• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
the problem was getting the heat from the die to the IHS, if you direct die then the problem is getting the heat from the die to said cooler, the problem still exists.
this is where liquid metal cam in. Die > IHS or Die > Cooler when using liquid metal this it was the really game changer

going Die > IHS > Cooler with liquid metal show next to no difference in in comparison to just Die > Cooler (again with liquid metal)
this is why i thought it was a thing of the past.
 
Permabanned
Joined
11 Jan 2019
Posts
3,214
Location
bedlam
TG Conductonaut liquid metal

refer to my last post..

the problem was getting the heat from the die to the IHS, if you direct die then the problem is getting the heat from the die to said cooler, the problem still exists.
this is where liquid metal cam in. Die > IHS or Die > Cooler when using liquid metal this it was the really game changer



its not the Direct die cooling that dropped your temps, its the liquid metals abilty to move the heat from the die to the IHS or in you case the cooler that as fixed the problem

going Die > IHS > Cooler with liquid metal show next to no difference in in comparison to just Die > Cooler (again with liquid metal)


Thermal Grizzly conductonaut - 73 W/mk
ARCTIC MX-2 - 5.6 W/mK (better than stock intel)
Pure indium 86 W/mK


the problem with the indium in intel made it about .4mm thick and that add thermal resistance, they also need to add other metals so thay can use it as a solder interface interface materiel. this will drop it below the well known liquid metal (Gallium) to about 47w/mK

https://www.indium.com/thermal-management/thermal-k-list/
 
Last edited:
Associate
Joined
30 Aug 2018
Posts
2,483
I previously mentioned Standby Voltage in this thread so I want to give an update regarding it. On the Maximus XI Formula leaving it on auto and exceeding a certain voltage or frequency would push the standby voltage much higher than the default 1v (ish).
The same behaviour occurs with the Maximus XI Extreme even when LN2 mode is not selected.

After some searching I found a post from Elmor stating that the standby voltage changes above 5.2GHz (?) according to a rule in the bios that informs the auto setting. He states that the increased voltage should be fine for benching.

That is from this thread
https://community.hwbot.org/topic/164325-maximus-ix-apex-overclocking-thread/page/2/

While I can see the question asked about whether or not increased standby voltage is fine for 24/7 overclocks I can't see an answer from Elmor about it.

I'm going to continue searching for the answer to this and setting the standby voltage to 1v manually until I've determined that a higher voltage is safe when not using exotic cooling.
 
Associate
Joined
18 Jun 2011
Posts
318
Location
Isle of Wight
I previously mentioned Standby Voltage in this thread so I want to give an update regarding it. On the Maximus XI Formula leaving it on auto and exceeding a certain voltage or frequency would push the standby voltage much higher than the default 1v (ish).
The same behaviour occurs with the Maximus XI Extreme even when LN2 mode is not selected.

After some searching I found a post from Elmor stating that the standby voltage changes above 5.2GHz (?) according to a rule in the bios that informs the auto setting. He states that the increased voltage should be fine for benching.

That is from this thread
https://community.hwbot.org/topic/164325-maximus-ix-apex-overclocking-thread/page/2/

While I can see the question asked about whether or not increased standby voltage is fine for 24/7 overclocks I can't see an answer from Elmor about it.

I'm going to continue searching for the answer to this and setting the standby voltage to 1v manually until I've determined that a higher voltage is safe when not using exotic cooling.
Set both PLL Termination and CPU Standby to 1.1v and forget.

https://community.hwbot.org/topic/186670-rog-maximus-xi/page/8/
 
Last edited:
Associate
Joined
18 Jun 2011
Posts
318
Location
Isle of Wight
Thats cool. Kindof crazy that even with solder this is "worth doing". I thought it was just intels toothpaste that was a problem. Can you lap the die or is that just madness?
Some people lap the die. Beware though, my previous 9900K died a few weeks back, I must have damaged some of the traces on the die when removing the solder, the IMC was never the same afterwards, and it died completely a few weeks after.

My new 9900K is just as good overclocking wise, uses about the same voltages, but the thermals are slightly better on my new one, so wont be Direct die cooling this one, no need.;)
 
Soldato
Joined
19 Nov 2015
Posts
4,867
Location
Glasgow Area
Some people lap the die. Beware though, my previous 9900K died a few weeks back, I must have damaged some of the traces on the die when removing the solder, the IMC was never the same afterwards, and it died completely a few weeks after.

My new 9900K is just as good overclocking wise, uses about the same voltages, but the thermals are slightly better on my new one, so wont be Direct die cooling this one, no need.;)
Oh don't worry, you wont catch me delidding a CPU any time soon. Too risky for me. I'm in my 30's so I am currently throwing all my cash at my mortgage, so don't really have much disposable to risk it. Maybe in the future I will experiment more.
 
Associate
Joined
30 Aug 2018
Posts
2,483
I've made some changes to my setup since the screenshots i posted last.
Okay, I finally pulled my finger out and uploaded some images.

5-GHz-Cinebench-2102.png

5-GHz-Realbench-1hr-Pass.png

1.128 was the lowest vcore I caught in a screenshot but not the lowest I observed.

5-1-GHz-Cinebench-2192.png

5-1-GHz-1hr-Realbench-pass.png

I think the lowest vcore for this was about 1.146v according to cpu-z

5-2-GHz-Cinebench-2189.png


The score on this one is quite low because I had other things running in the background while benching during the 5.2GHz run.


5-2-GHz-1hr-Realbench-pass.png

5-3-GHz-Cinebench-2224.png

5-3-GHz-1hr-Realbench-pass.png
5-4-GHz-Cinebench-2300.png

All runs were back to back and with an ambient temp of around 18 degrees ish.

  • CPU delidded and with liquid metal (no die frame right now :eek: )
  • RAD xspc tx360 with 3 be quiet! 120mm high speed PWM fans in pull config
  • CPU BLOCK EK Velocity
  • RESERVOIR Phanteks R160
  • PUMP EK DDC PWM
  • COOLANT Ice Dragon
  • VRMS also in the loop
I've got my GPU on a separate loop and when I have enough fans I am going to be adding a MO-RA3 as an external radiator to the CPU loop.

Some of the runs have lower than expected scores in Cinebench and this is due to other things going on in the background while I was benching. I only have the one PC and not a great deal of time so I have to multitask, sorry about that. I may revisit this and update the scores if I begin tweaking the RAM and working on lowering the vcore for a given frequency.
I can run Realbench at 5.4GHz but it starts giving WHEA errors, you'll see that all my screenshots include HWinfo to show that each run was WHEA error free. I 'think' I am at the limit of the board now in terms of power delivery, or perhaps my loop is at the limit in terms of heat but I am not going to try overclocking at below ambient.

For what it's worth, the VRMs on the Maximus XI Formula peaked at 42 degrees during the 5.3GHz 1hr Realbench run and don't go higher for 5.4GHz Realbench so I think they are doing fine. Even if the temp sensor is inaccurate and is reporting something crazy like 40 degrees lower than the hottest point I'm still well within the spec.


  • 2x 360 20mm xspc tx360
  • 1x 120 20mm xspc tx120
  • The GPU is now in the same loop
  • The VRMs are currently not in the loop
  • The CPU block is still the EK Velocity. Now mounted upside down (not goofy) with long M3 screws.
  • More liquid metal than usual on the die
  • Maximus XI Formula has been replaced with the Maximus XI Extreme
  • Still no chiller
  • Ambient temps are higher


Result:

5.4GHz 1hr Realbench stable, no AVX offset, no chilling of the loop. No WHEA.
vcore is set as 1.4v in BIOS and LLC is level 6. Underload the vcore droops to 1.332v

Also stable in Apex legends, Div 2, Overwatch etc.

Temps at idle/browsing are 30s and temps while using the machine for more demanding tasks are 40s. I don't know what it is that has changed, or whether it is the combination of all the recent changes I listed above, but temps on the core are much, much better now. So much better than they were previously. The difference is huge.

Screenshot-180.png

https://i.postimg.cc/FR6ZqWP8/Screenshot-180.png
 
Last edited:
Associate
Joined
30 Aug 2018
Posts
2,483
@Distracted. Cant see much in the way of numbers in your screenies, what are your Realbench temps like at 5.4ghz ?
I've added a direct link to the full size screenshot - sry, I thought you could reach it from the image in the forums. If you follow the link and hit download original (if you want to) you can see the thing in full size and it is much clearer. I'm not sure why my images are all blurry.

Here's a cropped version:
 
Associate
Joined
30 Aug 2018
Posts
2,483
I've noticed :mad: I've set the standby at 1v for now like I did previously. I'll adjust the PLL as you suggest. ty

Edit: @moorhen2 do you know what the default values should be and which pll value I should be adjusting in the bios? There is PLL Termination voltage and then in another tab there are several other PLL values :confused:
 
Last edited:
Back
Top Bottom