Constant EDP OTHER under Core, GPU and Ring

[Invertedstateofmind]

Hello,

I would like some opinions or suggestions on my PC. I follow this forum a lot for throttlestop and undervolting in general. As you can see from the files I attached ;
1. Undervolted on CPU core and cache, and also increase ICCmax to max.
2. Did some tampering on TPL's PL1 and PL2
Here are the questions I would like answered :
1. Why do I keep getting EDP other on my limits? And occasionally under load, PL2 would come flash red as well
2. Is there anything, in particular, I did wrong with my settings? Like anything counterintuitive?

PC specs :
GL503GE
I7 8750H
GTX 1050 TI

 

[unclewebb]

If IccMax is set to the maximum, Power Limit 4 in the TPL window can also be used to try and solve EDP OTHER throttling issues. Try setting this to 0 or to the maximum, 1023. Setting this to 0 is supposed to tell the CPU to ignore this setting.

Some computers will constantly show EDP OTHER in yellow even when the CPU is idle and you have just cleared this information in Limit Reasons. Not sure why this happens and I do not know any way to fix this.

Your log file shows some PL1 throttling at 17W. The 8750H has a 45W TDP rating so it should never be power limit throttling at such a low value. Someone at Asus dreamed up a good idea and decided to enforce power limit throttling way below the rated TDP. Not sure why Asus decided to do this or why anyone would think that doing this is a good idea. No laptop with a 45W CPU let alone a laptop marketed as a "Gaming Laptop" should ever include a throttling scheme like this.

Often times, these kind of problems cannot be solved. If you are using some Asus software to control your laptop, make sure it is set to the highest performance value. The Quiet setting often times will reduce the power limits like this.

The only other thing to try is removing the Intel DPTF driver. Brad shows how to do this.

Easily Disable Intel DPTF / IPF (Power Throttling) for Good - Brad's Hacks

Important: this is no longer necessary for many laptops because ThrottleStop‘s “Disable and Lock Power Limits” option can override DPTF. If that doesn’t work, try the following. If your CPU frequency is being reduced under load, even at low temperatures, you’re probably experiencing power...

bradshacks.com

For the 8750H, the core and cache voltages do not have to be set equally. Most users get better results by setting the core voltage higher than the cache voltage. Do some Cinebench R20 testing.

MAXON Cinebench R20.0 Download

CINEBENCH is a real-world cross platform test suite that evaluates your computer's performance capabilities. CINEBENCH is based on MAXON's award-winn

www.techpowerup.com

For a baseline, run Cinebench with the core and cache at -100.6 mV like your screenshot shows. Now leave the cache offset voltage as is and start bumping only the core in steps of -25 mV. You might get your best results or best temperatures with the core up around -200 mV. If this is not game stable, reduce the core to -175 mV or -150 mV.

Hopefully ditching that driver will fix the power limit throttling issues that Asus is using. Terrible design.

 

[Invertedstateofmind]

Hello unclewebb, thank you for the super speedy reply;

If IccMax is set to the maximum, Power Limit 4 in the TPL window can also be used to try and solve EDP OTHER throttling issues. Try setting this to 0 or to the maximum, 1023. Setting this to 0 is supposed to tell the CPU to ignore this setting.

Some computers will constantly show EDP OTHER in yellow even when the CPU is idle and you have just cleared this information in Limit Reasons. Not sure why this happens and I do not know any way to fix this.

I tried both 0 and 1023, and both yielded basically the same result with EDP yellow still on. So I guess no way around that.

Your log file shows some PL1 throttling at 17W. The 8750H has a 45W TDP rating so it should never be power limit throttling at such a low value. Someone at Asus dreamed up a good idea and decided to enforce power limit throttling way below the rated TDP. Not sure why Asus decided to do this or why anyone would think that doing this is a good idea. No laptop with a 45W CPU let alone a laptop marketed as a "Gaming Laptop" should ever include a throttling scheme like this.

Yeah, I was still tech-blind back then. Didn't really consider this, should've done more research instead of trusting tech youtubers.

Often times, these kind of problems cannot be solved. If you are using some Asus software to control your laptop, make sure it is set to the highest performance value. The Quiet setting often times will reduce the power limits like this.

The only Asus software I can think of when you mentioned this is probably the ROG gaming center. I've attached a picture with the default values below. Should I change anything here ?

The only other thing to try is removing the Intel DPTF driver. Brad shows how to do this.

I applied the registry before running cinebench, but I didn't find anything named "Intel(R) Dynamic Platform and Thermal Framework" in my devices manager prior to this.

For a baseline, run Cinebench with the core and cache at -100.6 mV like your screenshot shows. Now leave the cache offset voltage as is and start bumping only the core in steps of -25 mV. You might get your best results or best temperatures with the core up around -200 mV. If this is not game stable, reduce the core to -175 mV or -150 mV.

I've attached the benchmarks result below, and as you stated it's game stable at -175 mV. As for the log files, I did all the benchmarks at around 10 to 11 am.

Oh and I also attached my full specs using HWINFO that I somehow didn't upload on the first post.

Looking forward to reading your insights again unclewebb.

 

[unclewebb]

The log file looks better. No more dropping down to 17W. Were you logging data while Cinebench testing at full load? It will be interesting to see another log file while you are gaming. Sometimes when the Nvidia GPU is active, some computers will drop the power limit of the Intel CPU. I hope your computer does not have this "feature" (problem).

Basically your CPU goes to maximum power, the cooling system cannot handle this so it thermal throttles for a while until the 45W TDP power limit throttling takes over. This is not ideal but it is OK to run your computer like this. If you are ever bored, see if you can improve the cooling. Try a thorough cleaning including replacing the thermal paste. You will get smoother performance if the CPU is not constantly thermal throttling. Run some more log files and try to make the constant TEMP warnings go away. The thermal throttling might only be a problem when full load stress testing. While gaming, you might not have this thermal throttling problem.

You can try gaming at -200 mV on the core. Different voltages have improved your Cinebench scores. Most modern games use the AVX instructions in the CPU so different voltages can help games too. Some users continue to get better Cinebench results at -225 mV but are not game stable. The cache might still be stable at -120 mV or -125 mV. It never hurts to try another test. Just change one thing at a time and do some game testing. Everyone thinks you need to find the ideal voltage immediately. You are close to ideal as is. Best to make small changes and play some games for a day before adjusting further.

 

[Invertedstateofmind]

Hello unclewebb, thank you again for your reply;

Were you logging data while Cinebench testing at full load?

Run some more log files and try to make the constant TEMP warnings go away.

Am not really sure what you mean by these actually.

If you are ever bored, see if you can improve the cooling. Try a thorough cleaning including replacing the thermal paste. You will get smoother performance if the CPU is not constantly thermal throttling

I cleaned and repasted thermals just recently actually, around 5/6 days ago. I was getting really hot temps (95-97 celcius) after paying someone to clean it about a month ago. I asked for some guidance in the techsupport discord during the whole repasting process and they claimed that the cleaners did a horrible job on my laptop.

The thermal throttling might only be a problem when full load stress testing. While gaming, you might not have this thermal throttling problem.

I think you're right. After a three hour session of gaming ( League ), my temps were around 68 - 75 celcius and around 3.8 Mhz. I've attached the log files so you could take a look.

The cache might still be stable at -120 mV or -125 mV. It never hurts to try another test. Just change one thing at a time and do some game testing.

I'll post the logs for this as soon as I can.

Thanks again for your suggestions unclewebb

 

[unclewebb]

not really sure what you mean

Open the log file and look in the far right column. When a CPU is throttling, the reason will be displayed in the right column. This makes it easy for me to quickly scroll through a log file to look for any throttling problems.

   DATE       TIME    MULTI   C0%   CKMOD  BAT_mW  TEMP   NVIDIA GPU     VID   POWER
2021-08-04  10:30:20  35.92   99.8  100.0       0   95    1493    53   0.9810   56.1   TEMP
2021-08-04  10:30:21  35.88   99.9  100.0       0   95    1493    54   0.9573   56.0   TEMP
2021-08-04  10:30:22  35.92   99.8  100.0       0   95    1493    54   0.9813   56.3   TEMP
2021-08-04  10:30:23  35.86   99.8  100.0       0   95    1493    54   0.9814   56.0   TEMP
2021-08-04  10:30:24  35.78   99.8  100.0       0   95    1493    54   0.9813   55.9   TEMP
2021-08-04  10:30:25  35.58   99.8  100.0       0   94    1493    55   1.0167   55.2   TEMP
2021-08-04  10:30:26  35.51   99.8  100.0       0   95    1493    55   0.9568   54.9   TEMP
2021-08-04  10:30:27  35.57   99.8  100.0       0   95    1493    55   0.9580   54.9   TEMP
2021-08-04  10:30:28  35.52   99.8  100.0       0   95    1493    55   0.9563   55.0   TEMP
2021-08-04  10:30:29  35.64   99.8  100.0       0   95    1493    55   0.9806   55.3   TEMP
2021-08-04  10:30:30  35.61   99.7  100.0       0   95    1493    55   0.9871   55.2   TEMP

In the above example, the C0% is at 99.8%. That means your CPU is fully loaded. A Cinebench test will fully load all of your CPU cores. Most games do not.

The CPU is pinned at 95°C. The main screen of ThrottleStop shows PROCHOT 95°C. Your computer has been set to thermal throttle (PROCHOT - processor hot) at 95°C.

The far right column shows TEMP. That means your CPU temperature is too high. That is causing your CPU to thermal throttle and slow down. You need to avoid that.

Your latest log file shows that when you are gaming, the temperatures are much better. Gaming does not fully load the CPU like Cinebench does so your CPU is not overheating. Performance while gaming is great. Both the CPU and GPU are running at full speed. Hopefully your games are working well.

What thermal paste did you use? Some popular thermal pastes work very poorly when used in a laptop. Desktop computers have an integrated heat spreader over the top of the cores. Mobile CPUs do not. Some pastes can start to pump out in as little as a week. When this happens, your CPU temperatures will go up and up. At some point you might be forced to redo the thermal paste and you might have to try using something else. It is rare for anyone to do a great job the first time they replace the thermal paste. There is always room for improvement. This task is not suitable for someone with OCD.

Your laptop might just be poorly designed. There are lots of poorly designed laptops in the market these days. Some laptops use heatsinks that are barely adequate. When you try to run your CPU at full power, they choke and temperatures start bouncing off the thermal throttling temperature. The best paste in the world cannot fix poor design.

 

[Invertedstateofmind]

Hello again unclewebb,

Open the log file and look in the far right column. When a CPU is throttling, the reason will be displayed in the right column. This makes it easy for me to quickly scroll through a log file to look for any throttling problems.

I understand, I'll lookout for this to review the log files myself.

Your latest log file shows that when you are gaming, the temperatures are much better. Gaming does not fully load the CPU like Cinebench does so your CPU is not overheating. Performance while gaming is great. Both the CPU and GPU are running at full speed. Hopefully your games are working well.

Well yes, I've been hitting 120 fps in league consistently with these changes. Usually, it'll go very hot very quick and will drop fps to 80ish after like 10 mins. It's all thanks to your advice.

What thermal paste did you use? Some popular thermal pastes work very poorly when used in a laptop. Desktop computers have an integrated heat spreader over the top of the cores. Mobile CPUs do not. Some pastes can start to pump out in as little as a week. When this happens, your CPU temperatures will go up and up. At some point you might be forced to redo the thermal paste and you might have to try using something else. It is rare for anyone to do a great job the first time they replace the thermal paste. There is always room for improvement. This task is not suitable for someone with OCD.

I bought and used the Arctic MX4 Thermal Compound. I actually have no idea that there are laptop-specific thermal pastes.
Can you recommend me a good one so I'll replace it when the dreaded time comes?
And yeah, I feel like I did a mediocre job at repasting it. I have slight OCD, and getting that paste to be just right was straight up pain.

Thanks again unclewebb for your insights

 

[unclewebb]

Arctic MX4

I have heard a LOT of complaints about MX4 not lasting very long when used in some laptops. I have heard good things about Noctua NT-H2 but bad things about NT-H1. Here is a good example of what happens to some thermal pastes.

You will have to do some Google searching to try to find a good paste for your laptop that is available in your market. Most thermal paste reviews are almost always done on desktop CPUs. They will test 20 different pastes, install each one for 5 minutes, take some temperatures and call it a day. This kind of data is meaningless for laptops. Search laptop forums instead for user opinions.

Watch out for counterfeit pastes. Filling a tube with toothpaste is an easy way for some companies to make some extra cash.

It's all thanks to your advice.

You are welcome. Your last log file looked a lot better. Your CPU and GPU temps and performance look good. Play some games before worrying too much about your temperatures. If things go downhill in a week or two, now you know what to look for.

 

[Invertedstateofmind]

Hello unclewebb,

I have heard good things about Noctua NT-H2 but bad things about NT-H1.

Fortunately, Noctua NT-H2 is available in my local market. Gonna grab one when push comes to shove.

So after all this undervolting and all, I decided to give Apex Legends a try. I've attached the log file for it. I got mostly TEMP warnings but peculiarly some PL1 (before launching the game I pressume) on the side as well.
I'm currently on -175.08 mV for cores, -125 mV on cache and haven't changed anything on the TPL window. Do you have any advice on this? I'm afraid at that temperature, I can't play for prolonged hours on that game.

Looking to hear your insights again unclewebb.

 

[unclewebb]

Do you have any advice

It looks like when your CPU gets hot, sometimes it goes into a special low power mode. When the CPU is limited to 17W, it will throttle like crazy. This cools things down but it kills performance.

I think the DPTF driver was renamed by Intel. The original driver got a bad reputation because it sucked. Looks like the new driver sucks just as much. That is when a company calls in the marketing department to try to cover up the mess.

As I previously mentioned, your computer is never going to run properly if the log file is full of TEMP warning messages. Nothing has changed. Your computer is still running too hot.

 

[XT785]

I bought and used the Arctic MX4 Thermal Compound. I actually have no idea that there are laptop-specific thermal pastes.
Can you recommend me a good one so I'll replace it when the dreaded time comes?

Hi! MX-4 is considered a very bad paste for laptops. You can take a look at this forum for finding the best thermal paste for your laptop. Currently, SYY-157 and Thermalright TFX are the recommended pastes.

 

[Invertedstateofmind]

Hello Unclewebb,

Sorry for the inactivity, really busy lately.
So I decided to get a new SSD, and installed a fresh windows on it. And now I can't undevolt at all. The inside of the FIVR control was greyed out, I guess the shop gave me a plundervolt update. And to no luck till the time of writing, I haven't found an older version of my BIOS anywhere on the net. I'm really at wit's end here, really can't stand the laptop being so hot while gaming. Any insights or breakthoughs on this yet ? Or has Intel/Asus done anything on this ?

Hi! MX-4 is considered a very bad paste for laptops. You can take a look at this forum for finding the best thermal paste for your laptop. Currently, SYY-157 and Thermalright TFX are the recommended pastes.

Thanks for the recommendation mate, Thermalright TFX is available on my local market. Gonna snatch one when a sale pops up.

Edit 1: I tried downgrading my BIOS, FIVR control still grayed out.

 

[unclewebb]

has Intel/Asus done anything on this ?

Undervolting is a security vulnerability. Security has become more important than a laptop that is actually useable.

I tried downgrading my BIOS

With Asus laptops, you sometimes need to go back several BIOS versions before you get to one that is not locked. The one you need is probably not available on the Asus website.

Check out Notebook Review. Someone with your laptop model might have an old BIOS version available.

 

[Invertedstateofmind]

Hello unclewebb,

Undervolting is a security vulnerability. Security has become more important than a laptop that is actually useable.

Yea I know it's more of a security problem, but it's kinda of a bummer imo.

Check out Notebook Review. Someone with your laptop model might have an old BIOS version available.

I will go and take a look, thank you for the suggestion.

So I've been tinkering other ways to get my pc running cooler, I've tried :
1. Lowering my GPU frequencies in order to lower the temps ( from max of 1700 ish to 1404)
2. Lowering speedshift max to 33

I bought a thermalright FTX and it's on its way. Any other ways you can think of in order to at least make the laptop run not so hot ? I've attached the log files below for reference.

Looking forward to hear from you again unclewebb

 

[unclewebb]

Things are getting worse. The most recent log file shows constant thermal throttling even though power consumption is only 30W. Your CPU has a 45W TDP rating. Only being able to run it at two thirds of its rated power is definitely not good for performance. Hopefully some new paste will fix this problem.

Keep looking for an older BIOS so you can get CPU voltage control back. There has to be one out there somewhere.

 

[Invertedstateofmind]

Hello unclewebb,

The most recent log file shows constant thermal throttling even though power consumption is only 30W.

Yea, this was significantly worse. But that kinda happened because the gaming software ASUS (ROG gaming center) has wasn't installed yet, so the fans were working in a silent profile. I do hope the fans do some work on this
I'll edit and attach the log files to this post after today's gaming session.

Keep looking for an older BIOS so you can get CPU voltage control back. There has to be one out there somewhere.

I've checked notebook reviews and didn't find one. I'm gonna stay optimistic and try to find one in the ROG forums as well.

Looking forward to hear from you again unclewebb

Edit : I tried gaming with different values for speedshiftmax in this log file

 

[unclewebb]

The log file shows that the CPU and GPU are running at a consistent speed without any thermal throttling. Hopefully performance was more consistent.

Some laptops cannot be run continuously at their full rated speed. The heatsink and fan being used might not be capable of this even when the laptop was brand new.