Jump to content

Help with undervolting

Go to solution Solved by unclewebb,

@Frank Cisco

You need to undervolt the core and the cache. Start more conservatively. Do some testing at -100 mV or -75 mV for both. Try running the TS Bench 960M test. This is a good first test to check for stability. If it reports any errors, the CPU needs more voltage.

Hi , how are you, I just need advice , I did an undervolt to my 8700k , and I don't know of I did it right but I started benching using cinebench and hwifo and undervolting with throttlestop , started with -60mv and kept note of the scores of cinebench, then after the 10 min tests , undervolt by -10 mv until I reached a low score for about 50 points that was at -1.77mv and just to be sure I added another -10 mv and yes , the score went down, I was amazed that the CPU never crashed neither the cinebench program, so I ended puting back the number of mv that gave me the best results in the test, and started playing with +- 1mv until I noticed that the scores where stable at -1.63, after this I passed the 30 min stability test, so I proceeded to create a scheduled task for throttlestop, in order to start with windows. I know that it's better to make that change in the bios. So what I'm asking is , if you can help me,  what do I need to change?  because I really don't know where and what settings I need to check in order to put the values in the bios, my mobo is an Asus ROG. Thanks .

Link to comment
Share on other sites

Link to post
Share on other sites

4 minutes ago, Frank Cisco said:

Hi , how are you, I just need advice , I did an undervolt to my 8700k , and I don't know of I did it right but I started benching using cinebench and hwifo and undervolting with throttlestop , started with -60mv and kept note of the scores of cinebench, then after the 10 min tests , undervolt by -10 mv until I reached a low score for about 50 points that was at -1.77mv and just to be sure I added another -10 mv and yes , the score went down, I was amazed that the CPU never crashed neither the cinebench program, so I ended puting back the number of mv that gave me the best results in the test, and started playing with +- 1mv until I noticed that the scores where stable at -1.63, after this I passed the 30 min stability test, so I proceeded to create a scheduled task for throttlestop, in order to start with windows. I know that it's better to make that change in the bios. So what I'm asking is , if you can help me,  what do I need to change?  because I really don't know where and what settings I need to check in order to put the values in the bios, my mobo is an Asus ROG. Thanks .

I had a 9900k that ran quite hot during summer and I undervolted it. The only thing I did was lower the core voltage from 1.37 to 1.25v in the BIOS and it was stable, is it most optimal way of doing it.. Probably not, is it the fastest way, yep!  I went from 85c temps to around 60c in games, incredible.

Link to comment
Share on other sites

Link to post
Share on other sites

What it sounds like you're experiencing is called clock stretching, where if the CPU will automatically underclock itself if the voltage drops too much and not report it to software. Usually, the solution to avoid this is to set up a manual "overclock," where you set the clock speed manually to whatever you want it to be (say 4.8GHz), set the core voltage to manual mode and try to minimize the voltage that way. That said, if you don't have a Z- series motherboard, that method won't really work. 

Link to comment
Share on other sites

Link to post
Share on other sites

My 8700k did

4.7 at 1.2v

4.8 at 1.24v

4.9 at 1.28

5 at 1.32v

and 5.1 at 1.4v all at llc lv6 

 

r23 isn’t a good stress test really as u can pass that many times and still crash in gaming and video editing ect

 

if ur in the uk I have a delid tool if u want to borrow it il post it to you 

 

I would probly pick a clock speed you’re trying to run and let prime 95 run for 30 mins then drop .01v and test again untill you crash. Once you get a crash go back up 0.02v for good luck and let prime run for 4 hours or so

 

then go about using ur pc as normal. If you do start to get crashes u can always add more voltage back 

-13600kf 

- 4000 32gb ram 

-4070ti super duper 

Link to comment
Share on other sites

Link to post
Share on other sites

@Frank Cisco

 

When using ThrottleStop were you adjusting both the core and the cache voltages? Post a screenshot of the FIVR window so I can see your settings. 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, unclewebb said:

@Frank Cisco

 

When using ThrottleStop were you adjusting both the core and the cache voltages? Post a screenshot of the FIVR window so I can see your settings. 

image.png.847cba391924d7ef8c8642aa53dd6e43.png

 

image.png.b7666b1e8cdb983a28c674f94f697b1a.png

 

there , so how i can put this values in my BIOS? i have an asus Z370-F

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Hinjima said:

I had a 9900k that ran quite hot during summer and I undervolted it. The only thing I did was lower the core voltage from 1.37 to 1.25v in the BIOS and it was stable, is it most optimal way of doing it.. Probably not, is it the fastest way, yep!  I went from 85c temps to around 60c in games, incredible.

 

1 hour ago, RONOTHAN## said:

What it sounds like you're experiencing is called clock stretching, where if the CPU will automatically underclock itself if the voltage drops too much and not report it to software. Usually, the solution to avoid this is to set up a manual "overclock," where you set the clock speed manually to whatever you want it to be (say 4.8GHz), set the core voltage to manual mode and try to minimize the voltage that way. That said, if you don't have a Z- series motherboard, that method won't really work. 

 

1 hour ago, Ebony Falcon said:

My 8700k did

4.7 at 1.2v

4.8 at 1.24v

4.9 at 1.28

5 at 1.32v

and 5.1 at 1.4v all at llc lv6 

 

r23 isn’t a good stress test really as u can pass that many times and still crash in gaming and video editing ect

 

if ur in the uk I have a delid tool if u want to borrow it il post it to you 

 

I would probly pick a clock speed you’re trying to run and let prime 95 run for 30 mins then drop .01v and test again untill you crash. Once you get a crash go back up 0.02v for good luck and let prime run for 4 hours or so

 

then go about using ur pc as normal. If you do start to get crashes u can always add more voltage back 

thank you guys, i posted 2 screenshots of Throttlestop  , the thing here is that i don't know where to apply that same values in my BIOS 

Link to comment
Share on other sites

Link to post
Share on other sites

BTW i just played TLOU for about an hour and it was smooth,( 60fps all the time with latest patch and latest nvidia drivers and windows 11 updates) with that settings, should i keep trying and see how much can i undervolt the cpu or its just fine? also someone said that cinebench is not good to undervolt , what else do you gts suggest , this its my first time attempting the undervolting

Link to comment
Share on other sites

Link to post
Share on other sites

39 minutes ago, Frank Cisco said:

undervolting

You have VBS enabled so your undervolt is not working at all. You need to disable VBS including core isolation memory integrity if you want to undervolt while in Windows 11. Reboot and delete the ThrottleStop.INI configuration file before trying to use ThrottleStop. Post an updated FIVR screenshot after you do that.

 

https://beebom.com/how-disable-virtualization-based-security-vbs-windows-11/

Link to comment
Share on other sites

Link to post
Share on other sites

55 minutes ago, unclewebb said:

You have VBS enabled so your undervolt is not working at all. You need to disable VBS including core isolation memory integrity if you want to undervolt while in Windows 11. Reboot and delete the ThrottleStop.INI configuration file before trying to use ThrottleStop. Post an updated FIVR screenshot after you do that.

 

https://beebom.com/how-disable-virtualization-based-security-vbs-windows-11/

LOL i knew i was doing something wrong, thanks for the reply, I'll do that , and start again. I'll keep you posted.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, unclewebb said:

You have VBS enabled so your undervolt is not working at all. You need to disable VBS including core isolation memory integrity if you want to undervolt while in Windows 11. Reboot and delete the ThrottleStop.INI configuration file before trying to use ThrottleStop. Post an updated FIVR screenshot after you do that.

 

https://beebom.com/how-disable-virtualization-based-security-vbs-windows-11/

ok dude , done, deleted .INI and disabled VBS , here are the screenshots. I'll follow your lead. 

image.png.768e226b72f10c27737151bed8ac5161.png

 

image.png.bd3c6e1492ff32f10d5b72ea4b7fb5ad.png

Link to comment
Share on other sites

Link to post
Share on other sites

@Frank Cisco

You need to undervolt the core and the cache. Start more conservatively. Do some testing at -100 mV or -75 mV for both. Try running the TS Bench 960M test. This is a good first test to check for stability. If it reports any errors, the CPU needs more voltage.

Link to comment
Share on other sites

Link to post
Share on other sites

3 hours ago, unclewebb said:

@Frank Cisco

You need to undervolt the core and the cache. Start more conservatively. Do some testing at -100 mV or -75 mV for both. Try running the TS Bench 960M test. This is a good first test to check for stability. If it reports any errors, the CPU needs more voltage.

Ok thanks , if it passes the stability, do I keep decreasing voltage by -5 or just by -1 , so basically I keep testing until I reach a crash or the bench decreases in score , and then I just put the value that was before, do I understand right ? As I said I'll do what you tell me , because first of all I'm new doing this, and second one ,you're the one that is giving me clear instructions. So I trust you. And I appreciate it. 

Link to comment
Share on other sites

Link to post
Share on other sites

19 minutes ago, Frank Cisco said:

do I keep decreasing voltage by -5 or just by -1

When I am first testing for stability I tend to adjust the voltage in bigger steps like 25 mV. If you only adjust the voltage by 1 mV at a time it will take forever. If you find any errors at -100 mV then go back to -75 mV. If you are OK there then try somewhere in the middle like -85 mV. There is nothing too scientific here. Try to find a voltage that is 100% stable. Use that offset voltage for a few days to make sure there are no random crashes or blue screens. A lot of blue screens when undervolting happen when the CPU is lightly loaded, not fully loaded. That is why it is important to use your computer as you normally would. If your computer crashes during normal use when undervolting, it needs more voltage.

 

There is no need to try and find the perfect voltage value. You do not want to be on the knife's edge of stability. Adjusting by 1 mV is too small of an adjustment. If you are not stable at -100 mV then I would add on at least 5 mV and probably 10 mV. Perhaps an offset of -90 mV would be a good compromise in this situation.   

 

What voltage are you testing? Are you seeing any improvement in your temperatures by undervolting or improvements in performance? If your computer is able to run at full speed during Cinebench with default voltage, undervolting is not going to make it run any faster. It might run a little cooler.

 

8700K processors are unlocked which means you can overclock them and run them faster than their default spec. Most users are more interested in doing this compared to undervolting their 8700K. What is your goal?

Link to comment
Share on other sites

Link to post
Share on other sites

37 minutes ago, unclewebb said:

When I am first testing for stability I tend to adjust the voltage in bigger steps like 25 mV. If you only adjust the voltage by 1 mV at a time it will take forever. If you find any errors at -100 mV then go back to -75 mV. If you are OK there then try somewhere in the middle like -85 mV. There is nothing too scientific here. Try to find a voltage that is 100% stable. Use that offset voltage for a few days to make sure there are no random crashes or blue screens. A lot of blue screens when undervolting happen when the CPU is lightly loaded, not fully loaded. That is why it is important to use your computer as you normally would. If your computer crashes during normal use when undervolting, it needs more voltage.

 

There is no need to try and find the perfect voltage value. You do not want to be on the knife's edge of stability. Adjusting by 1 mV is too small of an adjustment. If you are not stable at -100 mV then I would add on at least 5 mV and probably 10 mV. Perhaps an offset of -90 mV would be a good compromise in this situation.   

 

What voltage are you testing? Are you seeing any improvement in your temperatures by undervolting or improvements in performance? If your computer is able to run at full speed during Cinebench with default voltage, undervolting is not going to make it run any faster. It might run a little cooler.

 

8700K processors are unlocked which means you can overclock them and run them faster than their default spec. Most users are more interested in doing this compared to undervolting their 8700K. What is your goal?

My goal it's to have lower temps and have the same or closest to same performance by undervolting. My CPU is old I just upgraded my GPU to an Asus ROG rtx 3070.  So what I'm looking for it's to achieve as I said lower temps, have the same performance and increase a little more the lifespan of my CPU in order to save for a new build, also I forgot to mention that where I live we have temps over 32 Celsius degrees, despite the CPU having a Corsair h110i aio, GPU package goes up to 70-75 ish on iddle after 6 hours of working. I do home office from 9 am to 7 pm , so I use this PC for work and gaming . So yes, basically that's why I'm trying to undervolt. 

Link to comment
Share on other sites

Link to post
Share on other sites

@unclewebb thanks for the advice man, i was able to use Intel XTU , after disabling what you said, also ran multiple benchmarks, now this is my final question, how do i know if my settings are really applied?


i can see that the core voltage offset is -0.170v , i have been using the Pc for 7 hours ( played TLOU for 30 mins ) and no crashes at all. My power plan its set to balanced now , because i was using max performance , here is a screenshot. So , basically that's my question and thanks.

 

image.png.4f00dbe495577f44e7c1ee6205a7aa9c.png

Link to comment
Share on other sites

Link to post
Share on other sites

@Frank Cisco

Intel XTU used to have problems with reliably applying voltages. This was especially true after resuming from sleep or hibernate.

 

Use HWINFO to see if your offset voltage is being applied correctly to both the core and the cache. 

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, unclewebb said:

@Frank Cisco

Intel XTU used to have problems with reliably applying voltages. This was especially true after resuming from sleep or hibernate.

 

Use HWINFO to see if your offset voltage is being applied correctly to both the core and the cache. 

thanks for the reply, here it's the screenshot : if you need an specific one, just tell me where to find that option, thanks again. image.png.19025ca96b6d07918af5bb6e274c0d96.png

Link to comment
Share on other sites

Link to post
Share on other sites

@Frank Cisco

Look for the Voltage Offsets section of HWiNFO. Check that your XTU voltages are being properly applied and they are applied when you boot up or when you resume from sleep or hibernate if you use that. 

 

image.png.09d084bccf4b2e7a18c7ba800423643b.png

Link to comment
Share on other sites

Link to post
Share on other sites

10 hours ago, unclewebb said:

@Frank Cisco

Look for the Voltage Offsets section of HWiNFO. Check that your XTU voltages are being properly applied and they are applied when you boot up or when you resume from sleep or hibernate if you use that. 

 

image.png.09d084bccf4b2e7a18c7ba800423643b.png

ok , i see them , after booting up , and resume from sleep. but i only see them in 2 of the 5 sections you highlighted:

image.png.bb888287c311d0ce318ec027cb002f3b.png 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Frank Cisco said:

i only see them in 2 of the 5 sections you highlighted

That is correct. You should only undervolt the core and the cache. Those are the two most important items to undervolt. 

 

Are you stable at -170 mV? That is more than most people are able to run and still remain stable. You are not overclocking your CPU so that might be why you can run such a big undervolt. If you are stable then everything looks good. 

 

If you do not want to use XTU, you can try setting a similar -170 mV negative offset voltage in the BIOS. 

 

Did this undervolt improve your temperatures? You can try running something consistent like Cinebench at default voltage for a baseline and then at an offset of -170 mV. The temperatures should improve quite a bit. 

Link to comment
Share on other sites

Link to post
Share on other sites

53 minutes ago, unclewebb said:

That is correct. You should only undervolt the core and the cache. Those are the two most important items to undervolt. 

 

Are you stable at -170 mV? That is more than most people are able to run and still remain stable. You are not overclocking your CPU so that might be why you can run such a big undervolt. If you are stable then everything looks good. 

 

If you do not want to use XTU, you can try setting a similar -170 mV negative offset voltage in the BIOS. 

 

Did this undervolt improve your temperatures? You can try running something consistent like Cinebench at default voltage for a baseline and then at an offset of -170 mV. The temperatures should improve quite a bit. 

i had a crash at -170 mV an hour ago, yesterday everything was fine, so i just put another +5 mv , and i'll keep testing and seeing if my cpu is stable. And yes this undervolt improved a lot my temps , from going to 90 celsius under load  , to 70 in package , and in idle from 50 to 30, it's a lot of improvement. (maybe because of the aio and the fans )  Also today , atm we have an outside temp of 31, and its expected to rise to 35 in 3 hours. 

anyway, thanks man for your advice, and i'll just keep and eye if the pc crashes again,to rise the voltage in increments of 5 or 10 mv. 

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×