ASUS X99 Motherboard Series - Official Support Thread

[Raja@ASUS]

This thread is for posting and sharing tips, tricks and results relating to ASUS X99 motherboards.

5960X OC tips PDF:

https://drive.google.com/file/d/0Bz2...it?usp=sharing


Latest UEFI Build for R5E:
RAMPAGE V EXTREME UEFI 0801
0801 Download Mirror


Latest UEFI build 1002 for the X99-Deluxe:
X99-Deluxe 1002 UEFI
Mirror

-Raja

[FlanK3r]

Thx for link, I lost it ...

[techguy2k5]

I've just purchased an X99 Deluxe and a 5960x, have it water cooled with a 3x120 rad, XSPC Raystorm block, and a D5 pump. Temps stay below TjMax so everything appears ok. However, I've noticed something strange when under load, while stock or overclocked, CPUZ reports the chip drawing significantly higher voltage than what is set in the BIOS when running Prime 95 or Intel Burn Test. At stock the chip may want up to 1.1 volts according to CPUZ under these conditions and @ 4.5GHz and 1.3V with the voltage set mnanually in the BIOS it shoots as high as 1.49V. Something's not right here. The only legitimate reason I can think of for this to happen is if it's specific to an AVX workload, which may be the case with the testing tools I've mentioned. However, my previous ROG boards never did this with Haswell or Ivy Bridge. I've updated the BIOS to the latest beta version available on the support site, 0701 dated 8/29/14. What I'm wondering is the following:
1) is this normal/can this behavior be reproduced by others?
2) is there a setting which controls this behavior? - again, voltage is set manually (not using adaptive or offset)

Edit: I've just run the standard OCCT test (not the Linpack test) which to my understanding should not use AVX and I see the exact same voltage spike. Either CPUZ is wrong, the sensor is damaged, or there's a BIOS setting that affects this behavior.

[SoulsCollective]

Try disabling LLC and see if that changes the readings CPU-Z is giving you. Also, I'm guessing this isn't the case, but you haven't accidentally switched the CPU overvolt jumper on the bottom right of the board, have you (see eg. page 1-20 of your manual)?

Have the same setup (5960X and X99 Deluxe) on its way to me, should be arriving late this week, so will be able to comment if I see the same behaviour then.

[techguy2k5]

Try disabling LLC and see if that changes the readings CPU-Z is giving you. Also, I'm guessing this isn't the case, but you haven't accidentally switched the CPU overvolt jumper on the bottom right of the board, have you (see eg. page 1-20 of your manual)?

Have the same setup (5960X and X99 Deluxe) on its way to me, should be arriving late this week, so will be able to comment if I see the same behaviour then.

Thanks for the suggestions, the BIOS layout is very different compared to the ROG boards I'm used to so I knew I was missing something. I'll check LLC and the board switch you mentioned here in a bit and report back.

[SoulsCollective]

Have paged through the manual and done a quick search, but my Google-fu is failing me - can't find anything on voltage read points for the X99 Deluxe (if any). It could be that CPU-Z is simply getting confused (can happen with new gen) and software readings are either wrong or being misinterpreted; ideally you'd check with a DMM but with no obvious read points I'm not sure how to go about doing that.

[techguy2k5]

Have paged through the manual and done a quick search, but my Google-fu is failing me - can't find anything on voltage read points for the X99 Deluxe (if any). It could be that CPU-Z is simply getting confused (can happen with new gen) and software readings are either wrong or being misinterpreted; ideally you'd check with a DMM but with no obvious read points I'm not sure how to go about doing that.

That's the direction I was going originally. I did a quick search for voltage read points as well and didn't find anything, might be a feature reserved for the Rampage V Extreme.

[FlanK3r]

I had X99 DLX in hands only for short time. I tried light OC via BIOS. I had no time for OCCT or AVX LINX, so I tried only Cinebench and CPUZ voltage was similar as in BIOS (+-0.05V)

[techguy2k5]

I had X99 DLX in hands only for short time. I tried light OC via BIOS. I had no time for OCCT or AVX LINX, so I tried only Cinebench and CPUZ voltage was similar as in BIOS (+-0.05V)

I assume you measured with CPUZ? Do you remember disabling LLC and checking the voltage switch on the mobo? I still haven't done those yet, need a bit more time this morning and was just curious if you recall.

[FlanK3r]

no, I had borrowed board for short time (30 minutes maybe ?), so I set only manual voltage in BIOS, LLC at auto and in benchmark I watched CPUZ.

[techguy2k5]

no, I had borrowed board for short time (30 minutes maybe ?), so I set only manual voltage in BIOS, LLC at auto and in benchmark I watched CPUZ.

That doesn't sound any different than what I've done which concerns me a bit. I'll double check things here before too long. Hopefully I just missed something.

[techguy2k5]

Looks like LLC needs to be taken off Auto to work "properly", at least in my case. Setting to Level 9 and manually setting voltage in the BIOS yields a rock steady voltage reading in CPUZ at almost exactly what I set (1.33V->1.328V). Thanks for the tips, guys.

[SoulsCollective]

Looks like LLC needs to be taken off Auto to work "properly", at least in my case. Setting to Level 9 and manually setting voltage in the BIOS yields a rock steady voltage reading in CPUZ at almost exactly what I set (1.33V->1.328V). Thanks for the tips, guys.

Interesting. Presuming accurate reporting, a 1.3V vs 1.5V core voltage should produce significantly different behaviour. Have you noticed any commensurate changes to temps or stability since changing LLC from auto? I'm wondering whether this has actually 'done' anything or has just resulted in CPU-Z giving a more accurate read.

[Charles Wirth]

Just a few notes.

On the R5E you can enable and disable specific cores and this is working, you can now find the best core easier or disable a problem core if you were lets say, shooting for max frequency. Other manufacturers do not have this feature.

In Windows, with all cores enabled the turbo functions as we normally see it on other chips, setting performance mode in power options (control panel) locks cores to max speed as intended. I mention this because it failed on a different MFG board.

Same setup, all cores enabled, if you adjust each core speed differently with XTU, all cores default core 0 speed regardless of other cores set speed. I emailed Intel on this and got a reply

"There is a feature broken since Ivy Bridge that I would like to see fixed. That is the turbo multiples and the way they work.

Example: in XTU you can set core speeds 0/1 4400 and 2/3 at 4200, meaning the first two cores are set to 4.4Ghz and the last two cores are set to 4.2Ghz, I hope this make sense. You would think that two cores could lock down to 4.4Ghz but they don't, they run at the lower speed set. This holds true if you try with mobiles, Sandy, Ivy and Haswell. AMD got this right and you can lock the cores at x speed.

Where this helps me and I wish this worked. I want to leave 3 cores down at low Mhz and force core 0 to max Mhz. I then lock affinity down for my benchmark to that one core. This does not work on Intel.

Thanks
Charles"

"From engineering:

Thanks for sharing, we?ve discussed similar ideas. Currently the Turbo OC Architecture does not support this. Maybe someday, but no plans currently."

Memory overclocking is working as intended and the R5E gets the most out of DDR4 so far, I say this because my other board has issues that I am sure will be fixed with BIOS updates.

My vcore is very stable, I have not seen the problem experienced on Deluxe.

[Raja@ASUS]

Thx Charles.

[nacho_arroyo]

Great guide Raja! i always read your guide, and always helps!

Thanks

[FlanK3r]

What nice I saw in London...
I think, I need luck for something similar. Next 14 days Im preparing my hand for good choice in shop

[l0ud_sil3nc3]

This thread is for posting and sharing tips, tricks and results relating to ASUS X99 motherboards.

5960X OC tips PDF:

https://drive.google.com/file/d/0Bz2...it?usp=sharing


More to follow soon. Yours truly is in the middle of an oversees trip and will be back in action full-time from 9/1

-Raja

Thanks Raja great read!

[XBrytanX]

^^^ +1
Thanks

[l0ud_sil3nc3]

Thanks, Asus this board is a true work of art









Will pick up a 5820K as that's all that's available at local brick and mortar and ram should show up the next few days. Looking forward to firing this up

[bullant]

Wow looks very nice man,looking forward to results,will have to give X99 play myself later looks like a exciting platform

[Don_Dan]

I'm looking forward to seeing your DDR4 results, George!

[Dumo]

Great thread Raja

What a beauty. Thanks Asus

It will be fun

4X4 and 4X8 clocking next..

[Dark-Master]

Cheers from Thailand

I just got myself these...

I have no idea how to OC right now. I just jumped off from my previous i7 920 hours ago as the mobo was dying. This is my very first UEFI mainboard. And I'm sticking to this thread waiting for all the info.

[CryptiK]

How controlled are the voltage transients above set VCore during load fluctuations when LLC is enabled to give little/no droop under load? I don't mean by looking at CPUZ, I mean true assessment with a high polling rate DMM or scope. Old boards would spike well over set vcore during load fluctuations (heavy to light load specifically) - Raja can you comment on the X99 VRM in this regard?