• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

What controls Turbo Core in Xeons?

Page 119 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Djordjemasinac

Junior Member
Apr 26, 2017
15
1
41
Unfortunately, it seems that post was wrong. I found no memory overclocking in that bios and the features that are there are only for unlocked xeons and all multiplier based.
 

Roshik

Junior Member
Dec 7, 2019
7
0
6
Good evening, boys. Have already asked the question on the Board to unlock turboboost processors 2696 v3. You have advised me to Asus Z10PE D16WS. What is needed power supply to be stable on turboboost? And the second question, the motherboard Gigabyte MW 70-3S0, can be an alternative to Asus Z10PE D16WS?
 

topmysteries5

Member
Jan 31, 2019
179
14
51
Good evening, boys. Have already asked the question on the Board to unlock turboboost processors 2696 v3. You have advised me to Asus Z10PE D16WS. What is needed power supply to be stable on turboboost? And the second question, the motherboard Gigabyte MW 70-3S0, can be an alternative to Asus Z10PE D16WS?
For dual 2696v3, you will need minimum 750 watts gold rated PSU.

Gigabyte MW70-3S0 has weaker VRM than Asus, though you can use this turbo mod but at lower intensity (lower performance boost).
 

Micrornd

Senior member
Mar 2, 2013
919
75
91
For dual 2696v3, you will need minimum 750 watts gold rated PSU.

Gigabyte MW70-3S0 has weaker VRM than Asus, though you can use this turbo mod but at lower intensity (lower performance boost).
Have you ever actually ran a Gigabyte MW70-3S0 with dual E5-2696 v3's, or is that just your theory? :rolleyes:
 

topmysteries5

Member
Jan 31, 2019
179
14
51
Have you ever actually ran a Gigabyte MW70-3S0 with dual E5-2696 v3's, or is that just your theory? :rolleyes:
I tried this mod with MW70-3S0 with dual 2686v3 in october this year. (I hav dual 2686v3 not 2696v3) For some reason this rig was freezing with u-code 27 under cinebench load, it only worked stable with u-code 39 and 40, but max all core turbo was x26 under heavy load. -70, -50, -20 was undervolting used. Same efi and u-code 27 gives x30 in asus Z10D16WS. I remember you succeeded with this mod on some gigabyte server motherboard earlier
 
Last edited:
  • Like
Reactions: Roshik

Micrornd

Senior member
Mar 2, 2013
919
75
91
I succeeded with the MW70-3S0. It ran with any microcode , although 43 or none gave the best results.
YMMV as memory, video card, bios, bios settings, bios revision, cooling, undervolting file, etc. all play into the final result.
I took the time to sort it out, my stability tests were 8hr full load on both the board and and video card simultaneously.
That is what I use as a stability test.
My MW70-3S0 ran all cores 2696v3 @ 3.4ghz (3.3ghz AVX) with 43 microcode, 24/7 without problems.
(With no microcode, all cores @ 3.5ghz (3.4ghz AVX))

Even running all cores @ 3.5 ghz, testing with Linpak Extreme and a Radeon VII, testing with the fuzzy donut (all AIO water-cooled) at the same time,
my Kill-a-watt never recorded higher than 735w @ the wall during an 8 hr test.
The cpu's did not exceed 61c, while the Radeon VII went 76c, (85c hotspot) after 8 hrs with Corsair H110s on the cpus and a Kraken x62 on the Radeon.
(The Kraken was chosen because of the extra tubing length needed for radiator placement)

Remember the Gigabyte board has no built in overclocking bclk ability, it was designed for long term stability, unlike the Asus's "let's just throw a lot of stuff on the board, so they can play with it" approach.
You want to remember that overclocking by bclk (as with Asus) is inherently unstable, as the pci-e bus is still tied to the same clock with v3's and v4's, so anything on the pci-e bus (including the board's sata controllers, nics, video cards, HDDs, etc) will add instability when you play with the bclk.
So if you want stable performance with a v3 mod, overclocking the bclk is not it (at least not unless you consider 1-2 MHz on the bclk bus useful, I don't)
Intel didn't separate the pci-e clock from the bclk until the scalable Xeons (3647s)
 
  • Like
Reactions: Roshik

Roshik

Junior Member
Dec 7, 2019
7
0
6
I tried this mod with MW70-3S0 with dual 2686v3 in october this year. (I hav dual 2686v3 not 2696v3) For some reason this rig was freezing with u-code 27 under cinebench load, it only worked stable with u-code 39 and 40, but max all core turbo was x26 under heavy load. -70, -50, -20 was undervolting used. Same efi and u-code 27 gives x30 in asus Z10D16WS. I remember you succeeded with this mod on some gigabyte server motherboard earlier
E5 2696v3 is top end CPU for haswell-EP generation. Before doing this mod, you need to check your motherboard VRM. 2x 2696v3 has TDP of 145x2= 290w. At stock your VRM can handle this. But with OC and turbo unlock power consumption will be around 600watts total for both CPUs, which is too much for server motherboards, you may end up damaging it. As far as i know only Asus Z10PE D8-WS / D16-WS can handle this much power without damaging anything.
I have tried this mod on a Z840 last year with dual 2690v3, it didn't even worked properly. During cinebench R15 it used to freeze and reboot randomly. Caused bootloops.
Good afternoon. I looked in your configuration on the Board Asus Z10PE-D16 WS, worth two Corsair H45. They have enough unlocked for under turboboost processors ? On characteristics looked, that TDP 95 W. have Corsair H45.
 

topmysteries5

Member
Jan 31, 2019
179
14
51
Good afternoon. I looked in your configuration on the Board Asus Z10PE-D16 WS, worth two Corsair H45. They have enough unlocked for under turboboost processors ? On characteristics looked, that TDP 95 W. have Corsair H45.
I'm using two Corsair H45s on dual E5 2686v3. But I'm not using with stock fans, i using two 3400rpm nidec servo high static pressure fans in push pull, but this setup is noisy, but work fine at cheaper cost.
 

Roshik

Junior Member
Dec 7, 2019
7
0
6
I'm using two Corsair H45s on dual E5 2686v3. But I'm not using with stock fans, i using two 3400rpm nidec servo high static pressure fans in push pull, but this setup is noisy, but work fine at cheaper cost.
Can you advise me what two coolers to buy ? ASUS z10pe-D16 WS Board, e-5 2696 V3 - 2x processors with unlocked turbo boost.
 

topmysteries5

Member
Jan 31, 2019
179
14
51
Can you advise me what two coolers to buy ? ASUS z10pe-D16 WS Board, e-5 2696 V3 - 2x processors with unlocked turbo boost.
280mm size / 360mm size liquid coolers will be sufficient and quiet too. But it doesn't make sense to invest money in this rig right now, when better cpus like threadrippers are available.
 

Geez

Junior Member
Jan 1, 2020
4
0
6
Noob question,
I'm running a Xeon E3 1246V3 on Asus B85 pro gamer. I cannot run at all on max turbo Freq even a single core. Max I'm able is 3.7
Was on 2203 bios revision, but currently downgraded to 2001 still no progress.

Previsously I was able to run at 3.9GHz, but not anymore. Is it possible to run all cores @3.9Ghz ?
Any guide/link with instructions to do a bios mod (If Possible)
 

Attachments

Last edited:

Geez

Junior Member
Jan 1, 2020
4
0
6
Noob question,
I'm running a Xeon E3 1246V3 on Asus B85 pro gamer. I cannot run at all on max turbo Freq even a single core. Max I'm able is 3.7
Was on 2203 bios revision, but currently downgraded to 2001 still no progress.

Previsously I was able to run at 3.9GHz, but not anymore. Is it possible to run all cores @3.9Ghz ?
Any guide/link with instructions to do a bios mod (If Possible)
Managed to reset with the jumper and back to normal turbo.
Single core max 3.9Ghz, all cores with load stays at 3.7Ghz.

Question remains can I run all cores at 3.9Ghz with load? (Asus officially removed this functionality which was supposed to work with initial bios revision and that is no where to be found. Version 205)
 

Micrornd

Senior member
Mar 2, 2013
919
75
91
(Asus officially removed this functionality which was supposed to work with initial bios revision and that is no where to be found. Version 205)
Unless I misunderstand what you are saying, "B85-PRO GAMER BIOS 0205 First release Bios" is readily available for download on the Asus site.
 

Geez

Junior Member
Jan 1, 2020
4
0
6
Unless I misunderstand what you are saying, "B85-PRO GAMER BIOS 0205 First release Bios" is readily available for download on the Asus site.
Bios release 205 is no longer there. (205 version was only version that supported multicore enhancement for non k processors.)
My CPU (E3 1246V3) is compatible only after 404 revision which was released later (According to CPU compatibility list.)
 

Micrornd

Senior member
Mar 2, 2013
919
75
91
Bios release 205 is no longer there. (205 version was only version that supported multicore enhancement for non k processors.)
My CPU (E3 1246V3) is compatible only after 404 revision which was released later (According to CPU compatibility list.)
I don't post based on guesses.
205, 403, 2001, 2103, and 2203 are all available.
It does require you to click on "Show All"
Perhaps that's what confused you?


To make your E3-1243v3 work with an earlier bios (205), you just need to embed any additional microcode added from a bios that works with your CPU (post 404) into the 205 bios.
Google how-to, it's a good learning experience ;)
 
  • Like
  • Wow
Reactions: Zorg33 and Geez

Geez

Junior Member
Jan 1, 2020
4
0
6
I don't post based on guesses.
205, 403, 2001, 2103, and 2203 are all available.
It does require you to click on "Show All"
Perhaps that's what confused you?


To make your E3-1243v3 work with an earlier bios (205), you just need to embed any additional microcode added from a bios that works with your CPU (post 404) into the 205 bios.
Google how-to, it's a good learning experience ;)
Many thanks. Yep I was redirected to BIOS straight from CPU support list section. (Sorry for being extremely stupid)
Will try this and post my results. :grimacing:
 

er557

Junior Member
Oct 11, 2016
14
5
51
anyone got issues with latest windows versions and initial boot? my c6 states are disabled and all seems to work ok, however a certain bios setting in the past prevented install with vtd extensions, currently i do have c3 and c6 for cpu, but the offending other setting remains off(max c states or whatever).is the safe os boot environment of windows setup more vulnerable to freezes?
i do have the turbo unlock working, but a hardware issue is suspected-as i cant upgrade to windows 1909 or 20h1
 

himtan

Junior Member
Sep 7, 2019
7
1
16
anyone got issues with latest windows versions and initial boot? my c6 states are disabled and all seems to work ok, however a certain bios setting in the past prevented install with vtd extensions, currently i do have c3 and c6 for cpu, but the offending other setting remains off(max c states or whatever).is the safe os boot environment of windows setup more vulnerable to freezes?
i do have the turbo unlock working, but a hardware issue is suspected-as i cant upgrade to windows 1909 or 20h1
Which cpu do you have ? Is it QS ?
 

er557

Junior Member
Oct 11, 2016
14
5
51
well, I got dual socket xeons e5-2686 v3 QS 18c/36t on asrock rack ep2c612-ws board, they both are unlocked on all 36 cores/72 threads @3.5ghz, using randir's efi driver with microcode x39 included, -50 undervolt on vcore, vssa and cache. very stable and a long running tested system.
 

himtan

Junior Member
Sep 7, 2019
7
1
16
well, I got dual socket xeons e5-2686 v3 QS 18c/36t on asrock rack ep2c612-ws board, they both are unlocked on all 36 cores/72 threads @3.5ghz, using randir's efi driver with microcode x39 included, -50 undervolt on vcore, vssa and cache. very stable and a long running tested system.
i am using payne efi file, the latest 43 microcode and latest windows 10 and i do not have any problem.
 
  • Like
Reactions: er557

karakarga

Junior Member
Apr 20, 2019
2
0
11
I have a Huanan x99-TF mainboard with Xeon E5 v4 type Engineering sample 16 Core 1600MHz processor.


I have seen that, CPU-Z tests multi core and single core. The all core result is 4944 at max turbo speed at 2000MHz. If there is 16 cores, 4944 / 16 = 309 per core. But for single CPU result is 290 only. If with single core, it must bump to 3000MHz. So 309 x (3000 / 2000 = 1.5) = 309 x 1.5 = 463.5 but the result is only 290! ~1.6 times lower than it must have to be.

The bios can not push to 30x divider. I have also used a single CPU intensive really aged programe called vReveal v3.2.0.13029. That is an old, distortion cleaning programe from videos. It run at 2000MHz all the time with one core.

So there is a bios error, that prevents CPU to shift to single core max speed at x99-TF Huanan board.
 

ASK THE COMMUNITY

TRENDING THREADS