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

I7 7700HQ Undervolting Proteus V, need some help.

Accutype

New Member
Joined
Nov 24, 2020
Messages
6 (0.00/day)
Hello everyone, how are you all

So this is my first post here, but i've been around for more than a year looking for any type of guide, cooling, overclock, advices, etc... And i have a problem. I was using Intel XTU app to undervolt my Proteus V since 3 years ago or so, and some months ago i noticed that it wasnt showing my temps on the core anymore. I didnt give it much credit cause i was able to check my core temps with a native app from pc specialist. Till yesterday. I was noticing some high temps and high sound. When i undervolted it, i always put -115 in XTU to make it stable and with good temps, as i will show in a capture, and i know 7700hq is kind of locked so XTU wont work anymore ( but i mention it now, yes). So i've been running some tests and got to undervolt it to -144.5 Core&Cache, and i ran P95 and checked RealTemp too, noticing that only appears to me Core0-1-2 ( no idea why, on HWMonitor and RealTemp) but, the thing is that my Core1 goes to 95Cº while my Core 0 and Core 2 goes only to 67-70? kind of. Btw My temps with XTU are after i repasted time ago with Liquid Metal, and it was all going good. And some months ago ( 4-5?) i repasted again the liquid metal, cause i was cleaning it and noticed that my liquid metal was kind of dissapeared, or at least a little bit of it after 2 years without opening it. So Things i need to know, if someone can help me:
1. I dont need performance at all, i am okay when it was working 3.4 or so in XTU without any problem, but at 60-70 Cº, thats what i liked, low fan noise, and everything was good.
2. Maybe i have to check how i applied my liquid metal last time? Cause i cant understand why only Core1 Temps goes to 95Cº max while Core 0&2 are pretty nice at +-70.
3. Any advice on the undervolting thing, maybe undervolting igpu or anything that could make it perform any better than right now? Again, i dont want every core to go 3.8ghz, im ok with them running at 3.4, but i dont want my llaptop to be +90 Cº under heavy load like it came from fabric, when it was hitting +90 before the first underclock.


Sorry if there is any grammar mistake, not my native language but i try my best. Hope anyone can help me reach some conclusions. Thanks everybody in this community

PS: First time i undervolted yesterday with ThrottleStop Limit Rin EDP Other was running red all the time when running TS Bench or P95, then appeared PL1, and now it is just yellow, so at least i think i did something good.



PS2: I dont know how, but it seems that there was a hiden core or something, i got to see 4 cores now, something happened with windows start or i dont know what, that was only showing 3 cores, i had to restart with 1 core and then with deattached option of number of processors, and now hwmonitor and everything shows Core3, the missing one.
 

Attachments

  • -115.png
    -115.png
    41.3 KB · Views: 261
  • FIVR-144.5.PNG
    FIVR-144.5.PNG
    245.2 KB · Views: 254
  • ThrottleStopConfig.PNG
    ThrottleStopConfig.PNG
    150.7 KB · Views: 336
  • SensorTest-144.5.PNG
    SensorTest-144.5.PNG
    119.6 KB · Views: 219
  • CPUMonitor-144.5.PNG
    CPUMonitor-144.5.PNG
    238.3 KB · Views: 201
  • TPL.PNG
    TPL.PNG
    82.2 KB · Views: 235
Last edited:
Joined
May 7, 2020
Messages
244 (0.17/day)
Such huge differences in cores means only one thing - TIM (paste/LM) gone bad, vary bad. Just reapply them.
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,271 (1.26/day)
It does not look good when multiple programs cannot read all 4 cores of your CPU. One core might be dead. Perhaps the liquid metal has shorted something out.

Try running CPU-Z and see what it reports for Cores and Threads.

1606239172372.png
 

Accutype

New Member
Joined
Nov 24, 2020
Messages
6 (0.00/day)
Sorry maybe i put too late the Postdata2, i found that lost core, it seems somehow it was hidding or something, after restarting windows with msconfig start number of cores 1, and then restarting with that square unattached of number of cores, my lost core appeared, and im now fine again. But the thing is that Core 1 is still showing huge difference over those other 3 cores, 95Cº or so vs +- 70 on heavy loads ( TS Bench, P95 ). Tomorrow ill try to repaste it, maybe i should just forget about the liquid metal and go with a thermal paste if the difference is not going to be that huge tho... Thanks for the answers and ill update tomorrow with what i get, if you have any specific test you wanna see, or any kind of config in ThrottleStop just tell me it here. Thanks everybody.
 
Joined
Feb 18, 2012
Messages
2,715 (0.61/day)
System Name MSI GP76
Processor intel i7 11800h
Cooling 2 laptop fans
Memory 32gb of 3000mhz DDR4
Video Card(s) Nvidia 3070
Storage x2 PNY 8tb cs2130 m.2 SSD--16tb of space
Display(s) 17.3" IPS 1920x1080 240Hz
Power Supply 280w laptop power supply
Mouse Logitech m705
Keyboard laptop keyboard
Software lots of movies and Windows 10 with win 7 shell
Benchmark Scores Good enough for me
Maybe the liquid metal compound dried out. I've read stories of liquid metal drying out and becoming harder.
Its never happened to me, ive been using liquid metal on and off for about 8 years.
 

Accutype

New Member
Joined
Nov 24, 2020
Messages
6 (0.00/day)
Okay guys, coming with news, just finished repasting and cleaning the pc and doing some testings.
First of all, ill post 1 FullView Cinebench (After a p95 passed) + Temps and ThrottleStop Limits and temps, where you can see it was doing something in ring edp and PL1( I must say that when i did the cinebench after the p95 test, limits where red, not yellow, but couldnt make a screen in that moment of cinebench, and they were red only in Multi test, on single core only yellow, if this gives you any info, cause i dont know a sht) all of this BEFORE the repasate yesterday 26. Also i'll add the log file.

And then ill post the testings from today, same again P95 then Cinebench with temps, log file, throttlestop showing Limits (But noone of them appeared), the really nice temps i got now and all of the above but i did it right now just after repasting and that,

Dont know why yesterday the vcore and cache didnt save my -0.145 offset, and i did all of this with only -.107 as you can see in HW Monitor. And the most important things, temps of core 1 are now like the rest of the cores, so everytings seems normal. But now i will need some advices about undervolting it, cause i dont know how, but after repaste cinebench test result was worse on single and multi than before xD Dont ask me.


Screens:
 

Attachments

  • CinebenchAntesDeRepasteo.PNG
    CinebenchAntesDeRepasteo.PNG
    2.3 MB · Views: 102
  • PreRepasteLOG.txt
    337.7 KB · Views: 84
  • CinebenchDespues.PNG
    CinebenchDespues.PNG
    2.3 MB · Views: 90
  • POSTRePaste.txt
    223.5 KB · Views: 79

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,271 (1.26/day)
Cinebench test result was worse
Your log file after the re-paste shows some serious Clock Modulation throttling. This means internally the CPU is processing data at a small fraction of its rated speed.

Look at the CKMOD column in the log file. This column should always be at 100.0%. Your example shows this column dropping down to 12.5%. Your CPU is running like a slug. Before this happens, your multiplier is being limited to 28. That means turbo boost is also disabled. More throttling. Did you do that on purpose or did your computer punch you in the groin?

Exit RealTemp and exit HWMonitor. ThrottleStop can monitor your CPU temperature and speed. HWMonitor does not report the correct voltages so I would not use that.

Set the Speed Shift EPP variable to 80.

You need to check the ThrottleStop - Clock Mod box and set that to 100.0 to try to counter the clock modulation throttling that is happening. This is very unusual. Very rare for a 7th Gen computer to use this throttling method. Most manufacturers abandoned this method a long time ago. The Mod column in the monitoring table on the main ThrottleStop screen should show 100.0 from top to bottom.

If you used ThrottleStop to change your voltages and your voltages did not get saved, make sure you have this option in the FIVR window set correctly.

1606447689602.png


After you get things fixed up, run another Cinebench test with the Log File option checked. Hopefully this will help you find your lost performance.
 

Accutype

New Member
Joined
Nov 24, 2020
Messages
6 (0.00/day)
Your log file after the re-paste shows some serious Clock Modulation throttling. This means internally the CPU is processing data at a small fraction of its rated speed.

Look at the CKMOD column in the log file. This column should always be at 100.0%. Your example shows this column dropping down to 12.5%. Your CPU is running like a slug. Before this happens, your multiplier is being limited to 28. That means turbo boost is also disabled. More throttling. Did you do that on purpose or did your computer punch you in the groin?

Exit RealTemp and exit HWMonitor. ThrottleStop can monitor your CPU temperature and speed. HWMonitor does not report the correct voltages so I would not use that.

Set the Speed Shift EPP variable to 80.

You need to check the ThrottleStop - Clock Mod box and set that to 100.0 to try to counter the clock modulation throttling that is happening. This is very unusual. Very rare for a 7th Gen computer to use this throttling method. Most manufacturers abandoned this method a long time ago. The Mod column in the monitoring table on the main ThrottleStop screen should show 100.0 from top to bottom.

If you used ThrottleStop to change your voltages and your voltages did not get saved, make sure you have this option in the FIVR window set correctly.

View attachment 177173

After you get things fixed up, run another Cinebench test with the Log File option checked. Hopefully this will help you find your lost performance.
But this only happens when p95 or cinebench starts tho, right now for example im not running any test and it is stable at 100.0 (Mod column). When p95 starts, it goes from 100.0 to 12.5 or idle in 12.5 jumps, thats why you see 100.0, 87.5, 75.0, 67.5, etc... And Multiplier is like that since i opened ThrottleStop for the very first time, but it is unchecked, so i was assuming it wasnt blocking it at 27, but no idea. Im going to run that cinebench now with speedshift at 80. Also thats the config on FIVR im running right now, if you see i put 4 cores on 36, but it will never go over 3400MHz, dont know why.

First sorry for double post, dont know if i should edit and make a postdata on last post, or it doesnt mind.

Ok so in cinebench something strange happened. First of all, it didnt ever go above 3000 Mhz, dont know why. Second, EDP and PL1 came back on red, dont know why. And the cinebench results where even worse than yesterday, not even 4000 points... ? no idea.

Screen 1, TS showing 100.0 Mod, 3000 MHz ( never went above that) and Limits on red EDP and PL1.

Cinebench Screen showing only multicore cause it was enough to show something happend that i went down to 3800 points, not even 4000.

Log FIle. You have to see the log file at 10:30 or so i think, the beginning of the log file is last night.


The configuration used was showed in last post.
--------------------------------------------------------------------------------------------
PD3?:

Okay sorry for this big post, i got news. I was running some cinebench tesst with some config i read of unclewebb in other forum of a 9750H, i check disable and lock turbo, i was fine, then i saww you saying that to lowering the core offset while a test, and i reached -190 ( Afraid of going lower, i saw a comment from other i7 7700HQ that was on -200 and cache at -60) While i was lowering this, i saw that, at the begining ( -145) i had +-2.900-3000 Mhz, and as i was lowering core it went higher, to 3020 or 30 maybe? Then i started lowering cache and everything was going nice, till i reached -170 ( 3160 Mhz i think, cant know right now) and it shutted down. When i restarted the pc, opened throttle and couldnt opened and shutted down again. That like 4-5 times, i deleted TS and downloaded again cause thoses offsets were saved and every time i was opening TS and shutting down all the time. Now that i downloaded again. I tried last stable offset while i was running cinebench ( BTW Mhz were going up, but edp and PL1 on red was being showed yet), so when i tried to put thoses offsets i was able to get before the first shutting down, i put -190 core and -165 cache ( which was stable on cinebench at 3160 3180 MHz) and it shutted down again. Now im with -190 core and -145 cache ( The first offset i got that was stable) and something changed, cause on a TS Bench it goes red on limits in Core PL1 and Ring EDP and PL1. Config is set as last time before the first shut down. And when showing EDp PL1 and PL1, at the begining of the test it goes to 3.4 Mhz and then when those limits appears it goes down to 2.8 and doesnt go higher. Stable temps at 55-57 Cº BTW, so no idea whats happening.

Now when ever i test at cinebench it goes to 3400 and then at 10 secs or so it goes down to 2790 Mhz and showing limits red on Core PL1 and RING PL1 and EDP. Thats the latests news. Im going to stop testing and wait for your answers.
 

Attachments

  • FIVR.PNG
    FIVR.PNG
    242.1 KB · Views: 114
  • TPL.PNG
    TPL.PNG
    82 KB · Views: 123
  • TS.PNG
    TS.PNG
    136.7 KB · Views: 133
  • DuringCINEBENCH.PNG
    DuringCINEBENCH.PNG
    42.9 KB · Views: 178
  • CinebenchonRed.PNG
    CinebenchonRed.PNG
    2 MB · Views: 119
  • 2020-11-27.txt
    593.4 KB · Views: 68
Last edited:

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,271 (1.26/day)
red on Core PL1
This indicates power limit throttling. If you look in the log file you will see that this is happening when power consumption is at 23 Watts. The 7700HQ has a 45W TDP rating. That means it should be able to run continuously at 45 Watts. Your laptop has been designed to throttle and slow down at half that power level. Make sure you are not using any cool and quiet software on your computer. These kind of programs can force a low power limit and result in constant power limit throttling.

To try and prevent this throttling, check the FIVR - Disable and Lock Turbo Power Limits box. This can help some laptops but not all. In the TPL window, go to the Turbo Boost Power Limits section at the top and check the Lock option. Press OK. Do another test and see if this helps keep your CPU from throttling below 45W.

It might not help. To be blunt, I am not impressed. Your laptop seems like it was purposely designed to underperform. Nice specs but they are meaningless if your computer constantly throttles.

When p95 starts, it goes from 100.0 to 12.5
A properly functioning laptop is not supposed to do this. It should only use clock modulation throttling after the CPU temperature goes over 100°C. It should never, ever use this during normal use. I understand that Cinebench and Prime95 are stressful tests. A properly designed computer should not be using clock modulation throttling during a stress test.

When your computer is using clock modulation at 12.5%, the CPU is running internally at about 300 MHz. I have a 14 year old laptop collecting dust in my closet that can run way faster than that. Check the Clock Mod box in ThrottleStop to prevent your computer from using this unnecessary throttling method. Intel CPUs will slow down if they get too hot. Manufacturers do not need to use multiple throttling methods.

1606500961895.png
 

Accutype

New Member
Joined
Nov 24, 2020
Messages
6 (0.00/day)
Okay okay... Final result is here. First of all, thanks everybody that answered and helped and @unclewebb you are a beast.

So the results are here. Everything worked fine, stable 3400 mhz during all of the test, max temps at 65Cº, +- 0.86v? Cant remember. No alerts on limits, everything working as it should be since the begining. Im really ok with this results, also cinebench scored 4500 on multi, 300+ points than expected so yeah, im fine. Maybe i could push a little bit more, but while i was on it, i had to restart 4 times due to not enough V on Cache and maybe Core too, i got to -192 core and -157 on cache as you can see. Everything seems to be fine and im really glad about the temps.

Wonderful forum and woderfull people here, hope to learn more and more for the next months around this forum. Thanks for all.
 

Attachments

  • 2020-11-29.txt
    105.9 KB · Views: 70
  • AfterAllFinal.PNG
    AfterAllFinal.PNG
    2.3 MB · Views: 128

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,271 (1.26/day)
you are a beast.
You are welcome. I always like trying to get to the bottom of a throttling issue. Kind of like a dog working over a bone!

When the Clock Mod box is checked, you need to also click on the Turn On button so ThrottleStop can monitor and correct for this type of throttling. Your previous screenshots showed that you were already doing this so I did not mention it. Your last screenshot shows that you did not do this. In the ThrottleStop title bar at the top, make sure it does not say Monitoring.

If you look in the log file at the C0% column, you will see that occasionally there are some dips that last for 4 or 5 seconds. This might be a sign of some clock modulation throttling going on with ThrottleStop rapidly trying to correct this issue. When the log file was run, I will have to assume that you clicked on Turn On and you were not in Monitoring mode at the time.

Your log file looks a lot better. Lots of 34.00 multipliers which is the multiplier your 7700HQ should be running when 4 cores are active. The FIVR window shows the default multipliers for your CPU are 38, 36, 35, 34. It also shows that you are requesting these be set to 38, 36, 36, 35. This is not going to do anything. You have a locked CPU that cannot be overclocked so the turbo ratios should be set to the default values. I am not sure if you changed these deliberately or maybe the BIOS did not set these correctly. Not too important. The CPU will ignore the requested values and will use the default 34 multiplier when 4 cores are active.

The log file shows consistent performance and great temps. What more could one ask for?
 

Accutype

New Member
Joined
Nov 24, 2020
Messages
6 (0.00/day)
About the C0%, it might be that in cinebench when the screen goes black after a render test, it gives like a little stop or like a little rest or nap, i can see that happening in temps going down, noise and also in the FID so im ok with it. The multipliers yes, when i changed them i thought oh sht, my processor is locked, and as it wasnt happening anyhting, i just forgot to change them to normal xD

And Clock Mod i didnt turned it on along the tests, so i have no idea if it did anything or not. I turned it on right now, gona save screens about it all and lets see if my repaste works for more than 6 months or so.
¡Thank you very much!
 

unclewebb

ThrottleStop & RealTemp Author
Joined
Jun 1, 2008
Messages
7,271 (1.26/day)
temps going down
When a CPU uses clock modulation throttling, you will see a drop in C0% and a drop in temperatures. I am not sure what the exact trigger is in your laptop that starts this type of throttling. Best to keep Clock Mod checked to try to minimize the impact of this design defect.

When clock modulation throttling is in progress and ThrottleStop is setup to correct this, you might not see the Mod column value change in the monitoring table. ThrottleStop is trying to correct this issue as fast as your computer is trying to set clock modulation to a value lower than 100.0.
 
Top