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

Y545 - Severe Frame Drops companied with severe drops in Vram clocks

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
Hey , I had have this issue for about two weeks by now , I have a legion Y545 (I7-9750H , RTX 2060 , 512NVMe , 1TB hdd , 16Gb 2666Mhz) , and what is happening is I'm having severe frame drops in every game I play , like the frames drops from like 200 to like 40 or 50 accompanied with stuttering when the drop occurs , I monitored my system with Msi Afterburner and what I noticed is right when the drop occurs the Vram clock speeds drops from 7001Mhz to 800Mhz and sometimes stays like that for a while and then repeat itself over and over again every 3 to 8 seconds , the temps for the cpu is always above 90 and the gpu 83 at max (this is nothing unusual i've had this laptop for over 8 months by now and the temps always been like that without any drops or whatsoever) , what I tried so far is checking every power management thing on my laptop (checked Lenovo Vintage software and made sure it is on Performance mode and checked windows power management plans and also made sure it is on 100% and on high performance , and checked Nvidia control panel and made sure the power is on "prefer maximum performance") and also I tried uninstalling the gpu drivers with DDU and installed a fresh clean drivers from back before 2 months (at that time this issue never existed) also I tried resetting my pc and chose the keep files option and nothing changed the drops still occur , also I tried undervolting my cpu (just in case that the cpu might be affecting the gpu temperatures but the undervolt did nothing to the cpu temps they are still above 90C) I also tried the balanced and quite thermal modes in Lenovo Vintage Software and nothing changed . Any help would be much appreciated because this issue is really driving me insane I can't play any game normally without the drops accompanied with the stuttering happening.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,265 (1.26/day)
Try using ThrottleStop. In the Options window check Nvidia GPU. On the main screen check the Log File option. Now go play a game for 15 minutes so you have a record of your CPU and GPU performance. Exit the game and exit ThrottleStop so it can finalize your log file. The log file will be in your ThrottleStop / Logs folder. Attach it to your next post so I can have a look. Also post some screenshots of the ThrottleStop main window, the TPL window and the FIVR window.
 

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
Try using ThrottleStop. In the Options window check Nvidia GPU. On the main screen check the Log File option. Now go play a game for 15 minutes so you have a record of your CPU and GPU performance. Exit the game and exit ThrottleStop so it can finalize your log file. The log file will be in your ThrottleStop / Logs folder. Attach it to your next post so I can have a look. Also post some screenshots of the ThrottleStop main window, the TPL window and the FIVR window.
Alright I will do in a couple of hours, I appreciate your help .

Try using ThrottleStop. In the Options window check Nvidia GPU. On the main screen check the Log File option. Now go play a game for 15 minutes so you have a record of your CPU and GPU performance. Exit the game and exit ThrottleStop so it can finalize your log file. The log file will be in your ThrottleStop / Logs folder. Attach it to your next post so I can have a look. Also post some screenshots of the ThrottleStop main window, the TPL window and the FIVR window.
ok , I got the files , I need to mention that the drop happened at the end of this session and stayed like that (I was playing R6S) the vram frequency was at 810mhz when it should be 7000mhz and the gpu load was at 99% all time when the drop happened. and I need to mention that I did the undervolt for the cpu with Intel XTU.
 

Attachments

  • 2020-09-26.txt
    103.6 KB · Views: 92
  • TS.PNG
    TS.PNG
    140.1 KB · Views: 103
  • FIVR.PNG
    FIVR.PNG
    254.9 KB · Views: 97
  • Screenshot (354).png
    Screenshot (354).png
    297.2 KB · Views: 101
  • TPL.PNG
    TPL.PNG
    82.9 KB · Views: 96

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,265 (1.26/day)
ThrottleStop was able to read that your CPU is using a -110 mV offset undervolt for the core and the cache. The advantage of ThrottleStop over XTU is that ThrottleStop allows you to adjust the core and cache offsets individually. These two voltages do not have to be set equal. XTU forces you to set them equal but ThrottleStop does not. Many 9750H CPUs will show improvement with the core set as high as -200 mV and will still be 100% stable. Use Cinebench R20 for testing. Exit Intel XTU. Leave the cache at -110 mV and start increasing only the core offset in ThrottleStop. You should see either better temps or better Cinebench performance. This should also help many modern games.

The Intel default thermal throttling temperature for the 9750H is 100°C. Lenovo decided to low ball this so your CPU starts thermal throttling at 94°C. In the Options window, if you do not see a lock icon, this can be adjusted higher by reducing the PROCHOT Offset value. If you see the lock icon, leave this setting alone. The Intel default for PROCHOT Offset is 0, not 6.


When you look at the right side of your log file, do you see the word TEMP? Your CPU is continuously throttling and its high temperature is the cause of throttling. The CPU is being rapidly cycled hundreds of times per second between 800 MHz and approximately 4000 MHz. This alone is going to interfere with smooth game play. While this is going on, your laptop cannot deal with the extra CPU heat so the GPU temp is going up and up. It peaks at 85°C at which point it is probably not very happy either. Some laptop manufacturers set the GPU max temperature lower than what Nvidia recommends so 85°C might be the max GPU temperature for your Lenovo laptop.

If your laptop used to be great but is not great anymore, you need to clean and replace the thermal paste for both the CPU and GPU. The 9750H has a 45W TDP rating. Your heatsink and fan are struggling to prevent thermal throttling when the CPU is running at only 35W or less. That's a problem. You should never continuously be seeing TEMP in the ThrottleStop log file.
 

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
ThrottleStop was able to read that your CPU is using a -110 mV offset undervolt for the core and the cache. The advantage of ThrottleStop over XTU is that ThrottleStop allows you to adjust the core and cache offsets individually. These two voltages do not have to be set equal. XTU forces you to set them equal but ThrottleStop does not. Many 9750H CPUs will show improvement with the core set as high as -200 mV and will still be 100% stable. Use Cinebench R20 for testing. Exit Intel XTU. Leave the cache at -110 mV and start increasing only the core offset in ThrottleStop. You should see either better temps or better Cinebench performance. This should also help many modern games.

The Intel default thermal throttling temperature for the 9750H is 100°C. Lenovo decided to low ball this so your CPU starts thermal throttling at 94°C. In the Options window, if you do not see a lock icon, this can be adjusted higher by reducing the PROCHOT Offset value. If you see the lock icon, leave this setting alone. The Intel default for PROCHOT Offset is 0, not 6.


When you look at the right side of your log file, do you see the word TEMP? Your CPU is continuously throttling and its high temperature is the cause of throttling. The CPU is being rapidly cycled hundreds of times per second between 800 MHz and approximately 4000 MHz. This alone is going to interfere with smooth game play. While this is going on, your laptop cannot deal with the extra CPU heat so the GPU temp is going up and up. It peaks at 85°C at which point it is probably not very happy either. Some laptop manufacturers set the GPU max temperature lower than what Nvidia recommends so 85°C might be the max GPU temperature for your Lenovo laptop.

If your laptop used to be great but is not great anymore, you need to clean and replace the thermal paste for both the CPU and GPU. The 9750H has a 45W TDP rating. Your heatsink and fan are struggling to prevent thermal throttling when the CPU is running at only 35W or less. That's a problem. You should never continuously be seeing TEMP in the ThrottleStop log file.
I tried undervolting with throttle stop rather than XTU and I was able to reach as low as -130 mV while it being stable lower than that the laptop becomes unstable and shutdown but temperatures wise nothing has changed for the cpu , it is always above 90°C , and yes the laptop used to run smoothly and great even with such temperatures, I will try using throttle stop again , but if that doesn't help what you would suggest to do ?

ThrottleStop was able to read that your CPU is using a -110 mV offset undervolt for the core and the cache. The advantage of ThrottleStop over XTU is that ThrottleStop allows you to adjust the core and cache offsets individually. These two voltages do not have to be set equal. XTU forces you to set them equal but ThrottleStop does not. Many 9750H CPUs will show improvement with the core set as high as -200 mV and will still be 100% stable. Use Cinebench R20 for testing. Exit Intel XTU. Leave the cache at -110 mV and start increasing only the core offset in ThrottleStop. You should see either better temps or better Cinebench performance. This should also help many modern games.

The Intel default thermal throttling temperature for the 9750H is 100°C. Lenovo decided to low ball this so your CPU starts thermal throttling at 94°C. In the Options window, if you do not see a lock icon, this can be adjusted higher by reducing the PROCHOT Offset value. If you see the lock icon, leave this setting alone. The Intel default for PROCHOT Offset is 0, not 6.


When you look at the right side of your log file, do you see the word TEMP? Your CPU is continuously throttling and its high temperature is the cause of throttling. The CPU is being rapidly cycled hundreds of times per second between 800 MHz and approximately 4000 MHz. This alone is going to interfere with smooth game play. While this is going on, your laptop cannot deal with the extra CPU heat so the GPU temp is going up and up. It peaks at 85°C at which point it is probably not very happy either. Some laptop manufacturers set the GPU max temperature lower than what Nvidia recommends so 85°C might be the max GPU temperature for your Lenovo laptop.

If your laptop used to be great but is not great anymore, you need to clean and replace the thermal paste for both the CPU and GPU. The 9750H has a 45W TDP rating. Your heatsink and fan are struggling to prevent thermal throttling when the CPU is running at only 35W or less. That's a problem. You should never continuously be seeing TEMP in the ThrottleStop log file.
OK I tried what you suggested and undervolted the CPU with throttlestop and it made the temperatures better significantly but the performance dips in the GPU wise it didn't change I will attach an in-game screenshot when the drop happened and stayed like that. I really appreciate your help so far but now the problem is certainly a GPU not a CPU problem , right ?
 

Attachments

  • Tom Clancy's Rainbow Six  Siege Screenshot 2020.09.26 - 23.26.15.30.png
    Tom Clancy's Rainbow Six Siege Screenshot 2020.09.26 - 23.26.15.30.png
    2.5 MB · Views: 112
  • FIVR2.PNG
    FIVR2.PNG
    262.1 KB · Views: 83
  • Options.PNG
    Options.PNG
    172.3 KB · Views: 88
  • 2020-09-26.txt
    470.4 KB · Views: 66

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,265 (1.26/day)
I tried what you suggested
Not really. Your screenshot shows that you are still undervolting the CPU core and CPU cache to the same values. That is not what you want to do with these CPUs. Set the cache back to -110 mV and leave it there. Run Cinebench R20 and only adjust the core offset voltage lower. Keep lowering it by -20 mV at a time. Run Cinebench R20 again and again and watch for increases in performance or a decrease in temperature. Take a break from your game. This is the best way to adjust your CPU voltage.

The throttling flag now showing up in your log file is XPRO. To disable this, try clearing the BD PROCHOT box on the main ThrottleStop screen. This could damage your laptop so only try doing this if you are willing to assume the risk of any damage. BD PROCHOT means some other component within your laptop is running too hot so it is telling the CPU to thermal throttle.

The problem is your laptop is too hot. Are you planning to replace the thermal paste? Unless you do that, you are not going to solve your problem. I have no idea what temperatures your CPU and GPU were running at when your laptop was brand new. All I know is that your laptop is running too hot now.

If you want to set PROCHOT Offset to a different value or to its original value, you will need to reboot for the change to take effect.
 

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
Not really. Your screenshot shows that you are still undervolting the CPU core and CPU cache to the same values. That is not what you want to do with these CPUs. Set the cache back to -110 mV and leave it there. Run Cinebench R20 and only adjust the core offset voltage lower. Keep lowering it by -20 mV at a time. Run Cinebench R20 again and again and watch for increases in performance or a decrease in temperature. Take a break from your game. This is the best way to adjust your CPU voltage.

The throttling flag now showing up in your log file is XPRO. To disable this, try clearing the BD PROCHOT box on the main ThrottleStop screen. This could damage your laptop so only try doing this if you are willing to assume the risk of any damage. BD PROCHOT means some other component within your laptop is running too hot so it is telling the CPU to thermal throttle.

The problem is your laptop is too hot. Are you planning to replace the thermal paste? Unless you do that, you are not going to solve your problem. I have no idea what temperatures your CPU and GPU were running at when your laptop was brand new. All I know is that your laptop is running too hot now.

If you want to set PROCHOT Offset to a different value or to its original value, you will need to reboot for the change to take effect.
I tried disabling BD PROCHOT and set my core offset to -20~50 mV and the cache to -100 mV and started Cinebench and as soon as I started the temps went to 100C and my laptop turned off , do you have a discord or something ? , I really wanna solve this problem and so far you are the only one who is helping me , if that is ok with you ofc .

Not really. Your screenshot shows that you are still undervolting the CPU core and CPU cache to the same values. That is not what you want to do with these CPUs. Set the cache back to -110 mV and leave it there. Run Cinebench R20 and only adjust the core offset voltage lower. Keep lowering it by -20 mV at a time. Run Cinebench R20 again and again and watch for increases in performance or a decrease in temperature. Take a break from your game. This is the best way to adjust your CPU voltage.

The throttling flag now showing up in your log file is XPRO. To disable this, try clearing the BD PROCHOT box on the main ThrottleStop screen. This could damage your laptop so only try doing this if you are willing to assume the risk of any damage. BD PROCHOT means some other component within your laptop is running too hot so it is telling the CPU to thermal throttle.

The problem is your laptop is too hot. Are you planning to replace the thermal paste? Unless you do that, you are not going to solve your problem. I have no idea what temperatures your CPU and GPU were running at when your laptop was brand new. All I know is that your laptop is running too hot now.

If you want to set PROCHOT Offset to a different value or to its original value, you will need to reboot for the change to take effect.
I lowered the turbo speed midgame and made the cpu at a consistent 3000mhz with a stable temperature but then again the vram clock speeds dropped and the frames dropped again , what do you think? , I will attach the log file , just take a look at the end of it .(ps I even tried disabling the turbo completely with no avail)
 

Attachments

  • 2020-09-27.txt
    238.5 KB · Views: 83
  • turbo.PNG
    turbo.PNG
    142 KB · Views: 60
  • fivr4.PNG
    fivr4.PNG
    244.3 KB · Views: 60
Last edited:

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,265 (1.26/day)
For a baseline use
cache -110 mV core -110 mV

Now lower only the core
cache -110 mV and core -130 mV
cache -110 mV and core -150 mV
cache -110 mV and core -170 mV
cache -110 mV and core -190 mV
cache -110 mV and core -210 mV

Run Cinebench each time you lower the core voltage. Also run a 1 or 2 Thread TS Bench test each time you lower the voltage. No need to run TS Bench for a long time. One or two minutes of light load testing is enough when starting out. If Cinebench can complete without errors then your laptop must be reasonably stable.

Once you are finished adjusting, then you can run some longer TS Bench tests or whatever test your like.

do you have a discord or something ?
I do not use Discord. I spend too much time in the forums as it is. If I start helping users on Discord, I can kiss my life good bye.
Time to go watch some football.
 

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
I do not use Discord. I spend too much time in the forums as it is. If I start helping users on Discord, I can kiss my life good bye.
Time to go watch some football.
Alright , I will try this although I already got a better performance with Cinebench and better temps just adjusting the turbo values , but I will try this too , I appreciate your help tho , much thanks and sorry for any disturbance.
 

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
Hey , in case you are still here , as I mentioed I got my cpu to the best performance I could , I increased my Cinebench score from 1800 to 2550 , but the problem of the Gpu Vram downclocking itself still persists , if you have anything in mind I would be glad to hear , as I mentioned you are the only who is helping me so far , I hope I'm not causing any disturbance for you but this problem is driving me insane I can't play game without it causing frame drops accompanied with stuttering, much thanks in advance.
[Edit] I attached an in game screenshot when the drop happened
 

Attachments

  • Tom Clancy's Rainbow Six  Siege Screenshot 2020.09.28 - 18.41.37.56.png
    Tom Clancy's Rainbow Six Siege Screenshot 2020.09.28 - 18.41.37.56.png
    2 MB · Views: 76

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,265 (1.26/day)
I increased my Cinebench score from 1800 to 2550
That is much better but a properly cooled 9750H should be over 3000 points during Cinebench R20. Your CPU is under performing because of bad cooling. The CPU is too hot. If you cannot fix the cooling then it will never run as it should.

The GPU VRAM throttling from 7000 MHz to 800 MHz is not normal and is the main cause of your poor gaming performance. The extra heat in your laptop might be causing the GPU VRAM to slow down. Normal throttling will slow a component down, wait for it to cool down and then it will try to go back up to full speed. This cycle will repeat over and over again.

It sounds like your VRAM is throttling, likely due to heat. The excessive heat in your laptop might have damaged the VRAM. Laptops that run at over 90°C for hours might be OK when brand new but over time, any electronic component can fail at high temperatures. If your laptop is still under warranty, you will have to convince Lenovo to fix it.
 

AliAlkazrji

New Member
Joined
Sep 26, 2020
Messages
7 (0.01/day)
That is much better but a properly cooled 9750H should be over 3000 points during Cinebench R20. Your CPU is under performing because of bad cooling. The CPU is too hot. If you cannot fix the cooling then it will never run as it should.

The GPU VRAM throttling from 7000 MHz to 800 MHz is not normal and is the main cause of your poor gaming performance. The extra heat in your laptop might be causing the GPU VRAM to slow down. Normal throttling will slow a component down, wait for it to cool down and then it will try to go back up to full speed. This cycle will repeat over and over again.

It sounds like your VRAM is throttling, likely due to heat. The excessive heat in your laptop might have damaged the VRAM. Laptops that run at over 90°C for hours might be OK when brand new but over time, any electronic component can fail at high temperatures. If your laptop is still under warranty, you will have to convince Lenovo to fix it.
I might have found the problem , I changed the room that I was in (to a much better cooled room) , the gpu temps never exceeded 76C , I'm gaming for an hour by now normally it appears that the Nvidia Gpu downclock itself severely when it exceeds 80C , I'm going to change the thermal pastes for both the gpu and cpu , since now it is clearly a thermal issue , and for the cpu unless I find a really good thermal solution for it (the options I have is either liquid metal or a thermal pad which I doubt it would do much) other than that , indeed my cpu would never be able to target its rated performance . Again much thanks with your help on the cpu it is already better , and sorry if I disturbed you .
 
Top