What controls Turbo Core in Xeons?

[C-Power/Tw0tch]

Hi @Dufus

Thanks for the reply, I am not terribly familiar with the UEFI Shell, but I am a quick learner
Started to read up on it a bit more, looks like a dos/unix sort of system, so I am sure that with some tinkering I could get somewhere at least lol.

Thing is, I have been overclocking since the pentium 1 era (when you could still use jumpers on the MB to change the FSB lol).

Until lately I have always been busy with unlocked CPU's so it's been easy for quite a while now, but my priorities/work have changed a lot and now I'd really like to tinker with these Xeons
It's just a completely different ballgame with modding biosses in ways I've never done before.

It seems I have a lot to learn when it comes to these type of bios/efi shell changes, but it's a nice challenge for a change!

In short though,
I'm kind of lost with what The Stilt is saying about locking those registers, I get what it does, but I don't understand how/where to actually do it.
So my guess is that RU is a shell where that is supposed to be changed in, but honestly I have no clue what to do with the RU files in the first place (although I guess that's what you load before Windows starts).

Could not find a guide on the site itself, so here I am hoping for some enlightenment from the Bios modding guru's

 

[C-Power/Tw0tch]

Ok here are some screenies etc.. Stuff that is boxed in red in the bios and windows get's completely ignored when set, or the CPu just goes in a slow mode whn it doesn't like it and get's stuck at x21 or x23 multi, with very poor performance, and some sort of C-State bug. (Hence I get it - you need to load stuff/change things in the efi shell, but that's where I am stuck atm )

Just added the screens with the no-microcode loaded bios and windows.. but still no luck. So I guess I really need to get my head around this EFI shell business.. any pointers (or pointers to guides) are very welcome (I'll start google-ing some more now to). I am certain there are more people in this thread that have the same "issue" and would love to hear from the modders too ^_^











Thanks for the help and reply's guys, much appreciated!

 

[Dufus]

@C-Power/Tw0tch Some BIOS already incorporate the UEFI shell as part of the BIOS but if we don't have that then we can download one and put it on a FAT32 USB flash drive or other suitable boot medium and boot that.

Here's a link to a shell (shell.efi). Should be okay for what we want.
https://github.com/tianocore/edk2/tree/master/ShellBinPkg/UefiShell/X64

1. Rename shell.efi to Shellx64.EFI and place on root directory of a FAT32 flash drive.

or

2. Rename shell.efi to BOOTx64.EFI then create a root folder named EFI and sub folder of that named BOOT and copy the BOOTx64.EFI to there.

Use the BIOS boot utility to boot the FAT32 flash drive in UEFI mode. You can type "help" for a list of commands. One of the most useful options is -b which will pause each time the screen is full, "help -b"

Get use to the "map" command and file systems (FS0:, FS1: and so on)
While your exploring that I'll put something together.


Thanks for the screen shots. Getting rid of the microcode and setting all cores to 30 should have been enough really. We are not overclocking so much other than maybe a little bclk so that extra core and cache voltage should not be necessary. You really want the opposite, less voltage.

Set CPU settings back to default, with the microcode update removed from the BIOS. All core at 30 should be okay if enacted upon, maybe try setting each core to 30 if that option is available in the BIOS. Let Windows load the microcode update as usual or use the vmware option mentioned earlier.

 

[The Stilt]

Remember that you cannot make any changes to the CPU settings (MSRs, power management and especially functions which are used through the "OC Mailbox"), after the CPU has been initialized.
Doing so will lock the CPU either to default or to LFM mode (happens with µCodes with "LFM Bug" present).

Remove the microcode from the bios and set the desired settings. Forget about running ThrottleStop, let alone trying to change any of the settings with it. Use CPU-Z & HWInfo for monitoring, they don't tamper with any of the settings.
If it still doesn't work, also try leaving as much of the bios settings to default for starters. ASRock might be changing some of the settings after the CPU has been initialized, which would cause the CPU to lock in default / LFM mode.

 

[C-Power/Tw0tch]

@Dufus @The Stilt

Thanks guys! Will be going at it today again and see if I can get somewhere further with it hehe.

Thanks Dufus for that little guide, very very useful indeed, that's pretty much what I was after in terms of actually getting further with my experiments

The Stilt, Thanks for the tips, I figured as much with bios settings when the CPU kept doing the opposite of what I thought it would do when changing any voltage/EIST/C-states settings lol.

I'll continue to experiment and will report back sometime later again

 

[GTZ]

Just read this thread and would like to say thanks!!!

I can't wait to try this with my 2683. One question, do you remove the microcode from the BIOS with the VMWare tool? Or is that just for injecting microcodes to Windows?

Thanks in advance

 

[C-Power/Tw0tch]

I'm back!

Anyway, what I have managed to do so far (but all with no actual results):

1. Used UBU/MMtool to remove the uCode from bios (Using MM to remove code and save unsigned bios, then use UBU to save it so it can be flashed)

2. Have windows loading different uCodes with the VMware updater (Tried 0x27, 0x36, 0x38, and 0x39)

3. Got into EFI Shell and figured out a thing or 2 (the "connect" command seems the useful one for loading a driver)

4. Got into RU.efi, have had a look around, but I have no clue where I can find those registers @The Stilt is talking about, I did find the MSR Registers, but no idea what aex or edx mean (or how and where to find and change them, or what bit 20:20 means, for that matter).

Having all bios settings at default seems to just have it run at stock turbo bins, having it set to "All Core 30x" seems to do nothing, and setting each multi to 30x per core also doesn't change the CPU's behaviour.
Having windows load different Microcode doesn't help either

If anyone would be so kind to give me some quick pointers of where/what to look for in Ru.efi - that would be highly appreciated.
If it doesn't want stick at 30x multi after that.. I don't know what else to try, maybe that driver @Dufus is brewing on his stove

Thanks again guys, i've learned a ton from this thread already, and I enjoy this journey into the unknown a lot (thank god I have dual bios to flash back a broken one LOL )

 

[GTZ]

I'm back!

Anyway, what I have managed to do so far (but all with no actual results):

1. Used UBU/MMtool to remove the uCode from bios (Using MM to remove code and save unsigned bios, then use UBU to save it so it can be flashed)

2. Have windows loading different uCodes with the VMware updater (Tried 0x27, 0x36, 0x38, and 0x39)

3. Got into EFI Shell and figured out a thing or 2 (the "connect" command seems the useful one for loading a driver)

4. Got into RU.efi, have had a look around, but I have no clue where I can find those registers @The Stilt is talking about, I did find the MSR Registers, but no idea what aex or edx mean (or how and where to find and change them, or what bit 20:20 means, for that matter).

Having all bios settings at default seems to just have it run at stock turbo bins, having it set to "All Core 30x" seems to do nothing, and setting each multi to 30x per core also doesn't change the CPU's behaviour.
Having windows load different Microcode doesn't help either

If anyone would be so kind to give me some quick pointers of where/what to look for in Ru.efi - that would be highly appreciated.
If it doesn't want stick at 30x multi after that.. I don't know what else to try, maybe that driver @Dufus is brewing on his stove

Thanks again guys, i've learned a ton from this thread already, and I enjoy this journey into the unknown a lot (thank god I have dual bios to flash back a broken one LOL )

Thank you for posting on how to remove the microcode.

The issue you are having is with Windows 10, I got it to work on window 7. But am having trouble trying to get it to stick on 10.

 

[The Stilt]

I'm back!

Anyway, what I have managed to do so far (but all with no actual results):

1. Used UBU/MMtool to remove the uCode from bios (Using MM to remove code and save unsigned bios, then use UBU to save it so it can be flashed)

2. Have windows loading different uCodes with the VMware updater (Tried 0x27, 0x36, 0x38, and 0x39)

3. Got into EFI Shell and figured out a thing or 2 (the "connect" command seems the useful one for loading a driver)

4. Got into RU.efi, have had a look around, but I have no clue where I can find those registers @The Stilt is talking about, I did find the MSR Registers, but no idea what aex or edx mean (or how and where to find and change them, or what bit 20:20 means, for that matter).

Having all bios settings at default seems to just have it run at stock turbo bins, having it set to "All Core 30x" seems to do nothing, and setting each multi to 30x per core also doesn't change the CPU's behaviour.
Having windows load different Microcode doesn't help either

If anyone would be so kind to give me some quick pointers of where/what to look for in Ru.efi - that would be highly appreciated.
If it doesn't want stick at 30x multi after that.. I don't know what else to try, maybe that driver @Dufus is brewing on his stove

Thanks again guys, i've learned a ton from this thread already, and I enjoy this journey into the unknown a lot (thank god I have dual bios to flash back a broken one LOL )

Which board are you using?

 

[C-Power/Tw0tch]

I'm on a ASRock X99 Extreme6/3.1 - More or less the same board as Dufus is using (the Taichi).
If it's down to the board, I'll be getting a Taichi, fits better with my build as well, and it's a good excuse to switch boards

Edit:
Dusting of my Windows 7, going to make a dual boot, just to see if it would work on that.
Then maybe Win10 is loading something besides the mc genuine intel file?

Edit2:
Ok Windows 7 doesn't change anything (but god do I miss it lol).
No microcode, or 0x27 both no go, maximum multiplier still is at 25x and also it will just downlock the multi to 12x (even when set to all core or per core 30x in bios), so it's not the OS being the issue..

I have an ASRock Extreme4 still laying around, maybe that board would be different, but I doubt it
Might give it a go though.... although I currently have it listed on eBay hmm.

Edit3:
Meh, going to see if that board is going to make a difference, there is also a ton of Biosses for it so a lot to play around with I guess. Unlisted from eBay and time for some more playing around!

Edit4:
DOH!
Doesn't work either, however I have seen difference in CPU behaviour on Win7 vs Win10.
7 stays at 20x multi, and goes properly to 30x on 2 cores, and never lower then 20, but Win10 goes down to 12x multi (all settings equal in bios).
Win7 also has 100points higher single core CPU bench then Win10 for some reason (all things being equal again) 1420 vs 1320ish, boggles my mind.

Anyway, not sure what else I can do apart from trying another motherboard (IE the Taichi), since Dufus has got it to work on that ^_^

Final Edit then:
Interesting find; Both Xtreme4 and Extreme6 latest biosses (3.60 and 3.30 respectively) remove the FIVR menu completely, and also remove all multiplier related settings:

 

[GTZ]

I'm on a ASRock X99 Extreme6/3.1 - More or less the same board as Dufus is using (the Taichi).
If it's down to the board, I'll be getting a Taichi, fits better with my build as well, and it's a good excuse to switch boards

Edit:
Dusting of my Windows 7, going to make a dual boot, just to see if it would work on that.
Then maybe Win10 is loading something besides the mc genuine intel file?

Edit2:
Ok Windows 7 doesn't change anything (but god do I miss it lol).
No microcode, or 0x27 both no go, maximum multiplier still is at 25x and also it will just downlock the multi to 12x (even when set to all core or per core 30x in bios), so it's not the OS being the issue..

I have an ASRock Extreme4 still laying around, maybe that board would be different, but I doubt it
Might give it a go though.... although I currently have it listed on eBay hmm.

Edit3:
Meh, going to see if that board is going to make a difference, there is also a ton of Biosses for it so a lot to play around with I guess. Unlisted from eBay and time for some more playing around!

Edit4:
DOH!
Doesn't work either, however I have seen difference in CPU behaviour on Win7 vs Win10.
7 stays at 20x multi, and goes properly to 30x on 2 cores, and never lower then 20, but Win10 goes down to 12x multi (all settings equal in bios).
Win7 also has 100points higher single core CPU bench then Win10 for some reason (all things being equal again) 1420 vs 1320ish, boggles my mind.

Anyway, not sure what else I can do apart from trying another motherboard (IE the Taichi), since Dufus has got it to work on that ^_^

Just wanted to update

I own an ASRock Extreme 4 X99

I could not get Windows 10 to play nice and for some reason beyond me I cannot get into my account. Everytime I load it, it makes a temporary account. Don't know if it was becuase of this or something else (fresh install though). No change. Will investigate later.

Windows 7 played a lot nicer but it was not stable. It hard locks often. Had to roll back to BIOS 2.1 (current is 3.6) and remove the microcodes. I could not use any 3.X BIOS becuase the option to change the multiplier and do any changes were missing. Without any microcode I'm windows installed the CPU performed worse than with microcodes. But after installing 27 with VMWare it performed well. But I think I had issues with the power limits though. It would start Cinebench with all cores loaded at 2.8 and slowly drop to 2.5. This gave me a score of almost 150 higher than normal. They cores never stuck to the max multi of a single core of 3.0 but an improvement overall. But the hardlocks exist. It can take 20mins or immediately at the splash screen.

If it takes a Tiachi to work then I will pass, I spent a lot of time working in my case and components painting them to match.

Is anybody else having the hardlocks?

 

[Dufus]

@C-Power/Tw0tch those 3.6 and 3.3 look the same as my Taichi 1.50 BIOS

I have linked the file 'V3.EFI' (V3.zip 633bytes) which can be copied to a FAT32 USB flash drive. Set CPU BIOS values back to default for now with 100MHz BCLK.
https://www.sendspace.com/file/ck1mlr

For this example I'm using a FAT32 formatted Sandisk USB flash drive with root folder 'EFI', sub folder 'Boot' containing shell.efi renamed as BOOTx64.EFI.

Spoiler: BIOS boot selection

Using the BIOS boot manager (F11 during BIOS boot on Asrock) we select the flash drive with 'UEFI:' prefix.

Spoiler: Running the Shell

Here we press 'ESC' after shell is loaded to stop the 'startup.nsh' from running. 'startup.nsh' is similar to the old DOS '.BAT' file, just a way of automating things on shell startup.

Our USB flash has been mounted as 'FS0:' as shown in the mapping table. Being the only USB device makes it an easy giveaway. So now we can test 'V3.EFI' which was copied earlier to the USB flash drive root by typing 'load fs0:\V3.EFI'. Hint, the TAB key can be used for auto-completion so sometimes we can save on a lot of typing. One should see 'V3 - All Turbo Set' if successful, something else if not. In this instance I've finished of with the 'exit' command which takes us back to the BIOS boot manager so we can select the 'Windows OS'. Alternatively we could have directly run Windows from the shell. For this system the Windows system partition is FS1: so being GPT typing "FS1:\EFI\Microsoft\Boot\BOOTMGFW.EFI" would do the trick.

Spoiler: Add to BIOS drivers

If we are happy with the driver and want to keep it we can get it to automatically load by placing it on the EFI system drive.In this instance I have copied V3.EFI from the USB flash drive 'FS0:' to the EFI system boot folder on 'FS1:' 'cp fs0:\V3.EFI fs1:\EFI\Boot'. Now using the shell boot configuration command we can add it to be executed before any OS with shell command 'bcfg driver add 0 fs1:\EFI\V3.EFI "V3 Full Turbo"'. After that type 'reset' to restart the PC as the EFI driver has not executed as yet.

Spoiler: Remove BIOS driver

To remove V3.EFI from the BIOS driver list, list the drivers to get its sequence number then remove with the bcfg remove command 'rm'. In this case 'bcfg driver rm 0'. The other driver CrScreenshotDxe.efi shown is an awesome utility from a guy who sometimes goes by the name of CodeRush. It was used to take these shell screen shots and can also be used for BIOS setup screen shots.

Don't forget to have Windows or the vmware utility update the microcode, version 0x38 works well for me.

@GTZ maybe some OC setting doesn't play well or seeing a bug with older microcode?

 

[GTZ]

@Dufus (thank you for everything)

When I use VMWare to install the micro codes and test them, l just install the new one. Should I delete the older one (and how if I need to) or is it safe just installing over the old one? Is there another method for installing microcodes to the OS or possibly before the OS loads? What method do you use?

I might try your shell command you wrote for C-Power since we have similar boards. If successful I will be a happy camper.

 

[C-Power/Tw0tch]

@Dufus

Edit: And @The Stilt too obviously - Could't have done it without you ^^

YOU ABSOLUTE LEGEND! 1000x Thank you sir!
Holy doley, over the moon, that driver works like a charm!





I still have a tiny problem,
I can't get it to load automatically, every drive I try in the shell, states "fs1:\EFI\Boot doesn't exist" Tried FS1 to 6, but no go.
Not entirely sure why, possibly because I have dual boot witn Win7 now?

 

[someEEguy]

Seems to be working on windows 10 with a Asrock X99M Extreme4 and a Xeon E5-2683v3. I did the following…..

1. downloaded latest bios


2. opened downloaded bios with UBU:

-press “7”

-press “1”

-select anything except “0/skip” for Broadwell-E, as this forces the update (I chose the latest microcode)

-select “0/skip” for Haswell-E (this causes there to be no microcode for V3s)

-press enter

-press “0/exit”


3. flashed modded bios


4. followed (OS needs to be a UEFI mode install for this to work):

Some BIOS already incorporate the UEFI shell as part of the BIOS but if we don't have that then we can download one and put it on a FAT32 USB flash drive or other suitable boot medium and boot that.

Here's a link to a shell (shell.efi). Should be okay for what we want.
https://github.com/tianocore/edk2/tree/master/ShellBinPkg/UefiShell/X64

1. Rename shell.efi to Shellx64.EFI and place on root directory of a FAT32 flash drive.

or

2. Rename shell.efi to BOOTx64.EFI then create a root folder named EFI and sub folder of that named BOOT and copy the BOOTx64.EFI to there.

and

@C-Power/Tw0tch those 3.6 and 3.3 look the same as my Taichi 1.50 BIOS

I have linked the file 'V3.EFI' (V3.zip 633bytes) which can be copied to a FAT32 USB flash drive. Set CPU BIOS values back to default for now with 100MHz BCLK.
https://www.sendspace.com/file/ck1mlr

For this example I'm using a FAT32 formatted Sandisk USB flash drive with root folder 'EFI', sub folder 'Boot' containing shell.efi renamed as BOOTx64.EFI.

Spoiler: BIOS boot selection

Using the BIOS boot manager (F11 during BIOS boot on Asrock) we select the flash drive with 'UEFI:' prefix.

Spoiler: Add to BIOS drivers

If we are happy with the driver and want to keep it we can get it to automatically load by placing it on the EFI system drive.In this instance I have copied V3.EFI from the USB flash drive 'FS0:' to the EFI system boot folder on 'FS1:' 'cp fs0:\V3.EFI fs1:\EFI\Boot'. Now using the shell boot configuration command we can add it to be executed before any OS with shell command 'bcfg driver add 0 fs1:\EFI\BOOT\V3.EFI "V3 Full Turbo"'. After that type 'reset' to restart the PC as the EFI driver has not executed as yet.

5. installed vmware driver/w 0x39 microcode:

Don't forget to have Windows or the vmware utility update the microcode, version 0x38 works well for me.

The microcode in Windows can be updated with a driver released by VMWare: https://labs.vmware.com/flings/vmware-cpu-microcode-update-driver
Here are version 0x27 & 0x39 microcodes for Haswell-Ex (0x306F2) in VMWare driver / Linux compatible format: https://1drv.ms/u/s!Ag6oE4SOsCmDhFnET3uw9wHeV4EA
Rename the desired version to microcode.dat, and proceed as instructed by VMWare.

microcode_amd.bin
microcode_amd_fam15h.bin


Before I was getting around 17000, now:

Spoiler: Full Turbo

Thanks Dufus and The Stilt.

 

[C-Power/Tw0tch]

@someEEguy

Great little compilation of steps to go through for the ones not wanting to read/go through everything posted up till now ^^

Edit:
Beware that using UBU only can cause your bios to not boot if your on one of those APTIO 5 biosses (UBU warns for that)

I had this a couple of times and had to use MMtool to remove the mirocodes, then just open the bios with UBU and just immediately press 0/Exit and save that way.
Worked every single time for me on both Extreme4 and Extreme6 Biosses

 

[someEEguy]

@someEEguy

Great little compilation of steps to go through for the ones not wanting to read/go through everything posted up till now ^^

Edit:
Beware that using UBU only can cause your bios to not boot if your on one of those APTIO 5 biosses (UBU warns for that)

I had this a couple of times and had to use MMtool to remove the mirocodes, then just open the bios with UBU and just immediately press 0/Exit and save that way.
Worked every single time for me on both Extreme4 and Extreme6 Biosses

Yeah, UBU's bios support is a bit spotty but read the following before going through with it.

B. Flashing a modded AMI Aptio V UEFI BIOS
(valid for the newest ASRock mainboards with Intel Chipset systems from X99 up)

The BIOSes of the latest ASRock mainboards do not have a capsule header anymore. That means, that the "Security Flash Check" is located somewhere else.
Our UBU and BIOS Guru SoniX has written a short guide about how to circumvent this special ASRock BIOS integrity protection.
>Here< is the link to his guide.
Update 1:
Meanwhile our Forum member lordkag found out, that the capsule header of the latest ASRock mainboard BIOSes is still present, but hidden. >Here< is his guide about how to circumvent the "Secure Flash check fail!" message while trying to flash a modded AMI Aptio V ASRock BIOS. Big thanks to him and Pacman for their work!
Update 2:
All UBU versions from v1.50 up are able to remove the hidden capsule header from ASRock AMI AptioV BIOSes automaticly and to make the modded BIOS flashable without getting the "Secure Flash check fail!" message.

http://www.win-raid.com/t455f16-Guide-How-to-flash-a-modded-ASUS-ASRock-Gigabyte-AMI-UEFI-BIOS.html

and
http://www.win-raid.com/t18f16-Guide-Manual-AMI-UEFI-BIOS-Modding-9.html#msg33546

 

[someEEguy]

I still have a tiny problem,
I can't get it to load automatically, every drive I try in the shell, states "fs1:\EFI\Boot doesn't exist" Tried FS1 to 6, but no go.
Not entirely sure why, possibly because I have dual boot witn Win7 now?

Were both your windows installs in uefi mode?

 

[C-Power/Tw0tch]

No - lol.

Which is why I'm backing up my drives at the moment and doing a clean install. Because it's worth it ^_^

Edit:
Yeap, all working now! This just makes me want to get a 2686 now and have 18 cores/36 thread at 3,5ghz+

 

[Ionstream]

No - lol.

Which is why I'm backing up my drives at the moment and doing a clean install. Because it's worth it ^_^

Edit:
Yeap, all working now! This just makes me want to get a 2686 now and have 18 cores/36 thread at 3,5ghz+

I'll let you know how that turns out

 

[Dufus]

@GTZ At this time there are not any known old microcode updates with the full turbo all cores bug still intact. The earliest one I've tried is version 0x19 and that's a no go so would need to be earlier than that. For now it's a case of appyling the turbo settings without an update present then applying the microcode update. Usually it's best to apply the microcode update early in the BIOS initialization but we don't have that luxury unless willing to rewrite a lot of the BIOS which is a lot of work. Microcode updates can be written over no problem, some versions even allow rolling back.

My first method to apply microcode update was just the OS, both Windows and Linux provide means to do this natively. Windows may require a KB update and Linux the microcode.dat file which Intel updates from time to time for this reason. Second method was to incorporate the microcode update into the EFI file which works similarly to the V3.EFI file and to date have started incorporating as BIOS firmware. The advantage of BIOS firmware is that settings remain after Sleep mode. With the EFI file approach settings are changed after BIOS post so do not stick with Sleep mode. The file method while being limited has the advantage of being simpler to apply and more universal.


@C-Power/Tw0tch If you are using MBR partitioning then you will not have a true EFI\BOOT directory. You could perhaps try and make one. I don't know if using an NTFS file system would be ideal but perhaps one could make a small additional FAT32 partition on one of the drives. BTW your CPUZ bench seems a lot lower than it should be?


@someEEguy Thanks for the confirmation. Technically it should work for all OS's and boards, I've personally run it with W7 and Linux Fedora 24. Some of those guys on win-raid may be able to help with a proper BIOS mod.


@Ionstream That would be interesting. I only have the Asrock X99 and a single 2683v3 to test with so seeing other results would be great.

The V3.EFI driver basically sets all turbo's to max core ratio, cache min/max ratio's to max cache ratio and core and cache voltage with -20mV offset then locks the OC'ing register so the settings other than cache min/max don't get affected by other software. Those settings seem to work well and help a little with the AVX2 offset which appears dependent on core and cache voltage, cache ratio and number of active cores rather than power dependent. Would be nice to get rid of that and reduce the AVX voltage if possible.

 

[katlen]

No - lol.

Which is why I'm backing up my drives at the moment and doing a clean install. Because it's worth it ^_^

Edit:
Yeap, all working now! This just makes me want to get a 2686 now and have 18 cores/36 thread at 3,5ghz+

I got an 2686v3 , let me know how it goes if you happen to get one , i doubt you could get 3.5ghz all core on 120W power but maybe 3ghz or a little higher would work , i currently have a supermicro motherboard and i think i can't do much about custom modules for efi bios, considering asrock taichi or extreme4 if someone confirms it works on it.

All the best

 

[Gaha]

hi folks - @Dufus, @C-Power/Tw0tch, @The Stilt ,
this is awesome thread !
I do have a Xeon 2660 v4 (Engineering Sample) ... am running linux (though have windows installed just in case) ... did read via readmsr on linux registers for turbo-bins (how many cores can turbo to which frequency) ... and it's exactly what i observer. It's 14 core processor ... 1-2 cores - 3.1GHz, 3 2.9 GHz,4 2.8GHz,5 2.7GHz, 6 2.6GHz and more then 6 2.5 GHz (its little bit different then regular 2660 V4 i guess cause it is ES). From thermal side I am sure CPU could cores at higher frequency (more then 2.5) ... so am hopeful you have some ideas.

Anyhow, motherboard is Supermicro X10SRM-F and do you think there is a way that microcode could be updated either in linux when it boots up or even if possible with BIOS (change ?) ?Or in Windows ?
I am ready to investigate and invest time if needed ...

Thanks for your opinion and advice !

Gaha

 

[C-Power/Tw0tch]

@Dufus
I just backed everything up and did a UEFI Windows 10 install, that worked fine and everything loads as it should now Also scores are fine now Was still in testing phase and cpu was probably in use by background stuff like windows update

http://valid.x86.fr/2eukgx

@katlen
I can confirm this works on the Xtreme4 (together with the 2683 at least) I think it will work for the 2686 just as well - but I would need to get a hold of one first to be sure

@Gaha
I think the V4 works slightly different from the Haswell's - I'm not sure if it works on those! I am sure some of the other guys on here can chime in on it

 

[Ionstream]

Still trying to find my feet.
I'm on a C612 chipset, which happens to be running on AMI Aptio V, so no easy UBU edit for me. Also, because my BIOS lacks MCE, I have to resort to MSR programming, which I have absolutely no knowledge about. Could any kind soul explain what's going on in this picture?

http://imgur.com/a/il16l