The ThrottleStop Guide

Discussion in 'Hardware Components and Aftermarket Upgrades' started by unclewebb, Nov 7, 2010.

  1. Avé César

    Avé César Notebook Consultant

    Reputations:
    23
    Messages:
    135
    Likes Received:
    74
    Trophy Points:
    41
    Thank you so much for those explaination, i was scared about a hardware failure.

    Last summer the laptop get hotter (of course some days here were hotter than others days) specially when i was playing CoD Warzone, cpu reache end 80's, to 90° so i had to slow, tweak, the turbo ratio limits and Speed Shift EPP to 93. I will test some other games with my config for now and compare if it reaches 84° max on some cores as before like in BF1. I had remove CoD warzone cause of those 30Go of patch updates, so annoying to reinstall half the game each time so can't compare on this one ^ ^

    For the cache , core data, i still don't understand why it's a good point, on multi threads, i thought more core, threads, is better, faster

    Edit : I played a bit of BF1 and max temp was 79° on several cores. I'm on to update throttlestop 9.2 ( finger cross )
     
    Last edited: Nov 25, 2020
  2. Jdpurvis

    Jdpurvis Notebook Evangelist

    Reputations:
    103
    Messages:
    457
    Likes Received:
    142
    Trophy Points:
    56
    Intel processors are designed to tolerate temperatures up to 100 degrees C - at that temperature, they will protect themselves by shutting down. You could certainly allow temps into the mid-90's under heavy stress without worrying about damage to your CPU. Undervolting with TS can certainly help with temps, and allow better performance - and it seems you have made some good progress with repasting and cleaning. You are on the right track - keep going.

    Best,

    Joe
     
    Avé César likes this.
  3. unclewebb

    unclewebb ThrottleStop Author

    Reputations:
    7,407
    Messages:
    6,157
    Likes Received:
    5,957
    Trophy Points:
    681
    I do not have any updated documentation but what Intel used to do is they would set the thermal throttling temperature (PROCHOT) to 100°C and then they would set the thermal shutdown temperature (THERMTRIP) 25°C to 30°C higher than that. The purpose of this was to provide a user with some headroom and some time to save their work. If thermal throttling was not enough to control the CPU temperature, only then would the CPU be forced to shutdown to protect itself.

    Thermal throttling works great. There would need to be a very serious problem like a loose heatsink before temperatures got completely out of control triggering a shutdown.

    [​IMG]

    The problem is that some cowardly laptop manufacturers decided to ignore Intel's advice and created their own throttling schemes. They reduced the temperature so thermal throttling begins at 95°C and then they prematurely force a shutdown at 100°C. It was never Intel's intention for their CPUs to be shutting down at such a low temperature.
     
    Papusan, t456 and Avé César like this.
  4. Avé César

    Avé César Notebook Consultant

    Reputations:
    23
    Messages:
    135
    Likes Received:
    74
    Trophy Points:
    41
    So i did update throttlestop and it looks, getting, better and better Unclewebb, thanks a lot for all your hardworks, sharing this to the people. I love the options to the Typo of the Main Interface,the Lock prochot option, some custom on the logo added + colors and typo. The "C10"
    tab looks better too.
    I report a little bug, different temp from the main board and the notification on min and max temp.

    In battlefield 1, 79° max today on 4 cores before was 84, but don't know why on Elders scroll online, i got big jump, i mean lowest temp compare to last monday 84° to 73° just right now...

    Edit, i forgot to say, when i delete the ini file, Switch off laptop, switch on to clear throttlestop before delete all file and update it, i wanted to perform a BIOS tweak reset too but i found my tweaks in it was reset too. Don't know if it's related to the "reset" for throttlestop ?! or if the bios reset when i did my maintenance on my laptop.
     

    Attached Files:

    Last edited: Nov 25, 2020
    unclewebb likes this.
  5. redmop

    redmop Notebook Consultant

    Reputations:
    13
    Messages:
    129
    Likes Received:
    28
    Trophy Points:
    41
    How can I verify that the undervolt is actually taking affect. I can set my undervolt to - 300 cpu and it runs just fine. The undervolt does show up in HWInfo.

    Alienware m17 r3 1.5.0 bios Windows 10.
     
  6. unclewebb

    unclewebb ThrottleStop Author

    Reputations:
    7,407
    Messages:
    6,157
    Likes Received:
    5,957
    Trophy Points:
    681
    The maximum package temperature is exactly the same.

    [​IMG]

    It looks like the Min data reported in the system tray is the minimum package temperature. It is possible to have individual cores running at a cooler temperature than the package. The Max data is the important data one should be looking at. The Max 65°C data in the button is the maximum Nvidia GPU temperature.

    @remop - Are you undervolting both the CPU core and CPU cache by -300 mV? Probably not or your CPU would crash. Post a FIVR screenshot. It has a monitoring table that will show the voltages the CPU is using. For a 10th Gen CPU, I would start testing with the cache at -75 mV and the core at -150 mV. Download Cinebench R20 and go from there.

    https://www.techpowerup.com/download/maxon-cinebench/

    Also run some light load TS Bench tests, 1 or 2 Thread, to sniff out any cache instability.
     
    Last edited: Nov 25, 2020
    Avé César likes this.
  7. Krzyslaw

    Krzyslaw Notebook Consultant

    Reputations:
    45
    Messages:
    238
    Likes Received:
    162
    Trophy Points:
    56
    On 10th gen I would start at even at -40 on cache for example my 8 core can only do -50mV on cache with ac dc Loadline at 30. This is my max UV that will not produce any errors in TS bench. And than my 10875h FULL loaded have around 1.106V with all core at 4.3GHz which gave me around 100W power consumption
     
    unclewebb likes this.
  8. redmop

    redmop Notebook Consultant

    Reputations:
    13
    Messages:
    129
    Likes Received:
    28
    Trophy Points:
    41
    -250.0 mV
    upload_2020-11-26_15-43-28.png

    -1000.0 mV
    upload_2020-11-26_15-44-27.png

    I9-10980HK

    Am I a victim of the plundervolt patch?
     

    Attached Files:

  9. Krzyslaw

    Krzyslaw Notebook Consultant

    Reputations:
    45
    Messages:
    238
    Likes Received:
    162
    Trophy Points:
    56
    Just red your posts. Probably Yes.
    Will write you tomorrow morning as it is midnight now in Poland and I need to sleep now. I Will write in about 9 to 10hours

    For now disable all UV and go to the bios. Use key combo to unlock hiden settings.
    Find following settings
    Overclocking lock, and set disable and any other locks that are enable under cpu settings or under security tab
    Find Xtu interface and set enable
    Voltage optimisation and set enable
    Sgx disable
    In addition Disable speedstep and enable speedshift

    Try again with UV in TS

    BTW if you have rtx2080super could you write me version of your vbios and share a dump of it

    EDIT:
    OH MAN, I tought I was at msi subforum, do you have even access to the options I have mentioned in bios?
     
    Last edited: Nov 27, 2020
  10. unclewebb

    unclewebb ThrottleStop Author

    Reputations:
    7,407
    Messages:
    6,157
    Likes Received:
    5,957
    Trophy Points:
    681
    Why would you think that?

    When voltage control is disabled, the turbo ratio adjusters will also be locked and disabled. Your turbo ratio adjusters are not locked and your voltages are showing up in the monitoring table. ThrottleStop FIVR will report when your CPU is Locked.

    [​IMG]

    When you set a ridiculous undervolt of -1000 mV, the CPU will ignore the vast majority of that request. Anything beyond about -225 mV or -250 mV is automatically ignored. If you undervolt only the core to -250 mV or -500 mV or -1000 mV; the CPU will use -250 mV or slightly less for each request.

    Do some Cinebench R20 testing.
    https://www.techpowerup.com/download/maxon-cinebench/

    Start with the core and cache set equally to -100 mV for a baseline. Also do some light load 1 and 2 Thread TS Bench tests. Make sure this test is not reporting any errors. Once you have a baseline score, start bumping the core up by -25 mV. Stop increasing this when you stop seeing performance or temperature improvements in Cinebench R20. At some point, increasing the core offset will make no difference to your results.

    When you were at -300 mV the other day, you had already gone too far.
     
    Last edited: Nov 27, 2020
Loading...

Share This Page