What controls Turbo Core in Xeons?

[cekim]

I get the exact same results with the same board and cpu.
....
On the other hand I'm interested in if the asus z10pe d8 ws would work with one and also with two 2696s?
Maybe I'll change the taichi board to that so when I want to upgrade to 2 cpus the board will be ready.
What are the drawbacks using double-socket boards with only one cpu?
Would I achieve 3.3 or 3.4 ghz with one and with two cpus too?

1. Interesting - so there is something about the Taichi SMM causing this. Someone (ucode?) mentioned that there were some querks with the Taichi boards using multiple system registers to limit clocks, but I didn't get any details.
2. "drawbacks" of the dual boards are:
a. need ECC memory
b. cannot changing timing of memory (as you can with normal DDR4 on a single board) - I run 2133@CAS 10 and see corresponding improvement in application level performance.
c. larger board limits and complicates case choices.

It can be run with 1 processor though. No issue there. I'll have an EP2C612-WS shortly to try out. I hope that performs as well as the Z10PE-D16-WS.

 

[cekim]

Looking at getting a programmer, but curious if anyone has managed to flash a modified bios in an EP2C612-WS? I presume the bios tool rejects the hand-edited and ubu edited bios?

 

[XDrego]

Looking at getting a programmer, but curious if anyone has managed to flash a modified bios in an EP2C612-WS? I presume the bios tool rejects the hand-edited and ubu edited bios?

read back a page Welsper did but his links dead. he used ch341a I think.in contact with asrock about that bios adds cluster on die fixed my memory issue by populating banks from outside in not by manual instructions turbos work out of box at least very stable

 

[cekim]

read back a page Welsper did but his links dead. he used ch341a I think.in contact with asrock about that bios adds cluster on die fixed my memory issue by populating banks from outside in not by manual instructions turbos work out of box at least very stable

ch341 is what I meant by a "programmer". That's how you program the BIOS ROM outside the system. You plug it into a chip that is able (with the help of USB) to drive the programming sequence.

I have one on the way, but until then, I was wondering if it was even possible (as it is with ASUS flashback) to write the image in without doing this.

 

[Gellért]

If I run cpu-z bench I get lower points when no other programs are used. When I run the benchmark when some other program generates some cpu load the points goes up higher. (~7300 vs ~8300)
Why is that?

I changed the ram order on the x99 taichi board like @XDrego said. (2-2 from outside, not A1 B1 C1 D1 slots) Seems ok.

 

[v8pixel]

Hello guys! I've read all the 36 pages, last night i tried to mod my bios on my Gigabyte X99 Ultra Gaming Rev 1.0 and was unsuccessful, unable to remove the microcodes.
Flashed the bios and when i load the V3.efi it says the i allready have the microcode...
I see that this is common for MSI and Asus...
Filko84 you have a similar MotherBoard, can you help me?
I have a Xeon 2683 V3 OEM runing now at max x25 (2500mhz) and should turbo to 3GHZ.
Thank you guys again, i really appreciate what are you doing here!

 

[Filko84]

Hello guys! I've read all the 36 pages, last night i tried to mod my bios on my Gigabyte X99 Ultra Gaming Rev 1.0 and was unsuccessful, unable to remove the microcodes.
Flashed the bios and when i load the V3.efi it says the i allready have the microcode...
I see that this is common for MSI and Asus...
Filko84 you have a similar MotherBoard, can you help me?
I have a Xeon 2683 V3 OEM runing now at max x25 (2500mhz) and should turbo to 3GHZ.
Thank you guys again, i really appreciate what are you doing here!

You can remove the driver with "bcfg driver rm 0"

Try this bios:
http://dox.bg/files/dw?a=26cd40db72

Install the bios, boot in Shell, remove driver, copy V3_1.efi to EFI\Boot partition, auto load the new driver from the EFI\Boot partition, restart and boot in windows, delete windows default microcode, install 39 microcode with vm tool.
That is the order but first of all you need to install Windows in UEFI mode or it will not work!

 

[v8pixel]

Wow! You rock , did you remove the microcodes? I think this in the worst x99 motherboard. I don't even know how to install windows in uefi, it has very little options in the bios...
I will flash the bios tonight and have find V3_1.efi (I have only the V3.efi). If it tells me turbo all cores it will be great when I load the V3_1.efi ! And I will try to reinstall windows or maybe install windows 10 to be sure.

 

[Filko84]

You need to choose in bios to load the OS in UEFI not in Legacy mode and then install Windows.

 

[Filko84]

Wow! You rock , did you remove the microcodes? I think this in the worst x99 motherboard. I don't even know how to install windows in uefi, it has very little options in the bios...
I will flash the bios tonight and have find V3_1.efi (I have only the V3.efi). If it tells me turbo all cores it will be great when I load the V3_1.efi ! And I will try to reinstall windows or maybe install windows 10 to be sure.

You can use V3.efi also but for me I get wrong CPU with it and other V3xx.efi files.
Only V3_1.efi is working correctly but my CPU is ES version so you should be fine with V3, V3x2 or any other.
Also you can search for V3x2_cp39.efi with already loaded microcode in the file, so no need to load it in windows.

 

[v8pixel]

I've set that when a installed windows. Maybe is in uefi boot the windows... is there a way to find put if is in uefi or lagacy when I am in windows?
Anyway I will search on Google, i'm going off topic now.
I will let you all know after I flash the bios from you. Thank you!

 

[v8pixel]

Filko84 i had no luck with the bios from you, it says ivalid bios image...

http://imgur.com/a/GxPQe

What can i do next?

Here are more screens from the bios
http://imgur.com/a/rRnxH
http://imgur.com/a/AgtIH
http://imgur.com/a/18yrV

 

[Gellért]

1. Interesting - so there is something about the Taichi SMM causing this. Someone (ucode?) mentioned that there were some querks with the Taichi boards using multiple system registers to limit clocks, but I didn't get any details.
2. "drawbacks" of the dual boards are:
a. need ECC memory
b. cannot changing timing of memory (as you can with normal DDR4 on a single board) - I run 2133@CAS 10 and see corresponding improvement in application level performance.
c. larger board limits and complicates case choices.

It can be run with 1 processor though. No issue there. I'll have an EP2C612-WS shortly to try out. I hope that performs as well as the Z10PE-D16-WS.

This is what I achieved with the taichi - fully stabile - on 18 cores so far. Only blck changed (104.2) and with the v3x2_50_vcc.efi activated


This one with 10 cores. I don't know if it's stabile or not. (edit: not) Seems like there is a limit to increasing blck right? (while you don't freeze out at start) (105.6) Any ideas? Could you test it with an asus d8 or d16 ws what you can achieve for comparison purposes?


This one is interesting. The '7.' one. 36 cores, 2 cpu, not 8 gig ram with high latency like the first one's but 131 gb (?) and 3898.55Mhz.
http://valid.x86.fr/top-cpu/496e746...0552045352d32363936207633204020322e333047487a

@custom90gt imagehosts are changed

 

[custom90gt]

Yeah don't click on those pictures - reported above post.

@Gellért - you need to find a different image host.

 

[cekim]

Yeah don't click on those pictures - reported above post.

@Gellért - you need to find a different image host.

lol, I was going to post the same thing. I've seen some sketchy image hosting before, but.... wow... ;-)

FWIW, H80i v2 seem to be doing well for 2696v3 3.3/3.4 all-core so far. Much better than the Noctua 3U sinks. They were hovering right at 70C in stress-app and worse in heavier things.

So far, it seems to have settled at 48-49C for stress-app. Going to try some heavier loads now.

Awesome on a stick - 18 core heavy load in a 36 core system:
cpu MHz : 3799.941
cpu MHz : 3770.292
cpu MHz : 3710.726
cpu MHz : 3799.671
cpu MHz : 3775.773
cpu MHz : 3796.796
cpu MHz : 3800.390
cpu MHz : 3689.164
cpu MHz : 3799.851
cpu MHz : 3793.652
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3793.652
cpu MHz : 3793.652
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3708.570
cpu MHz : 3673.890
cpu MHz : 3673.082
cpu MHz : 3778.917
cpu MHz : 3665.804
cpu MHz : 3619.894
cpu MHz : 3657.539
....

coretemp-isa-0000
Adapter: ISA adapter
Physical id 0: +56.0°C (high = +68.0°C, crit = +78.0°C)

coretemp-isa-0001
Adapter: ISA adapter
Physical id 1: +55.0°C (high = +68.0°C, crit = +78.0°C)

 

[cekim]

Filko84 i had no luck with the bios from you, it says ivalid bios image...

http://imgur.com/a/GxPQe

What can i do next?

Here are more screens from the bios
http://imgur.com/a/rRnxH
http://imgur.com/a/AgtIH
http://imgur.com/a/18yrV

I don't have any experience with the GB boards, but generally speaking the "normal" method of bios update doesn't work with any of them. You usually need a more permissive method that doesn't do all the integrity checks. I'd go through this thread again and see what tool people are using with those boards.

I use flashback for ASUS as it will accept bios images that ezflash rejects. The Taichi allowed a UBU modified bios to flash. Looks for the EP2C612-WS, I'm going to have to use an external writer to burn the ROM outside the system.

 

[Cata40]

Can you explain how to remove v3.efi via shell and try the v3_1.efi or V3_2.efi?
When I go to shell and load the v3_1.efi I get message that it is already presented.
I also copied it to the fs1:\efi\boot to auto load it but want to remove it and auto load the v3_1.efi.

This is the link for modded latest stable bios for Gigabyte GA-X99-Gaming5 if someone want to use it.
http://dox.bg/files/dw?a=334fd2c669

In windows 7 work this:

The reason why you can't access it on Windows Explorer (File Explorer) is because by default it does not run with administrative privileges. If you let's say kill Explorer and execute it again from an elevated command prompt then you will have access to the EFI partition through File Explorer.

First open an elevated command prompt and give the EFI partition a drive letter, you can do it with mountvol as shown in jmiserez's answer (recommended) or with diskpart. In the following example the EFI partition is on disk 0, partition 1 and will be assigned to letter b:

diskpart
list disk
select disk 0
list partition
select partition 1
assign letter=b
exit

Now kill the File Explorer process and start it with administrative privileges:

taskkill /im explorer.exe /f
explorer.exe

Now you can access the EFI partition with File Explorer.

Note that the EFI partition is a FAT32 file system partition. FAT partitions don't have the Security tab, this is a NTFS feature only.

I don"t know if it works in windows 10 like this

 

[pianosso]

Hello. My ASROCK X99 ITX and XEON E5 2683 V3 with bios v. 3.60 still shows the microcode value.
If I mod by using microcode value 27 or 39, it shows those values in CPU-Z. But if I skip the microcode in UBU, it shows microcode 38 (default value)
I run an UEFI Windows 7, I modded the UEFI boot by loading V3 and the full turbo config but it seems there's no change. Max speed is 2.5 GHz.
Would anyone please let me know what do I do wrong?
Thanks a lot.

 

[Gellért]

lol, I was going to post the same thing. I've seen some sketchy image hosting before, but.... wow... ;-)

FWIW, H80i v2 seem to be doing well for 2696v3 3.3/3.4 all-core so far. Much better than the Noctua 3U sinks. They were hovering right at 70C in stress-app and worse in heavier things.

So far, it seems to have settled at 48-49C for stress-app. Going to try some heavier loads now.

Awesome on a stick - 18 core heavy load in a 36 core system:
cpu MHz : 3799.941
cpu MHz : 3770.292
cpu MHz : 3710.726
cpu MHz : 3799.671
cpu MHz : 3775.773
cpu MHz : 3796.796
cpu MHz : 3800.390
cpu MHz : 3689.164
cpu MHz : 3799.851
cpu MHz : 3793.652
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3793.652
cpu MHz : 3793.652
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3799.941
cpu MHz : 3708.570
cpu MHz : 3673.890
cpu MHz : 3673.082
cpu MHz : 3778.917
cpu MHz : 3665.804
cpu MHz : 3619.894
cpu MHz : 3657.539
....

coretemp-isa-0000
Adapter: ISA adapter
Physical id 0: +56.0°C (high = +68.0°C, crit = +78.0°C)

coretemp-isa-0001
Adapter: ISA adapter
Physical id 1: +55.0°C (high = +68.0°C, crit = +78.0°C)

Why are those weird Mhz numbers? Are they fluctuating? Could you check it with cpu-z what mhz and how much points you get? You could try enabling only 10 cores/cpu in uefi bios. I get 3,9ghz stabile on 10 cores (out of 18). I wonder what that would be like on a dual cpu system.

 

[Cata40]

Hello. My ASROCK X99 ITX and XEON E5 2683 V3 with bios v. 3.60 still shows the microcode value.
If I mod by using microcode value 27 or 39, it shows those values in CPU-Z. But if I skip the microcode in UBU, it shows microcode 38 (default value)
I run an UEFI Windows 7, I modded the UEFI boot by loading V3 and the full turbo config but it seems there's no change. Max speed is 2.5 GHz.
Would anyone please let me know what do I do wrong?
Thanks a lot.

Maybe you forget to rename in system 32, mcupdate_GenuineIntel

 

[cekim]

Why are those weird Mhz numbers? Are they fluctuating? Could you check it with cpu-z what mhz and how much points you get? You could try enabling only 10 cores/cpu in uefi bios. I get 3,9ghz stabile on 10 cores (out of 18). I wonder what that would be like on a dual cpu system.

Yes, they fluctuate.
Running linux, so can't easily get cpuz numbers.
This is a dual system.

 

[pianosso]

Maybe you forget to rename in system 32, mcupdate_GenuineIntel

Actually I manually removed the microcode data from the BIOS file then put it in UBU, saved it without any other tweaks, flashed the BIOS, set the cores to All Turbo in EFI shell and DONE.
Cores all running at 3.0 GHz on 2683 V3.

 

[Gellért]

Yes, they fluctuate.
Running linux, so can't easily get cpuz numbers.
This is a dual system.

Can you disable some cores in the bios on the asus board like on the asrock one?
Then you should get higher frequencies even at stock. (without the 'hack')
I mean I get ~3,4ghz with 18 cores but with only 10 cores enabled I get ~3,9 Ghz on them.
That is what I'm interested in if you activate only 10 cores/cpu you would get the same result with 2 cpus or not.

 

[n31l]

Strange.. performance scores are roughly half when turbo is enabled on all cores.. all cores run 3199, no temperature issues

I have an Asrock x99m Fatal1ty (3.30 Bios) 2695 v3 -- tried 0x27 & 0x39.

I'm currently manually disabling the "mcupdate_GenuineIntel.dll" and manually loading V3.efi whilst I'm testing.. anybody had any similar issues?

 

[Gellért]

Strange.. performance scores are roughly half when turbo is enabled on all cores.. all cores run 3299, no temperature issues

I have an Asrock x99m Fatal1ty (3.30 Bios) 2695 v3 -- tried 0x27 & 0x39.

I'm currently manually disabling the "mcupdate_GenuineIntel.dll" and manually loading V3.efi whilst I'm testing.. anybody had any similar issues?

http://ibb.co/kHsgoF
http://ibb.co/d4zSTF

You should definately try 'v3x2_50_39_vcc.efi' with microcode 39 built in or 'v3x2_50_vcc.efi' without a microcode.

https://drive.google.com/open?id=0B4nDSxzsKIeOWk1NenhDSEFRbnM - I uploaded them but have no credit

You should delete microcode and 'mcupdate_GenuineIntel.dll' from system32 first.
If you delete the microcode but you have the 'mcupdate_GenuineIntel.dll' present in the system32 folder, it will automatically load ucode 36 as we see in one of your pictures. (hwinfo - at the top - uCU - 36) If you have no microcode loaded it will show 'uCU - '

how to delete an 'undeletable' file (mcupdate_GenuineIntel.dll): https://www.howtogeek.com/howto/windows-vista/how-to-delete-a-system-file-in-windows-vista/