the turbo limit of the core. my core does 8-48 or 8-49 (I still don't know why it changes), so how do I deduce the max from that?

Also, I used to have my Memory DDR have an actual number. Now it's just "--". Also, the second box (next to what SHOULD be the memory ddr) says muCode 3C. Is that relevant in any way to what I do?

Also, I should keep sleep defaults checked, right? So that I don't run into errors when i use this at uni when opening and closing.

I got stuck in a bluescreen loop last night, I thought I'd lost my computer completely, because I did something that didn't immediately seem too bad. This is probably the only time I'll ever appreciate a laptop battery; power mode gets enforced from startup and the loop ends. I just unplugged and the loop stopped happening. Useful knowledge for the future yk?

What is the reason behind "static" voltage (the option beside adaptive). I pressed apply once on my CPU cache, and this was before I got the offset to apply, and then my laptop just immediately died. What is the point of having static voltage if it's just gonna end your computer lol?

Maximum Graphics Power

• Up to 95 W

There is the main problem, very bad cooling. It's a low powered 3060. A normal 3060 is rated at a 115w. With the nvidia powerboost, or whatever it's called, the max is 140w.

No it's very good cooling, it never goes above 60 degrees. I overclock it too, and it still doesn't go over 60 degrees. Is Nvidia Powerboost an intergrated system or is it something I gotta install?
Edit: GPU cooling is great, CPU cooling is not. My IETS GT500 on full power (I have noise cancelling headphones) just tears the air out of the laptop, which is why my NVMe is no longer a problem. Despite what other people say about cooling pads, this thing is an absolute beast. I'm not talking 5 degrees, I'm talking 15 degrees on average. GPU idle temperature drops 25 degrees on a very low RPM on the cooling pad, even when both built-in computer fans are not even spinning. It's also reduced my CPU's thermal throttle frequency a tonne too (frequency as in how often it happens). My main concern is the spikes, which are pretty intense with an i9-11900H, as you can imagine. With my PROCHOT at 84, the worst spike I've had is 89 degrees. That pretty munted. This is also the main reason I turned BD PROCHOT off. Idk if the iGPU causes the CPU to throttle too, but there definitely is a lot of BD PROCHOT going on with my CPU in games and its pretty heavy. All this heat is really just isolated to the CPU, the Nvidia GPU is more than fine.

 

[gQx]

Hmm seems just like me you like to experiment and tweak things here and there so much maybe a bios reset + clean windows install should fix that for you I know I messed with my pc so much didn't realize when or what caused an issue many times even though I tought it can't be something I did. well sometimes it meant hardware damage so format + bios reset didn't help couple of times

 

[VortexSupernova]

Hmm seems just like me you like to experiment and tweak things here and there so much maybe a bios reset + clean windows install should fix that for you I know I messed with my pc so much didn't realize when or what caused an issue many times even though I tought it can't be something I did. well sometimes it meant hardware damage so format + bios reset didn't help couple of times

My thermals are a thing that occurred since the beginning. I'm just trying to do what Acer couldn't be fricked to do.

I never do or recommend doing that. If you have a Nvidia GPU, there is nothing to be gained by undervolting the Intel GPU.


When I was young, you could overclock an Intel CPU by 50% or more. Now Intel ships their high end CPUs so they are running on the edge of stability. The default voltage is cranked up which creates a lot of heat. There is very little room for any further overclocking because even at default settings, you are already at the thermal limits of what the cooling can manage. Your CPU includes a feature so you can overclock it another 500 MHz. Trying to use that feature would create a ridiculous amount of heat in your laptop. It is difficult for any manufacturer to create a heatsink big enough to manage the heat an 11900H can produce. The full potential of this powerful CPU is limited by the cooling.

Intel is doing the same thing with their desktop 13900K and 13900KS. It is difficult to keep these CPUs under 100°C during a stress test even when using top end AIO coolers. Intel is running their top end processors flat out so they can compete with AMD.


Some manufacturers set much more restrictive power or current limits which are used as soon as you switch to battery power. EDP is current throttling. This is designed to protect your battery. You might need something the size of a car or motorcycle battery to fully power your CPU and Nvidia GPU. Users do not want that so some throttling is used to help keep the battery from being damaged.


You used to have this when VBS was enabled. The ThrottleStop FIVR window does not work correctly when VBS is enabled. VBS prevents ThrottleStop from examining the CPU to see what voltages are available. Some CPUs do not have a iGPU Unslice setting. I am not sure why. I only know that this is normal so don't worry about it.


The cache ratio controls how fast information can be transferred between the CPU and the on chip cache memory. Slowing the cache down can improve stability. If you slow the cache down too much, some apps or games will slow down significantly. If you want to do some testing, use ThrottleStop to set the Cache Ratio Max to 8. Run Cinebench and see if your computer slows down. A fast cache is good but 100 MHz or 200 MHz slower than max makes very little difference in performance for most apps and games.

View attachment 297444

I'd just like to give an update. I seem to have actually exceeded the i9's default clock. I blue screened at -105mV, then at -95mV, so I decided to just back it down to -80mV on both core and cache. It runs stable now. The temperatures still spike but, paired with the IETS GT500 on full blast, my own (internal) fans on auto, playing a game that chews the CPU and its cores, I ran a complete 4.49GHz turbo clock CONSTANTLY. My temperatures typically varied. I could get 55-60 stable on some matches, or 65-70 on other matches. My max temperature recorded was 86, which is 2 degrees over the PROCHOT threshold. I only occasionally got PL2 yellow on my core and GPU in the limits, and my PL2 is set at 75W, with PL1 50 clamped.
If I ever thermally throttled (I did, dunno how many times, maybe a few but only very briefly) it was unnoticeable.
This is amazing. I'm essentially running a limit of 45 where it should be running 44. My i9's turbo is rated to 4.4GHz turbo on 8-core, yet I'm running 4.49GHz (which is basically 4.5) on 8-core turbo and the thermals are being maintained exactly as I wanted. I've also put my core and cache at the same mV, for some reason they both can only undervolt to the same extent. The cache ratio min/max are both on 8, I don't really know if it's done much but since it's not hurt my temperatures and clocks, I'll keep it like so.

Here's my settings btw:

Yes, I fully know I won't be hitting 5.4GHz anytime *ever*, but it can't hurt. Just thought you should know how I did it, considering I'm running the i9 better than it was sold.

 

[unclewebb]

My i9's turbo is rated to 4.4GHz turbo on 8-core, yet I'm running 4.49GHz

The 11900H also supports +5 bins of overclocking. That means you should be able to run the 8 Cores turbo ratio to a maximum of 49. You might have to move to the North Pole to make that a practical setting.

I used to have my Memory DDR have an actual number. Now it's just "--".

It might have been showing a big number when you had VBS enabled but the number it was showing was not accurate. Many items in the FIVR window do not work correctly when VBS is enabled. Not all Intel CPUs report a DDR memory speed through the FIVR register. I am not sure why. Use CPU-Z or similar to track your memory speed.

sleep defaults checked, right?

You do not need to check anything unless you have a problem. If your laptop has a problem resuming from sleep, you can try checking the Sleep Defaults boxes.

What is the reason behind "static" voltage

Many desktop users like to run their CPU at a fixed speed. For them, a static voltage that never changes makes sense. If I had a laptop I would always leave the voltage setting on Adaptive. That works well in most situations.

I still don't know why it changes

You have Thermal Velocity Boost checked in the FIVR window. This tells the CPU to slow down 100 MHz at about 70°C. I do not like that feature so I clear that box. This might be why your MHz are not consistent. Intel calling this feature Boost is misleading. It actually causes throttling. The marketing department gave it a name so users would be excited by this new throttling feature.

 

[VortexSupernova]

View attachment 297586

The 11900H also supports +5 bins of overclocking. That means you should be able to run the 8 Cores turbo ratio to a maximum of 49. You might have to move to the North Pole to make that a practical setting.


It might have been showing a big number when you had VBS enabled but the number it was showing was not accurate. Many items in the FIVR window do not work correctly when VBS is enabled. Not all Intel CPUs report a DDR memory speed through the FIVR register. I am not sure why. Use CPU-Z or similar to track your memory speed.


You do not need to check anything unless you have a problem. If your laptop has a problem resuming from sleep, you can try checking the Sleep Defaults boxes.


Many desktop users like to run their CPU at a fixed speed. For them, a static voltage that never changes makes sense. If I had a laptop I would always leave the voltage setting on Adaptive. That works well in most situations.


You have Thermal Velocity Boost checked in the FIVR window. This tells the CPU to slow down 100 MHz at about 70°C. I do not like that feature so I clear that box. This might be why your MHz are not consistent. Intel calling this feature Boost is misleading. It actually causes throttling. The marketing department gave it a name so users would be excited by this new throttling feature.

A few things:
1- what does ring down bin do? I'd like to know exactly what it is if I do end up changing it.

2- I only have TVB on because I don't see the point of running 4.9GHz on only 2 cores, at least half of my cores are active whenever I use my laptop for anything. If it simply acts as a small buffer on 1-2 cores when reaching higher temperatures, I'm fine with that. Again, the idea of my performance is to only skim the 84C thermal limit, not plunge through it. Considering that I'm running this thing pretty fast now, I have really no intention of pushing it further.

3- why did my cache ratio like not impact anything? By constricting it to 8, I don't really know if anything happened, whether it be performance or temperatures. What *would* happen, theoretically?