ROG Maximus XI

[unityofsaints]

On 3/19/2019 at 9:36 PM, aerotracks said:

No Extreme BIOS for Extreme Board, what has it come to in 2019 ?

By "Extreme", ASUS means extreme watercooling, It's been that way since at least the IX. The VIII wasn't the best board for that socket but a least there was some LN2 benching at the time. 

Edited March 21, 2019 by unityofsaints

[aerotracks]

14 minutes ago, unityofsaints said:

By "Extreme", ASUS means extreme watercooling, It's been that way since at least the IX. The VIII wasn't the best board for that socket but a least there was some LN2 benching at the time. 

I am quite certain we are not talking about a watercooling board. ?

The XI Extreme sitting in front me has LN2 Mode, RSVD switch, slow mode, retry and safe boot as well as Dual BIOS which even the Gene doesn't have.

[unityofsaints]

2 minutes ago, aerotracks said:

I am quite certain we are not talking about a watercooling board. ?

The XI Extreme sitting in front me has LN2 Mode, RSVD switch, slow mode, retry and safe boot as well as Dual BIOS which even the Gene doesn't have.

Yes but 4 dimm slots and 0 LN2 results on HWBot...

Edited March 21, 2019 by unityofsaints

[aerotracks]

I am aware Gene only has 2 slots, 3 if you use a hammer. Sometimes 4 DIMM config is an advantage.

[dumo]

If there any feedback to improve performance on MXIE, pretty sure they will release new bios.

Even Strix I-Gaming comes with beta bios to test memory.

[Alex@ro]

Bios 0050 for Maximus XI Apex and Gene, can change BCLK with OC PANEL : 

Maximus XI Apex

 

Maximus XI Gene 

 

A few tips for memory validation  from the master himself  : 

 

"

 

Cpu ratio 16x

Cache 16xmin max

Dram ratio 5500

Max tweak mode2

cl-24-31-63-3t

max out trfc

vccio/sa 1.3/1.75

corevcore=1.45

dram v 1.6v

pll term 1.5v

pll bandwidth level 6

System agent pll v=1.005

Mem controller pll v=1.005

ME state disable

Mrc fast boot enable

 

Disable all mem training except:

Wr time centering 1d

Early wr time centering 2d

Jedec wr leveling

Receive enable training

Early readmpr time centering 2d

Senseamp offset training

 

Train ~ -100 to 140c

Keep retrying till you get the ‘post code’ 50 stall for some seconds due to me disable

Immediately torch mem pot to ~ -80C

Usually after training mem needs 70 to 80c

 

You want the max freq you can train pass from start to post code ‘50’ so try to train as hi as possible here. Right after 50 you can quickly press oc panel down bclk about 3~5 mhz before finish post + enter os to boot easier

 

Always use 100mhz bclk in bios so mrc fast boot will always not retrain, for the actual training bclk at any point, be at before or during, just use oc panel to adjust the actual training bclk.

When validating:

Say you need 104 bclk validation:

Open cpuz up bclk to 104 with ocpanel and apply but do not press f7 now, immediately down bclk a 1mhz or some and then apply then press f7 quickly. Because cpuz is slow it will f7 the higher mhz, f7 at high mhz always crash. Always make sure you get the file out or use sync to force flush data to disk.

"

 

Edited April 1, 2019 by Alex@ro

[flanker]

bad question, but how to connect OC Panel/Panel II to Maximus XI series? I have old OC Panels and also cables from old series of Maximus boards. New one connector on board seems a little different ?

[Alex@ro]

You need the new cable Michal....

[flanker]

Where I can buy it or find it?

[gkmltd]

48 minutes ago, flanker said:

Where I can buy it or find it?

+1

[SoF]

Did anyone here had trouble reproducing results when testing memory?

I'm having a very strange issue that my new b-dies (a1) won't be able to reproduce results which have just been fine in 32m.

I could narrow it down to VTT DDR and RAM voltage.

VTT DDR is VERY picky and it will only pass at 0.84375V (higher = instant crash or very unstable at least). RAM is also very unpredictable and needs to be adjusted between 1.74 and 1.75V a couple of times.
All of a sudden everything seems to run nice and stable, then doing a reboot, setting the EXACT same voltages in BIOS or now after some testing, just a reboot -> completely unstable again.

32m passed

AM3 passed

pifast passed (btw pifast is a good indicator if there is a good setting or not)

Flashed the newest 0050 bios but still the same (was 0805 before).

When rebooting even 1m or pifast will crash almost instantly and I need to adjust voltages. Tried to disable most of the training stuff and checked MemTweakIt settings but all seem to be the same. It all comes down to adjusting voltages (vram, vttddr) within 1 or 2 ticks in turbovcore a couple of times and all of a sudden it seems to be stable again (at the same values as before).

Once stable I can run all 3D Marks, 32m and everything...reboot -> start the struggle again.
I'm out of ideas right now ^^

€: I might have one bogus stick...need to check this tomorrow...TRFC values are way to high, sometimes boot, sometimes need to be ~700 to be stable..god damn, I hate testing memory...

€²: and it's not the cpu clock...5.5 is completely stable for everything at this point...tried with 5 gig and it's all the same. 100% memory issue.

Edited April 1, 2019 by SoF

[websmile]

At 1,75v, did you set maxmem to available max of 2,5gb? I left you a note at AF that B-die needs maxmem at x64 bit OS because of its characteristics at tight timings and/or high volts, when I look at your 32m I don´t think you set this...

[SoF]

Haven't set maxmen yet - haven't read that it is that important for mem overclocking yet.
A quick test showed I'm no longer having issues now with low TRFC and so on...damn I owe you something man

€: okay when I get to boot, it looks stable but I just had a strange reboot where previous settings won't boot...must be this strange "training" stuff...need to work around that but windows-wise the maxmem was all it took...still can't boot 1T...but that's for the next days, had enough of that memory-crap for today.

Thanks for helping me out with memory another time btw: It's a F4-4133C19D-16GTZA which was in your hands already

Edited April 1, 2019 by SoF

[marco.is.not.80]

2 hours ago, gkmltd said:

+1

+1

[SoF]

@websmile Wasted 3-4 hours of testing but you saved my night...now I can sleep well

AM3 @2222MHZ

32m@2222MHZ

No issues anymore!

Tomorrow is sub-timing day ?

Edited April 1, 2019 by SoF

[PatitoStark]

Hi (sorry por ask in this topic.) 

I recently changed a formula x for an apex xi.

I do not understand how the overclock of this Mb works.

I have a 9900k at 5 ghz with water.

In formula x i have 5 ghz with 1.18 vcore. At 5.3ghz a need 1.32 vcore.


Temp of package and cpu (peci) are on 73-86º.

However in apex xi need 1.12 vcore and i cant 5.3 ghz. Only can 5.2 at 1.25vcore.

With 1.3 vcore at 5 ghz. I have this. I have 20 º more than formula X . If I release 4 cibenech the pc turns off



why? How I fix this?

[aerotracks]

6 minutes ago, PatitoStark said:

With 1.3 vcore at 5 ghz. I have this. I have 20 º more than formula X . If I release 4 cibenech the pc turns off
why? How I fix this?

Maximus XI has different VCore measuring method from Maximus X, so you can't compare like that. Explanation by elmor here:

https://www.overclock.net/forum/27686004-post2664.html

Edited April 3, 2019 by aerotracks

[Alex@ro]

Check that you don't have ln2 mode turned on by accident. Also the temperature difference can be because of too high auto pll termination and standby, put them manually at 1.1V and see if you still have big temps. 

[subaruwrc]

On 4/2/2019 at 1:07 AM, SoF said:

@websmile Wasted 3-4 hours of testing but you saved my night...now I can sleep well

AM3 @2222MHZ

32m@2222MHZ

No issues anymore!

Tomorrow is sub-timing day ?

what is your system agent voltage ? 

[PatitoStark]

3 hours ago, aerotracks said:

Maximus XI has different VCore measuring method from Maximus X, so you can't compare like that. Explanation by elmor here:

https://www.overclock.net/forum/27686004-post2664.html

 

Thanks for the help.

Something Elmor says he knew.

My problem is before reaching 5.3 ghz with 1.32 in maximus X and yet in maximus xi with the options that have not arrived. It fails me because of temperatures. Since they are 20 degrees more than in formula X,

If I leave everything automatic and only change vcore and frequency, it depends on the day and the grades of the room I pass a cibenech.

at 5 ghz I have good temperatures playing because I have a good liquid rl custom and  my 9900k need only 1.15 vcore  .if I had to put it at 1.3 vcore I could not.

[aerotracks]

Again, you can't compare VCore readouts. If your CPU is warmer on Maximus XI then you are giving it more VCore than on your old Maximus X board.

Edited April 3, 2019 by aerotracks

[SoF]

9 hours ago, subaruwrc said:

what is your system agent voltage ? 

1.275V and tried up to 1.3V which makes the system more unstable. I had no luck running 1T yet (even down to 2020 MHZ memory)...the system will just go into a boot-loop and reboot into safe mode.

I am pre-testing for AM3 currently and aim for max eff. there..I managed getting from 501k to over 517k tonight...

[subaruwrc]

try higher

I had the same sympthoms at the begining of my b-die learning.

try 1.3 io 1.3 Sa for start then raise the SA when you can boot 1t

 

 

[PatitoStark]

15 hours ago, aerotracks said:

Again, you can't compare VCore readouts. If your CPU is warmer on Maximus XI then you are giving it more VCore than on your old Maximus X board.

sorry, I do not compare voltage, I only take it as reference. I want pass cb at 5.3 like maximus X  but i cant and i dont know .

some bios parameter will be wrong

[Niuulh]

On 4/1/2019 at 1:21 PM, Alex@ro said:

Bios 0050 for Maximus XI Apex and Gene, can change BCLK with OC PANEL : 

Maximus XI Apex

 

Maximus XI Gene 

 

A few tips for memory validation  from the master himself  : 

 

"

 

Cpu ratio 16x

Cache 16xmin max

Dram ratio 5500

Max tweak mode2

cl-24-31-63-3t

max out trfc

vccio/sa 1.3/1.75

corevcore=1.45

dram v 1.6v

pll term 1.5v

pll bandwidth level 6

System agent pll v=1.005

Mem controller pll v=1.005

ME state disable

Mrc fast boot enable

 

Disable all mem training except:

Wr time centering 1d

Early wr time centering 2d

Jedec wr leveling

Receive enable training

Early readmpr time centering 2d

Senseamp offset training

 

Train ~ -100 to 140c

Keep retrying till you get the ‘post code’ 50 stall for some seconds due to me disable

Immediately torch mem pot to ~ -80C

Usually after training mem needs 70 to 80c

 

You want the max freq you can train pass from start to post code ‘50’ so try to train as hi as possible here. Right after 50 you can quickly press oc panel down bclk about 3~5 mhz before finish post + enter os to boot easier

 

Always use 100mhz bclk in bios so mrc fast boot will always not retrain, for the actual training bclk at any point, be at before or during, just use oc panel to adjust the actual training bclk.

When validating:

Say you need 104 bclk validation:

Open cpuz up bclk to 104 with ocpanel and apply but do not press f7 now, immediately down bclk a 1mhz or some and then apply then press f7 quickly. Because cpuz is slow it will f7 the higher mhz, f7 at high mhz always crash. Always make sure you get the file out or use sync to force flush data to disk.

"

 

Thanks for thoses infos. Where can i find ME State in this BIOS please ?