Xeon E5-2600 V3 Turbo Boost Unlock

Method 1: Inject FFS driver straight into the BIOS.

This is a step by step guide of how to unlock Xeon E5-2600 V3 Turbo Boost frequency on all CPU cores using Chinese X99 motherboards and modified BIOS with a build in FFS driver. This method is OS independent and does not require EFI driver installation / re-installation.

Required files

Download archive required for this guide.

Hashes of the x99-tu-ffs.zip file:

MD5: 6eeeff1123513e08499283f3719f16d9
SHA1: 52299ba3f0988551bcdf9bc51b13e07c5c1d3ec6
SHA256: 04d32938b330328097b88b2b5c03b976e7a7a3dbe6adbdd9e66d1994f698ef66

To validate hashes of the file execute the following commands:

certutil -hashfile x99-tu-ffs.zip MD5
certutil -hashfile x99-tu-ffs.zip SHA1
certutil -hashfile x99-tu-ffs.zip SHA256

Google Slides Presentation


Method 2: Manual EFI driver installation.

This is a step by step guide of how to unlock Xeon E5-2600 V3 Turbo Boost frequency on all CPU cores using Chinese X99 motherboards and manual EFI driver installation. This method was written for and tested with Windows 10 Pro x64.

  • ❗ IMPORTANT: The step “6. Windows driver installation” does not seem to be required any more. First validate if Turbo Boost Unlock works without it. Install Windows drivers only if unlock does not work without them.
  • ❗ IMPORTANT: Use the V3DUAL.EFI driver for dual socket motherboards. Google Slides presentation has been updated, but the video guide is not possible to update.

Required files

Download required archive for this guide.

Hashes of the x99-tu.zip file:

MD5: 272621266752a1df69cfa57c88a3f2df
SHA1: e6c43b03cabf3d21b8b023f21c7d64c2fc09bc6e
SHA256: 88a9cc5d15d3ac59e896600c6c2706ca0cd99a55b45be6e18874df2b7c23d2bd

To validate hashes of the file execute the following commands:

certutil -hashfile x99-tu.zip MD5
certutil -hashfile x99-tu.zip SHA1
certutil -hashfile x99-tu.zip SHA256

Google Slides Presentation

Review of the Chinese X99 Motherboards

JingSha X99 D4

JingSha X99 Dual

Huananzhi X99-TF

Huananzhi X99-8M

Machinist X99Z

Runing X99 V1.2

PlexHD / Atermiter X99

Written by on
Views: 7506 Comments: 100
Вадим on 10 January 2020 12:16 says

Use afudos instead of FPT for Z99 v102. Other steeps are made by this instruction. Proof: CPU-Z Validation, userbenchmark.com/UserRun/23542225.

Miyconst on 10 January 2020 14:29 says

Вадим, do you mean Machinist X99Z? So, AfuDos worked for you? In my case AfuDos failed to write the BIOS.

Вадим on 10 January 2020 21:07 says

Did you tried with /gan key?

LUI Siu Man on 30 March 2020 11:25 says

I stumbled a thread on guide on win-raid, where the xx.efi file is inserted into his ASUS board ROM as xx.ffs usinf the UEFITool after the efi-to-ffs conversion.Link here: https://www.win-raid.com/t3874f16-GUIDE-Haswell-E-EP-Xeon-CPUID-F-Turbo-Unlock.html

I think this methods seems more convenient than loading the v3.efi with the USB EFI method, since the overclock process is done on startup, and the boot process into the EFI is not required anymore

Do you think it is doable with those Chinese board? I hope you can have a try :)

Miyconst on 30 March 2020 11:30 says

LUI Siu Man, I plan to try this method to validate Turbo Frequency Unlock on Ubuntu, but I am not sure when I will have time for it.

LUI Siu Man on 30 March 2020 16:07 says

I have successfully tested the ffs method and it worked! I put the v3.ffs and another .ffs linked by the ASUS thread to the ROM.

It is done on my Huananzhi X99-F8 with E5 2678 v3. All cores at x33 mutiplier

One important thing is that for some reason, or maybe just me and some others, Windows 10 will lockup very very soon if I just delete the microcode . I had to set the "IntelRCSetup> Advanced Power Management Configuration>Power technology" to "disable" to boot normally into Windows 10 (not sure about other OS). I used AMIBCP v5.02 and set the "CPU C State Control" to USER so I could disable the C3 and C6 report in the UEFI, like what the user did to his ASUS board.

My resource x99_ffs_v3.rom, guide & notes.

The X99_ffs_v3.rom is the rom I use now. And the document is a messy but useful steps I took to achieved the ROM mod.

Miyconst on 30 March 2020 22:20 says

LUI Siu Man, thanks a lot for the info! This will be the next thing I try out, if it works for me as well - then the Turbo Boost unlock procedure will be way more stable and easy.

Benyba on 06 April 2020 15:44 says

Great content thank you! I'm facing issues with x99-tu-ffs.zip archive which seems corrupted every time I download it.

Also regarding method 2 and V3DUAL.EFI I systematically get the message "V3DUAL.EFI is not an image" when trying to load it while V3.EFI loads correctly (but I've a dual CPU system). Checksum is correct on this archive so I'm not sure what I'm doing wrong here, any idea ?

Miyconst on 06 April 2020 16:10 says

Benyba, it seems like my hosting provider is having some issues, thus the file gets corrupted while download. I have add extra download options via Microsoft One Drive. Please validate.

Benyba on 06 April 2020 17:28 says

Miyconst, and it works ! Method 2 confirmed on Dell Precision 7910. For some reason injecting FFS driver was not enough and I had to manually install the EFI driver. Still no luck with V3DUAL.EFI but using freecableguy's v3x4 did the job. Also had to use service mode jumper to flash the modified bios. Thank you so much, you just made my day!

Miyconst on 06 April 2020 18:03 says

Benyba, thank you for the info! I will double check the V3DUAL.EFI issue, maybe I have included a wrong file. Where did you download the freecableguy's drivers from? I am usually using Payne's drivers. Also, which FFS driver did you inject?

Benyba on 06 April 2020 18:51 says

Miyconst, the driver I manually installed is from here. For some reason I wasn't able to load any other ones like your v3dual.efi or v3x2_payne_0_0.efi. The FFS driver I injected is the v3x2_payne_0_0.ffs you provided. I still have something slightly under optimized here as my max turbo seems to be stuck at x35 while in full load now while it should theoretically be 36x on 2697v3.

Miyconst on 06 April 2020 19:11 says

Benyba, one more thanks, I will add your link to the list of resources for others to follow. Xeon E5-2697 V3 is a 14-Core monster, but its TDP is limited to 145 Watts. All Xeon E5 V3 CPUs reduce the clock speed to stay within the TDP package limit even with unlocked Turbo Boost. If you reduce the CPU/Memory controller voltage you might be able to get full 3.6 GHz on all cores under full load.

Benyba on 06 April 2020 19:16 says

Miyconst, got it, I'll fine tune this then! Thanks again and stay safe wherever your are!

Alejandro on 06 April 2020 20:26 says

Hello Miyconst, thank you very much for all your work, I like your content a lot, you`ve got yourself another Brazilian fan.

Given your new permanent boost turorial, Iam curious as to the need to remove microcode updates from the system (windows and linux), because I imagine that it would be applied again on boot if it isnt removed from the OS right? (as per your method 2).

Also, if possible, I think it would make for great content if you explored removing spectere and meltdown microcode paches, talked about what malware actually uses them (I have searched and found no metion if those were exploited by anyone in real life (in a feasable manner)) and the performance gains by doing so.

Thanks again for the awesome content! and stay safe

Miyconst on 06 April 2020 21:08 says

Alejandro, thank you for the support. Regarding your questions:

  • It does not seem to be needed to remove CPU micro-codes from OS any more. Even with the manual EFI driver installation, I no longer remove the mcupdate_GenuineIntel.dll file. Only in a few rare cases I had to do that.
  • As far as I know, all Chinese X99 motherboards do not have the latest Spectre/Meltdown updates from Intel and with branded motherboards it's very easy to remove them - use a previous version of the BIOS.
Alejandro on 08 April 2020 01:02 says

Miyconst, thank you very much for the assistance, I will start exploring this topic as soon as my motherboard arrives.

As a last question (sepecific to linux since I pretend to use it), I read in the Arch wiki while researching this topic that the microcode patches are loaded by the OS as soon as the system starts, why would this no be the case with these chinese motherboards? link:https://wiki.archlinux.org/index.php/Microcode (first section explaining the early loading procedure is what bugs me the most)

Thanks again for your help

Miyconst on 08 April 2020 01:08 says

Alejandro, unfortunately, I am quite bad at Linux and the boot process, thus I can't give you a clear answer. From what I found online, is that the "trick" happens before any CPU microcode is loaded, it "fools" CPU to use Turbo Boost settings of one CPU core on all Cores, after that CPU microcode can be loaded. Personally I have tested Turbo Unlock on Ubuntu 18.04, but please keep me updated after you validate Arch Linux.

Alejandro on 08 April 2020 17:34 says

Miyconst, as soon as I recieve the motherboard I will be testing Manjaro, and because I made the mistake of buying the plex HD X99 (I bought it before watching your videos, and the sller only had good reviews) I will be reporting back on that as well, do you know the version of your reviewed plex HD X99 board? and it`s BIOS version? this is where I bought mine

also, I bought a programmer, and read somewhere that FPT doesen not read/write the whole BIOS, is that true? where can I read more about how that works?

Miyconst on 08 April 2020 20:51 says

Alejandro, don't be too sad about Plex HD X99, yours might be a better unit. I have got multiple different reports, for some this board works well, for others it has all sort of issues. Hopefully, you have got a good sample. Currently I am super busy with my work and other videos, but I still plan to validate Plex HD X99 with Huananzhi BIOS and see if that's gonna work and/or improve its performance. This GitHub repository might give you some answers regarding FPT.

Joseph on 09 April 2020 15:25 says

Miyconst, I am trying to follow your steps on a evga x99 micro 2 with xeon e5 2660 v3. When I try to run dump.bat it tells me "Error 7: Hardware sequencing failed. Make sure that you have access to target flash area!" Any advice you can give me?

Miyconst on 09 April 2020 15:30 says

Joseph, with branded motherboards you have to use proprietary utility to read/write BIOS, the FPT tool does not usually work with EVGA/MSI/ASUS and other branded boards.

Joseph on 09 April 2020 15:53 says

Miyconst, after running it again it was able to export the bios files and later flash it back. After a reboot I am getting the full 3.3 turbo on all cores. Thanks for all your help! Your guide/video made the process very straight forward!

Miyconst on 09 April 2020 15:55 says

Joseph, wow. So, you used FPT tool on your EVGA X99 baord? It's good to know that it worked well. Having a branded motherboard you can also try to increase BCLK to something like 105-107 MHz.

Joseph on 09 April 2020 15:59 says

Miyconst, Yep! At first it failed with the error code, but when I closed/reopend cmd and ran the bat file again it worked. It was strange, but everything is working correctly. I will try increasing my BCLK next!

Miyconst on 09 April 2020 16:02 says

Joseph, sounds great!

Joseph on 09 April 2020 17:08 says

Miyconst, Set the BCLK to 103.84 (will slowly push further until I find the highest stable BCLK) and it is running stable at 3426.81mhz on all 10 cores!

HotnikQ on 15 April 2020 02:43 says

Thank You for share these content with us! i use fptw64.exe to flash my bios.

  • Huananzhi X99-TF 2680V3 v3_payne_90_50 Success.
  • Huananzhi X99-F8 2690V3 v3_payne_90_50 Success.
Miyconst on 15 April 2020 07:32 says

HotnikQ, thank you for the info. On your screenshots you show macOS, did you manage to successfully install macOS Mojave on Huananzhi X99-F8?

Niko on 15 April 2020 07:36 says

Did u know that on x99-tf after going to sleep and back again, unlocking turbo boost has no effect? You should turn reboot once again to take effect.

Miyconst on 15 April 2020 07:39 says

Niko, yes, I have seen similar reports, but didn't have chance to validate myself, yet.

Niko on 15 April 2020 07:52 says

Also after reboot windows cant start properly such like there is no bootable device. And u gave to enter bios and set defaults again.

Mikey on 20 April 2020 11:03 says

Doesnt work with MacOs ( Hackintosh )...

Miyconst on 20 April 2020 11:04 says

Mikey, do you have a bit more details? Which motherboard, which CPU, which BIOS, which driver. How exactly does not work? Does Hackintosh work without Turbo Boost Unlock?

Mikey on 21 April 2020 00:18 says

Miyconst, x99-tf with your modded bios ( incl tubro boost unlock + driver), 2678v33 opencore build. MacOs can work with it for sure.

Also will u fix issue about turboboost and sleep?

Miyconst on 21 April 2020 00:21 says

Mikey, thank you for the info. I am not sure how to fix Turbo Boost issue after computer awakes from sleep, someone with better knowledge of how all of these things work is needed.

Riko_crimea on 22 April 2020 11:49 says

А есть уже готовый БИОС, чтоб прошить и ничего не добавлять вручную? Дайте ссылку, пожалуйста. Материнская плата - x99-tf, xeon 2678.

Is there a ready-made BIOS to flash and not add anything manually? Give me a link please. Motherboard - x99-tf, xeon 2678.

Riko_crimea on 22 April 2020 13:22 says

Riko_crimea, я нашел. Оказывается, ссылки на последнем слайде в Google Slides Presentation. Я думал это видео. Спасибо!

I found. It turns out the links on the last slide in Google Slides Presentation. I thought this video. Thanks!

Miyconst on 22 April 2020 16:58 says

Riko_crimea, I have got some reports that there are two revisions of Huananzhi X99-TF, for some my BIOS works, for others it does not. That is why you shall always have a backup of your original BIOS, then, if my BIOS does not properly work on your motherboard, you can restore you original BIOS.

Riko_crimea on 22 April 2020 18:25 says

Miyconst, I set yours - while it works fine. If there are problems, I will write.

Nicola Sanna on 23 April 2020 00:16 says

Hi Miyconst, I'm Nicola from Italy. I just buy a Jingsha X99 Dual with two xeon E5 2678 V3 and I'm trying to unlock the turbo boost ... I've tryed both procedure ... with inject the ffs driver in the bios I didn't have any success while, with manual EFI driver installation : if I use del V3DUAL.EFI the system don't recognize the file valide... only if I use the V3.EFI I obtained that the only one CPU has turbo boost unlock ... what can I do for have both CPU with turbo boost unlock?

Thanks a lot for your kindly
Regards

Miyconst on 23 April 2020 14:30 says

Nicola, hi. Please try to use the v3x2_payne_0_0.efi driver, it might work with your motherboard.

Nicola Sanna on 23 April 2020 19:39 says

Miyconst, Hi. I've already tryed the v3x2_payne_0_0.efi and the result was negative. I don't understand .. the V3.efi for single CPU works good ...

Miyconst on 23 April 2020 22:55 says

Nicola, what do you mean by "negative"? It did not turbo boost or did not boot?

Nicola Sanna on 23 April 2020 23:04 says

Miyconst, No turbo boost ... the system respond V3x2_payne_0_0.efi is not an image the same message that for V3DUAL.EFI ... instead with the V3.efi only one of the two CPU go to Turbo Boost unlocked ...

Thanks a lot for your help

Miyconst on 24 April 2020 10:31 says

Nicola, that sounds really wrong. Could you please try to download the drivers from their original source - https://peine-braun.net/public_files/XEON_V3_BIOS_MODS/EFI-Drivers/dual CPU/? It could be that the files get corrupted when you download from my website.

Nicola Sanna on 25 April 2020 11:25 says

Miyconst, HI, thanks but as I said I've alredy tried the V3x2_payne_0_0.efi and It didn't works ... I finally found another driver EFI on web (V3x4-0.10b-i306f2-rc9_90_70_50) that it works very well and now I have both my CPU Xeon E5-2678 V3, 24 cores at full turbo ... Thanks a lot for your help.

Dimitri on 25 April 2020 13:11 says

Many thanks for the guide! I can confirm it working in the X99-TF with a e5 2678v3 using -30 mV at VID and -20mV at mem (your -30 - 20 file :D ). However, I am wondering what is the lowest undervolting this CPU could support. I could try it myself but, do you know what would be the worst case scenario if I get the VID too low? I just can't fry this system but I need it to be just a bit faster. Thanks again for your helping!

Riko_crimea on 25 April 2020 15:24 says

Miyconst, Hello! I have a question. The computer boots up for a very long time (BIOS) - maybe you know the solution to this problem?

Miyconst on 25 April 2020 17:21 says

Riko, hi. Unfortunately, long booting time is a common problem of the Chinese X99 boards with modified BIOS, I am not aware of any solution for this problem, yet.

Miyconst on 25 April 2020 17:22 says

Nicola, that's very good to hear. Could you please share the exact link to the file you were able to successfully use?

Miyconst on 25 April 2020 17:23 says

Dimitri, thank you for the update, it helps me with collecting statistics. You system will hang if you reduce the CPU voltage too much, but it definitely will not fry.

Nicola Sanna on 25 April 2020 21:22 says

Miyconst, This is the link of efi driver that I use with my Jingsha X99 Dual and two Xeon E5 2678 V3 ... It works very well ...

Michel on 27 April 2020 15:40 says

Hello Miyconst, Thanks for the amazing detail work you are producing.  So much good info for pc bilders... I did tried to follow your steeps. I purchased a Machinist X99Z V102, paired with a E5-2640 V3.  I was having stability issues at one point.  I notice that a RAM stick was not properly installed.  I touch it when the power was ON. (I know that is stupid) and the system died.Now the board is dead, no power no fan spins, no led lite ????   Do you think the board is now finish? Do you think I can flash the Bios?

Thank again. Michel from Canada

Miyconst on 27 April 2020 15:49 says

Michel, I am sorry to say that, but it seems like something got fried on the motherboard. Restoring BIOS most likely won't help here. If you touched the motherboard heatsink and it died - it could have shortened some contacts, thus, you can try to completely remove the heatsink (it's electrically conductive) and start up the motherboard. If no signs of life - then your best bet would be to replace the motherboard and hope that it didn't fry up your CPU/RAM.

Michel on 27 April 2020 16:20 says

Miyconst, Thanks for the quick response. PS. I did'nt touch the heat sink. I just moved the RAM stick a bit. ???

Michel on 27 April 2020 16:47 says

Miyconst,

So you think my CPU could be fried buy moving the RAM? So If I purchase a new board, Can I test the CPU and RAM on that new board? or will the damaged part induce more damage to the board?
You think it is safer to get rid of all the kit?

Thanks a lot. Michel

Miyconst on 27 April 2020 18:53 says

Michel, I am sorry for not reading your message properly, mixed your issue with other guy's issue. In your case I would recommend you remove all memory sticks, install just one, make sure it's properly installed, and try to start. The motherboard should show signs of life even without memory installed.

Also, I don't think your CPU is fried, it could happen if you would shorten some wrong contacts, but you didn't seem to to do anything like that.

Michel on 28 April 2020 23:00 says

Miyconst Thank a lot for the responses. I really appreciate. I think my board is dead. Based on your advises I will order a "Kllisre X99 D8" from your link to support your channel (I think it is your best recommendation.

Two last technical questions if you have 2 minutes:

A- Is the life spend of a CPU like E5-2640 V3 reduced by keeping the turbo mod on all the time ( do you think the cpu will last 2 years ?) B - In two weeks, when I will receive my new board, can I try my RAM stick in? Can a damaged ram stick damage my new board?

Thanks again

Miyconst on 29 April 2020 01:43 says

Michel, please always validate if you can find a better deal on AliExpress, if there is a significant price difference between my affiliated links and other offers - don't buy from the affiliated links.

Life span of CPU won't be significantly affected with Turbo Boost unlock, these CPUs are designed to work 24/7 for decades, even under super extreme conditions, your CPU shall work for 2 years and more.

I think, you can try your memory module in the new motherboard, but first validate the motherboard with another RAM stick.

JukkaX on 01 May 2020 14:40 says

Dimitri, Whaat? There's a -30 mV VID BIOS to download in smwhr?

Dimitri on 02 May 2020 00:24 says

JukkaX yes there are ffs and efi files to lower the VID, just as sir mi8 instructed. By the way, my e5 2678v3 with x99-tf works amazing with -70 cpu and -50 mem. I get solid 3.3 Ghz all cores at 119 W TDP max. It t also runs Arch very well (no longer need to use a virtual machine in windows to take advantage of the faster clocks in linux). Really good mod that deserves its praise

Marcos on 02 May 2020 02:22 says

Hi! I have managed to get the turbohack on the old way working (Kllisre x99 d8 and 1630v3 here), I'll be trying your method on the weekend. I do have a question though, is there any way to change BCLK on these chinese x99 motherboards?

Cheers from Brazil!

Miyconst on 02 May 2020 10:25 says

Marcos hi. I am a bit surprised that Turbo Boost unlock changed something for Xeon E5-1630 V3. Could you please describe what was the CPU behavior without the hack and with it? What are the scores in Cinebench without and with the hack? This hack is supposed to work on E5-2600 V3 & E5-4600 V3 CPUs only. Regarding BCLK - no, there is no way to adjust it in BIOS on the Chinese X99 boards.

Brian on 02 May 2020 12:34 says

Hi, may I know where can I download the lazy pack for Huananzhi X99-F8 (stock voltage) as mentioned in your video?

Thank and Regards, Brian

Miyconst on 02 May 2020 12:49 says

Brian, hi. The "lazy" pack is available for download from the last page of the Google Slides presentation.

JukkaX on 02 May 2020 19:39 says

Dimitri, Got it! -50 mV now and seems to be solid. Helped to get Cinebench R20 go at 3.1 GHz instead of 3.0 Ghz (@ -20 mV). In CB R15 VID is 0.919 V but in R20 it's 0.963 V.

Marcos on 02 May 2020 20:20 says

Miyconst, before the hack it'd go to 3.7Ghz all cores, now they go up to 3.8 Cpu-z and cinebench r20 got a range of improvement from 2 to 4%. Trying to mess around with Intel XTU and found out there's some throttling from current limit... Is there anything that can be done so that current doesnt limit the processor? Hwmonitor only shows Max 80w being drawn by the package and no heating problems...

Miyconst on 02 May 2020 20:49 says

Marcos, E5-1630 V3 has maximum turbo boost frequency of 3.8 GHz, there is no way you can bypass this limitation.

Marcos on 03 May 2020 01:40 says

I see! But if XTU is throwing this "Current/EDP Limit Throttling: Yes" during a full load on the CPU doesn't it mean that there is something holding the CPU back?

JukkaX on 03 May 2020 13:13 says

Now at -90 mv for the core and Cinebench R20 runs at 3.2 GHz all the way through :-) During R20 VID is 0,948 V and R15 0,874 V (CPU-Z) LinX runs at 2,6 - 2,7 GHz and yields 360 GFlops with relaxed DDR3 1866 timings. Will start experimenting with lowering the memory controller voltage next.

JukkaX on 03 May 2020 14:05 says

-90 mV soon proved to be unstable on my 2678 v3; PUBG crashed twice in a short time.. Gotta to go back to -80 mV.

Dimitri on 10 May 2020 04:50 says

I can confirm that the CPU goes back to its stock speed and voltages if the computer is left to sleep. I am running an e5-2678v3 on a x99-TF DDR3 1866 (-70,-50 mV as stated above). A quick restar solves the issue but the system seems to take twice as long to start up.

ystal on 15 May 2020 00:00 says

Alejandro, Did you managed to do turbo boost unlock on Manjaro?

Marco on 15 May 2020 17:50 says

Hi, thanks for the great guide; one question I flashed the modified Huanhanzhi X99 F8 bios on a Kllisre D4 and everything works as expected but one core is stuck at 80% load and I get poor single core benchmark results; did you experience any of that? Thanks

Miyconst on 15 May 2020 17:56 says

Marco, hi. What is your CPU? Did you by any chance install Windows drivers from the previous Turbo Boost unlock method?

Marco on 15 May 2020 18:14 says

Miyconst, Hi Miyconst, thanks for the reply; my cpu is a Xeon 2620 V3, no I directly flashed the modded bios. I must say that I'm using an old windows installation so I should try with a clean install or linux and report back, just wanted to know if it was a common problem

Miyconst on 15 May 2020 18:23 says

Marco, yes, I have seen this problem with this exact processor a few times. In my case it was caused by installed custom cpumcupdate driver. In your case it's probably caused by either dirty Windows installation or incompatible FFS driver. Did you take my BIOS or prepared yourself?

Marco on 15 May 2020 19:00 says

Miyconst, I took your prepared bios for the F8 with 20 mv undervolt, glad to hear it's a common problem. I'll try to prepare another bios with another the FFS driver; thank you for your support.

Julien on 16 May 2020 12:26 says

First, thank you for knowing that you share here. It is very interesting to read your tests. Have you already tested the Kllisre x99 dual Processor? I saw that on the Jinshua dual test, the gain is ridiculous and you advise to not unlock the boost.

My configuration is : Kllisre x99 Dual 2 xeon 2678 V3 2 XFX RX 580 8GB 8 DDR 4 8go 2667Mhz Kllisre 2 SSD Nvme Corsair MP510

Everythings in open air.

Do you think it will be the same for me and and that I would have no gain with the Turbo Unlock ?

Miyconst on 16 May 2020 12:50 says

Julien, I have not received the dual socket Kllisre X99 motherboard yet, when I receive it - it will be tested on my channel. In general, I strongly advice not to build dual socket configurations, unless you use specialized software which can leverage this many cores & threads. Turbo Boost Unlock has the best impact in games, because in games CPU does not exceed its power consumption limitations and can stay at the highest possible frequency all the time, but dual socket configurations are terrible for games on its own.

Julien on 16 May 2020 14:36 says

Miyconst, I mount this configuration because my current PC has trouble making 4k video editing smoothly and to play too. My current config: Ryzen 2600x on MSI B450 Pro, 16go DDR4, rx 580 8gb and ssd MP510 I will wait for your test before I do anything. Thank you very much and good luck with your Youtube channel !!!

Polar on 22 May 2020 15:50 says

Miyconst,thank you for your great work! Great tutorials! I tried to unlock the turbo boost on a Supermicro X10DAX dual socket board with two E5 2683 v3 cpus (ES/QS, stepping 2, according to Cpu-Z). I used both the dual socket ffs-file provided in your ffs-files, as well as a home brewn ffs-file based on V3x4-0.10b-i306f2-rc9_90_70_50.efi. I used the newest bios for X10DAX (rev. 3.2) as a staring point. After the modifications according to your tutorials, I used SuperMicros own software to flash the bios. No success in both cases. Modified bios do work, system remains table, but the turbo works not we would like. E5 2683 v3 should boost up to 3 GHz, and they do usually in my system, though not all core. After this bios mod, with either of the two ffs-files, the frequency of all cores remained limited to 2.5 GHz. So, instead of "all core unlock" to 3.0 GHz, I got a "all core lock" to 2.5 GHz. Do you have any suggestions? Thank you!

Miyconst on 22 May 2020 16:22 says

Polar, hi. Unfortunately, I don't have any experience with the Supermicro LGA 2011-3 boards. In your case I would suggest you to start with manual EFI driver installation. Create a BIOS without 06F2 CPU microcodes and experiment with different EFI drivers, see if any of them is going to work. Try to enable/disable CPU C3/C6 states, on some boards it has to be enabled to make Turbo Boost Unlock work. Also experiment with "energy efficiency" settings in the BIOS.

Polar on 22 May 2020 16:28 says

Miyconst, thank you! Great suggestions. Re: settings: yes, these were exactly the settings I have been dabbling with.

Polar on 23 May 2020 17:34 says

Miyconst, success! Thank you very much! (hardware see above, Supermicro SM 10DAX, 2x E5 2863v3, ES/QS, stepping 2, Win 10).

Manual installation worked, ass suggested, without the last steps of removing stock drivers. Basically all I needed was a modified up to date bios file without the 06F2 microcodes as well as the 3VDAUL.EFI - driver file. Loading the driver through in the efi shell worked, however, the efi shell provided by SM does not support the bcfg-command. Therefore at this point, I have to load the driver on each restart of the computer. Adjustment to the bios were done manually, as the provided tool to change settings permanently did not work well in my case. It was absolutely necessary to disable to C6 state report in the advanced power settings, otherwise the system would hang on startup. Power management settings are well hidden in the bios interface, under CPU settings, you have to scroll down and down.... Advanced power settings must be set to "custom" in order to allow to modify the C6 state setting. Many other bios settings I explored seem to have no specific impact on the turbo unlock. It is possible to combine this mod with the provided blck overclocking for this board (Hyper-Speed).

Results: CPU 0 showed a max multiplier of x29 (no idea, why, even with no load it would never reach 30), CPU 1 was able to boost up to x30. Overall probably a 12-15% performance gain was observed in CPU heavy applications. Frequencies were, obviously not fixed at max turbo, but changed with load. I tried again to incorporate the V3DUA.EFI into the bios.rom, converting it to ffs, but to no avail. x multipliers again were limited to x25. Thermals and power consumption increased, with the combined speed improvement, but remain reasonable.

Miyconst on 23 May 2020 18:33 says

Polar, thanks for the update! It seems like the Supermicro board is loading the DXE drivers in some other or different order, which makes Turbo Unlock not function properly. At least manual EFI driver installation worked well. To get a few extra % of the performance you can try to reduce CPU voltages and see how that works.

Polar on 23 May 2020 20:55 says

Miyconst, btw, as I mentionned SM EFI shell does not support the bcfg-command to add drivers. My quick and dirty fix is to simply set the EFI shell as first step in the boot process, hd as second. Using the Shell on the HD fs0: I created the file called "startup.nsh" with the command "edit startup.nsh" The file contains just simply load command and exit. load fs0:\efi\boot\V3dual.efi exit Works. :) Just a workaround. I guess there is a more professional way to do this. :)

Polar on 24 May 2020 11:06 says

Miyconst, tried different setting for the undervolting settings and it helped further improving performance and reducing power draw. Using the Payne 50-50 .efi file, both cores now worked at x30 multiplier. I have been using a moderate bclk overclocking setting 2 of SM Hyper-Speed available for the SM X10DAX motherboard, resulting in a constant 3.07 GHz on all 28 of the two e5 2683 v3 CPUs' cores under full load. Even tried 90-90, and the system still worked like a charm under full load for hours. No critical values nowhere. Not one sign of instability in the system, no restart, no nothing. :) Question: How would you go about finding the ideal setting for the undervolting?

Miyconst on 24 May 2020 11:11 says

Polar, I have two methods for CPU undervolting:

  • Lazy one: just apply -50mv on CPU and CPU memory controller. Works in 99% of the cases. Only absolute worst Xeons can't tolerate such voltage reduction.
  • Smart one: reduce CPU voltage until the system is not stable under one of the following tests: Prime95, Aida64, Cinebench R20, Corona. Add 20mv to the unstable value and stay happy. For example, if a system becomes unstable at -100mv, I leave it at -80mv.
Polar on 25 May 2020 08:52 says

Miyconst Thank you again for your help! To summarize, I was aware of this mod for quite some time, but I was intimidated by possible complications. As it turned out, with my branded and trusted Supermicro motherboard SM X10DAX and two XEON e5 2683-v3 chips, it was quite easy. Basically just removing one line of microcode patch from the bios using the respective tool and adding the cpu-driver in the UEFI-shell was all it took. Bios updates are very easy on these SM boards. You can flash a new bios right from the UEFI-shell. No bootable external medium needed. Very few modifications in the bios settings very necessary, basically only disabling the C6-state monitoring was critical. Undervolting using the 100-50 efi-file from Payne was very helpful mostly improving performance and to a lesser extent reducing power draw. I have been using the SM X10DAX board with these Xeons chips since 2016 for heavy duty calculations, often 24/7 and I have to say Supermicro lives up to its reputation of making rock solid products with a great longevity. Practically no restarts need, even with 24/7 use and 100% load bearing tasks. Software maintenance needed was minimal. The board offers ton of features, probably more pci slots you will ever use, sli and baseclock overclock capabilites and much more. The board is big, and a non standard form factor, prepare your drill :), and still expensive at around 400$ to 450$, but I have been very happy with this choice. This is a great investment, giving you performance, reliability and peace of mind. :) Now with a performance gain of about 15% in my main applications, my workstations continues to work on with no hickups at all, purring happily. :) Thank you!

Miyconst on 25 May 2020 12:17 says

Polar, I am glad to hear that it worked well for you and thank you for providing feedback with additional information. I am sure, someone else with a Supermicro board will benefit from your findings.

AzzY on 29 May 2020 00:23 says

Miyconst, great tutorial. Thank you!

I followed your guide but my CPU can't reach full turbo multiplayer. My CPU is 2678v3 and it stays at 32x instead of full 33x. My motherboard is Huananzhi X99-TF. Do you have any suggestions?

Miyconst on 29 May 2020 09:18 says

AzzY, does your CPU go to 3.3 GHz when only one CPU core is utilized? If yes - then it's a normal behavior. In order to increase CPU frequency to it's maximum 3.3 GHz you may try to reduce the CPU voltages. Not every Xeon E5-2678 V3 can stay within its power envelope at full load and 3.3 GHz. If your CPU never goes up to 3.3 GHz - then you have to try some other FFS driver.

Jose on 29 May 2020 10:23 says

Hi, is MACHINIST X99Z bios compatible with Kllisre X99Z?

Miyconst on 29 May 2020 10:35 says

Hi Jose. As far as I know, these are the same boards, just a different name, and the BIOS is compatible as well. Still, the risk is yours.

евгений on 30 May 2020 18:05 says

Спасибо делал также только с руским переводом на другом форуме не вышло а с тобой вышло

Miyconst on 30 May 2020 18:07 says

евгений, я всегда рад помочь.

AzzY on 31 May 2020 12:59 says

Miyconst, Hi Miyconst, finally I got it what this FFS drivers do after your explaination. . Thank you very much - I am fighting with this CPU for 2 Weeks now. After your clarification I injected -70 cpu and -50 mem as somebody mentioned above and my CPU is now working on 3,30GHz on all cores even when stress tested! Now I have 2 more questions: Where can I buy you a beer? (PayPal, Monese, Revolut or something else) How to OC the BCLK in the BIOS? I am not sure I see the settings there. I want to increase the BCLK with 5% to achieve 3,465GHz. Thank you in advance PS: If possible please delete my post above. I forgot to write my nickname and there is my e - mail now :(

Miyconst on 31 May 2020 14:33 says

AzzY, no worries, I have deleted your previous comment and I am glad to hear that slight CPU voltage reduction helped with achieving 3.3 GHz under full load. You can buy me a beer through a PayPal Donation, which is not required, but very much appreciated.

Regarding the BCLK overclock, as far as I know, not a single Chinese X79/X99 board supports BCLK overclocking via BIOS settings, thus I am quite confused when you say that the settings are there. Could you please share a screenshot with such settings?