9900KS - Maximus XI Extreme OC Help - Overclock.net - An Overclocking Community

Forum Jump: 

9900KS - Maximus XI Extreme OC Help

Reply
 
Thread Tools
post #1 of 274 (permalink) Old 11-28-2019, 10:38 AM - Thread Starter
New to Overclock.net
 
Join Date: Mar 2018
Posts: 105
Rep: 0
9900KS - Maximus XI Extreme OC Help

I am new to overclocking in general, but I have done alot of reading on the subject and I'm working on overclocking my 2nd Gaming PC.

I have a 9900KS on an Asus Rog Maximus XI Extreme eATX board.
H115i 280mm AIO on it as well, and all the fans I could fit.

I followed Der8auer's youtube video on overclocking the 9900K, and he mentioned to adjust LLC from 6 up to 7 if need be, or increase the voltage higher if you see Vdroop kicking in when testing Cinebench R15 or Prime95.
I've tried LLC 6 at 1.30-1.33 and then i move to LLC 7 from 1.30 up to 1.35, and it's always dropping the voltage during stress testing.

It seems to stay stable through the tests, but it is having Vdroop Kicking in.

Will I have any issues later that I should try and fix now?

If it goes through Prime95 v26.6 for about 30 minutes at 1.30V and LLC7, but has Vdroop kicking in during the test, should I leave it as is? Or can I drop it further to LLC 6 and run again to try and pass it through for another 30 minutes?
If the voltage drop isn't going to cause crashes in my games, then I can push the voltage/LLC/AVX offset a little more, correct?

Last edited by MikeJeffries; 11-30-2019 at 04:20 PM.
MikeJeffries is offline  
Sponsored Links
Advertisement
 
post #2 of 274 (permalink) Old 11-28-2019, 10:46 AM
New to Overclock.net
 
skupples's Avatar
 
Join Date: Apr 2012
Location: Fort Lauderdale
Posts: 20,189
Rep: 580 (Unique: 324)
what happens @ 1.4?

i haven't tuned my 97k much, I dialed in an ez 5.0 oc, tuned via adaptive + offset, llc7, (which I assumed meant the least LLC, not completely off) flipped on xmp ran a few tests, and went on my way.

R.I.P. Zawarudo, may you OC angels' wings in heaven.
If something appears too good to be true, it probably is.

Last edited by skupples; 11-28-2019 at 10:49 AM.
skupples is online now  
post #3 of 274 (permalink) Old 11-28-2019, 11:50 AM
New to Overclock.net
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 6,391
Rep: 398 (Unique: 284)
Quote: Originally Posted by MikeJeffries View Post
I am new to overclocking in general, but I have done alot of reading on the subject and I'm working on overclocking my 2nd Gaming PC.

I have a 9900KS on an Asus Rog Maximus XI Extreme eATX board.
H115i 280mm AIO on it as well, and all the fans I could fit.

I followed Der8auer's youtube video on overclocking the 9900K, and he mentioned to adjust LLC from 6 up to 7 if need be, or increase the voltage higher if you see Vdroop kicking in when testing Cinebench R15 or Prime95.
I've tried LLC 6 at 1.30-1.33 and then i move to LLC 7 from 1.30 up to 1.35, and it's always dropping the voltage during stress testing.

It seems to stay stable through the tests, but it is having Vdroop Kicking in.

Will I have any issues later that I should try and fix now?

If it goes through Prime95 v26.6 for about 30 minutes at 1.30V and LLC7, but has Vdroop kicking in during the test, should I leave it as is? Or can I drop it further to LLC 6 and run again to try and pass it through for another 30 minutes?
If the voltage drop isn't going to cause crashes in my games, then I can push the voltage/LLC/AVX offset a little more, correct?
There will ALWAYS be vdroop at every LLC except 8. And don't use such high LLC anyway. Vdroop is your friend, contrary to what you may think.

Here is what happens if you jack up the LLC too much.

https://elmorlabs.com/index.php/2019...ne-visualized/

Most people don't understand that it isn't the sensor reading or the "RMS" voltage that determines stability in the end. It's the -transient minimum- that does.

The Maximus XI Gene/Xtreme boards work best with LLC5. LLC5 is the best combination of medium vdroop and good transient response. LLC6 isn't worth it because transient 'dips' increase too much below the vcore sensor reading. You will need a higher BIOS voltage for the extra vdroop but this is NOT a problem. 1.40v BIOS set + LLC5 is perfectly safe to use for day to day usage (don't try to run small FFT AVX Prime95 or LinX 0.9.6 at this setting though. Stick to Realbench 2.56 or Cinebench R20 with 7200 custom second loops if you want to stress test that).

In the 16k prime95 AVX test by buildzoid, you can see that "average to minimum" voltage distance is closer at LLC5 than LLC6. That's why LLC5 works so well. (Use his Asus Gene results).


Also some games hate too high LLC+Decreasing BIOS voltage to reduce idle voltage too much.
Apex Legends is known to crash if you use a too low BIOS set voltage+too strong LLC.

For example, in Prime95 29.8 build 6, small FFT with AVX disabled, the following two settings give the exact same max load vcore after vdroop, and same temps/current: (on my Gigabyte Master): at 5 ghz

1.250v + Vcore LLC "Turbo" (LLC6)
1.30v + Vcore LLC "High" (LLC5)

Both give 1.195v full load in Prime95 and is stable.
However Apex Legends will crash randomly at 1.250v +LLC6. (Requires 1.260v bios set to not crash).
But will run just fine with 1.30v +LLC5!
Why?
Because the "low load" voltage is higher at 1.30v LLC5, compared to 1.250v LLC6.
So it's often not a good idea to raise LLC too high and reduce BIOS voltage too low just because a stress test passes at the same load voltage.

Use HWinfo64 to look for WHEA CPU L0 errors in the sensor bottom section.

Another LLC video on torpedoing your voltage regulation


[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM

Last edited by Falkentyne; 11-28-2019 at 12:28 PM.
Falkentyne is online now  
Sponsored Links
Advertisement
 
post #4 of 274 (permalink) Old 11-28-2019, 12:44 PM - Thread Starter
New to Overclock.net
 
Join Date: Mar 2018
Posts: 105
Rep: 0
Quote: Originally Posted by Falkentyne View Post
There will ALWAYS be vdroop at every LLC except 8. And don't use such high LLC anyway. Vdroop is your friend, contrary to what you may think.

Here is what happens if you jack up the LLC too much.

https://elmorlabs.com/index.php/2019...ne-visualized/

Most people don't understand that it isn't the sensor reading or the "RMS" voltage that determines stability in the end. It's the -transient minimum- that does.

The Maximus XI Gene/Xtreme boards work best with LLC5. LLC5 is the best combination of medium vdroop and good transient response. LLC6 isn't worth it because transient 'dips' increase too much below the vcore sensor reading. You will need a higher BIOS voltage for the extra vdroop but this is NOT a problem. 1.40v BIOS set + LLC5 is perfectly safe to use for day to day usage (don't try to run small FFT AVX Prime95 or LinX 0.9.6 at this setting though. Stick to Realbench 2.56 or Cinebench R20 with 7200 custom second loops if you want to stress test that).

In the 16k prime95 AVX test by buildzoid, you can see that "average to minimum" voltage distance is closer at LLC5 than LLC6. That's why LLC5 works so well. (Use his Asus Gene results).

https://www.youtube.com/watch?v=eUPzADCcuTc

Also some games hate too high LLC+Decreasing BIOS voltage to reduce idle voltage too much.
Apex Legends is known to crash if you use a too low BIOS set voltage+too strong LLC.

For example, in Prime95 29.8 build 6, small FFT with AVX disabled, the following two settings give the exact same max load vcore after vdroop, and same temps/current: (on my Gigabyte Master): at 5 ghz

1.250v + Vcore LLC "Turbo" (LLC6)
1.30v + Vcore LLC "High" (LLC5)

Both give 1.195v full load in Prime95 and is stable.
However Apex Legends will crash randomly at 1.250v +LLC6. (Requires 1.260v bios set to not crash).
But will run just fine with 1.30v +LLC5!
Why?
Because the "low load" voltage is higher at 1.30v LLC5, compared to 1.250v LLC6.
So it's often not a good idea to raise LLC too high and reduce BIOS voltage too low just because a stress test passes at the same load voltage.

Use HWinfo64 to look for WHEA CPU L0 errors in the sensor bottom section.

Another LLC video on torpedoing your voltage regulation

https://www.youtube.com/watch?v=R6TkHBhqO_k
Unbelievable thorough explanation! Thank you for this
So do you think LLC5 and 1.30V for a 9900KS on my Maximus XI Extreme should be good?
This PC is primarily used for VR sim racing and will have a motion simulator attached to it, if that matters.
I used my 8700K @ 4.8ghz all core currently and am switching the rig to the new computer
The Maximus X Hero I have is at LLC6 at 4.8ghz and I believe it was 1.32V (maybe less - i cant remember off the top of my head) - but the first pc ran fine for over a year now.

EDIT:
I just ran LLC5 and 1.33V and it drops to 1.217 during stress testing Prim95 and Cinebench R15 and R20. It drops more than the LLC 6 or 7 that i tried before, but temps seems better. It isnt hitting 90's Celcius like it was during the previous stress tests i tried.
Is this normal?
Should I leave it 1.33 or go up/down?

Last edited by MikeJeffries; 11-28-2019 at 01:11 PM.
MikeJeffries is offline  
post #5 of 274 (permalink) Old 11-28-2019, 01:52 PM
New to Overclock.net
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 6,391
Rep: 398 (Unique: 284)
Quote: Originally Posted by MikeJeffries View Post
Unbelievable thorough explanation! Thank you for this
So do you think LLC5 and 1.30V for a 9900KS on my Maximus XI Extreme should be good?
This PC is primarily used for VR sim racing and will have a motion simulator attached to it, if that matters.
I used my 8700K @ 4.8ghz all core currently and am switching the rig to the new computer
The Maximus X Hero I have is at LLC6 at 4.8ghz and I believe it was 1.32V (maybe less - i cant remember off the top of my head) - but the first pc ran fine for over a year now.

EDIT:
I just ran LLC5 and 1.33V and it drops to 1.217 during stress testing Prim95 and Cinebench R15 and R20. It drops more than the LLC 6 or 7 that i tried before, but temps seems better. It isnt hitting 90's Celcius like it was during the previous stress tests i tried.
Is this normal?
Should I leave it 1.33 or go up/down?
Yes, your temps are lower because the "RMS" (averaged load voltage) voltage is lower because of more vdroop.

You can and should reduce BIOS voltage as low as you can go without crashing. But be sure to look for CPU Cache L0 errors (Virtual register store corruption, deals with hyperthreaded cores) if voltage is too low.
With HT disabled, you tend to just BSOD instead (at an even lower voltage of course).

[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM
Falkentyne is online now  
post #6 of 274 (permalink) Old 11-28-2019, 02:02 PM - Thread Starter
New to Overclock.net
 
Join Date: Mar 2018
Posts: 105
Rep: 0
Quote: Originally Posted by Falkentyne View Post
Yes, your temps are lower because the "RMS" (averaged load voltage) voltage is lower because of more vdroop.

You can and should reduce BIOS voltage as low as you can go without crashing. But be sure to look for CPU Cache L0 errors (Virtual register store corruption, deals with hyperthreaded cores) if voltage is too low.
With HT disabled, you tend to just BSOD instead (at an even lower voltage of course).
So LLC 5 is best, I am testing now at 1.32V Vcore.
It's dropping to about 1.20-1.21V when running Prime95
I'm going to let it run for an hour or so, and then I'll drop Voltage to 1.31?
Should I keep doing this until it becomes unstable? Is that the best way?

Also, I have AVX at 4, is that good?
MikeJeffries is offline  
post #7 of 274 (permalink) Old 11-28-2019, 02:06 PM
New to Overclock.net
 
Join Date: Sep 2018
Posts: 58
Rep: 0
Consider upping voltage +.015v to .02 in bios if dropping from llc 7 to 6 on the asus. Though 7 is not recommended.

And +.005 to .01 if ur going from 6 to 5

This will keep ur minimums relatively close from ur base line if it was stable.
DrunknFoo is online now  
post #8 of 274 (permalink) Old 11-28-2019, 02:22 PM
New to Overclock.net
 
skupples's Avatar
 
Join Date: Apr 2012
Location: Fort Lauderdale
Posts: 20,189
Rep: 580 (Unique: 324)
Quote: Originally Posted by Falkentyne View Post
Yes, your temps are lower because the "RMS" (averaged load voltage) voltage is lower because of more vdroop.

You can and should reduce BIOS voltage as low as you can go without crashing. But be sure to look for CPU Cache L0 errors (Virtual register store corruption, deals with hyperthreaded cores) if voltage is too low.
With HT disabled, you tend to just BSOD instead (at an even lower voltage of course).
idk about anyone else, but I also notice a specific kinda load lag when my CPU's volts aren't properly tuned. a "working" voltage, against a functional voltage feel quite different in some titles, like mh:w. or like you highlighted to an extreme w. apex crashing.

i'm glad to read LLC hasn't changed all that much . I had my head in the sand from 5xxx series up til i got this 9700k this time last year.

R.I.P. Zawarudo, may you OC angels' wings in heaven.
If something appears too good to be true, it probably is.
skupples is online now  
post #9 of 274 (permalink) Old 11-28-2019, 03:18 PM
New to Overclock.net
 
Join Date: Nov 2019
Posts: 11
Rep: 0
I have my 9900KS @ 5.2 Ghz running at 1.325 vcore and llc set to turbo on z390 gigabyte aorus master.

In Prime95 my vcore drops down to 1.266 under full load. That has been the lowest I get. I am not gaming though. But I am doing heavy 4k video editing and render operations. So I get a vdroop of 0.6 volt which is pretty good. Do you guys agree? You want to tweak around to get the vdroop as low as possible and as close as possible to your fixed vcore in bios.

Hwinfo is still reading out VID. Which is much higher, but I am not mistaking this not a real voltage. Just a value intel would have set to run at your current scenario.
Attached Thumbnails
Click image for larger version

Name:	Screenshot_20191128_230931_1574979167134.jpg
Views:	7
Size:	243.7 KB
ID:	308696  

AlexauwaHL is offline  
post #10 of 274 (permalink) Old 11-28-2019, 04:43 PM
New to Overclock.net
 
criskoe's Avatar
 
Join Date: Aug 2010
Posts: 183
Rep: 3 (Unique: 3)
Quote: Originally Posted by MikeJeffries View Post
I am new to overclocking in general, but I have done alot of reading on the subject and I'm working on overclocking my 2nd Gaming PC.

I have a 9900KS on an Asus Rog Maximus XI Extreme eATX board.
H115i 280mm AIO on it as well, and all the fans I could fit.

I followed Der8auer's youtube video on overclocking the 9900K, and he mentioned to adjust LLC from 6 up to 7 if need be, or increase the voltage higher if you see Vdroop kicking in when testing Cinebench R15 or Prime95.
I've tried LLC 6 at 1.30-1.33 and then i move to LLC 7 from 1.30 up to 1.35, and it's always dropping the voltage during stress testing.

It seems to stay stable through the tests, but it is having Vdroop Kicking in.

Will I have any issues later that I should try and fix now?

If it goes through Prime95 v26.6 for about 30 minutes at 1.30V and LLC7, but has Vdroop kicking in during the test, should I leave it as is? Or can I drop it further to LLC 6 and run again to try and pass it through for another 30 minutes?
If the voltage drop isn't going to cause crashes in my games, then I can push the voltage/LLC/AVX offset a little more, correct?
As falkentyne wrote. Don’t be so concerned about the droop. It’s perfectly fine to have droop. Even good in our case. Maximus xi extreme owners.

The voltage that really matters here is the voltage vcore reading at full load. When you do your testing make sure you have a pen and paper and write values down as you test. And Don’t change multiple settings between tests. Get you base settings down and then slowly lower vcore voltage and retest until it’s not stable anymore. Note the LOAD vcore and work with that number. When you find you lose stability. Then pop it a notch back up. The quickest way might be. Start at 1.35v LLC 5. And run tests for 1 to 2 hours. After each pass. Lower it .01. Keep doing that till it’s not stable and fails. Then go back the .01 and retest for a long final stability test. How long is up to you. Everyone on this site has different opinions as to what’s adequate. For me personally I go for a 8 hour final run. Note Finding your absolute lowest voltages can be time consuming so prepare for that. Try to be patient and go at it methodically.

After days of changing settings and countless hours of stress tests with that exact same board and cpu. Ive come to the conclusion that LLC 6 is garbage on that board. Finding stability was hard on LLC6 and I wasted so much time with it. I found it actually required higher then needed LOAD voltages to be stable. So I used LLC 5 and it has worked out good.

If you want to see more about my OC. Read the 9900KS Lottery thread. I have ALOT of posts in there that include bios settings and values and such. I got to play with 2 different 9900KS’s so you will see the range and where your chip might sit. I had one mediocre chip and one really good one. I also have the same board so all of the settings I refer to will be the same as you.

For me I wanted to be full P95.29 AVX SMALL FFT Stable. This test requires the highest vcore to be stable cause it really stresses the cpu. ALOT. But temps can get out of control real quick if your cooling is inadequate so be careful if you try.

My final settings I settled with was

5ghz -0 AVX @ 1.325v LLC 5 = 1.172V P95 AVX Full 175 amp LOAD.

as you can see it’s a huge drop but it’s fully stable with AVX P95 and absolutely ANY tests I throw at it. Which is what I was going for. I went for a 8 hour test but I’m crazy. Lol.


Last edited by criskoe; 11-28-2019 at 05:12 PM.
criskoe is online now  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off