cancel
Showing results for 
Search instead for 
Did you mean: 

[MAIN\MERGED] Intel Skylake & Kaby Lake Hyperthreading Bug

pepper_chico
Level 7
It has been confirmed that Intel's Skylake/Kaby Lake hyper-threading can present serious issues that can be fixed by a BIOS/UEFI update that targets Intel processor errata KBL095, KBW095 or the similar one for Kaby Lake processor.

Any plans for such update to come? On Kaby Lake it seems the most safe route at the moment is to simply disable hyper-threading until ASUS provides a new BIOS that addresses this. On Skylake there's the option on ArchLinux side, for example, to install extra/intel-ucode to get microcode updates since there's a fix in revision 0xb9/0xba and later. Windows doesn't have a microcode update for this yet, afaik.

Source: https://lists.debian.org/debian-devel/2017/06/msg00308.html
86,654 Views
133 REPLIES 133

Leoplate25 wrote:
Hi, i did what Da-BOSS posted and RWEverything and HWINFO64 show "BA" microcode, so i updated it, i think, because my BIOS keeps showing "9E". I have an i7-6700k (Intel64 Family 6 Model 94 Stepping 3) on an ASUS Z170 Sabertooth S. Can someone tell me if it's ok? Thanks!!!


65710

i used this
http://www.win-raid.com/t154f16-Tool-Guide-News-quot-UEFI-BIOS-Updater-quot-UBU.html

AKBAAR wrote:


i used this
http://www.win-raid.com/t154f16-Tool-Guide-News-quot-UEFI-BIOS-Updater-quot-UBU.html


And looks like he has BIOS Flashback. Probably doesn't know how to use it.

eldata wrote:
And looks like he has BIOS Flashback. Probably doesn't know how to use it.


I converted the file with ubu, copied and pasted into my usb stick, shutdown the PC, pressed the BIOS flashback button for 3 seconds, like ASUS says, then stops blinking within 5 seconds and the BIOS won't update. 😞
If we can't live together, we are gonna die alone.

I did it!!! I put the usb stick in the usb 3.0 port and it flashed the BIOS. Sooo scared because it takes too long and 3 restarts. Thanks to everybody who gave me a hand with this!!!

65721
If we can't live together, we are gonna die alone.

Leoplate25 wrote:
Hi, i did what Da-BOSS posted and RWEverything and HWINFO64 show "BA" microcode, so i updated it, i think, because my BIOS keeps showing "9E". I have an i7-6700k (Intel64 Family 6 Model 94 Stepping 3) on an ASUS Z170 Sabertooth S. Can someone tell me if it's ok? Thanks!!!


Hi! I have an i7-6700k on an ASUS z170 Sabertooth S and i tried everything, modded BIOS with UBU (told me it is not a valid BIOS file), installing UBUNTU on a virtual machine and i could not do anything either, and all the "solutions/fixes" given in this post. Can you help me? Can you attach for me a modified BIOS that actually work or what can you do? Thanks in advance! Cheers from Argentina!"

Try UBU again but use BIOS Flashback to flash the modified CAP file instead of EZ Flash.

Korth wrote:

~
This bug has affected both Skylake and Kabylake. Seems like a good chance it's also going to be in Coffeelake.
~


Hardly likely given the **fact** that they have already sent SKYL and KBL uCode fixes to their partners in April/May. In our case, it's Asus who has been dragging their feet instead of releasing the uCode in BIOS updates. The ball is now squarely in Asus' side of the court.

eldata wrote:
Hardly likely given the **fact** that they have already sent SKYL and KBL uCode fixes to their partners in April/May. In our case, it's Asus who has been dragging their feet instead of releasing the uCode in BIOS updates. The ball is now squarely in Asus' side of the court.

@eldata,

No, not quite yet, lol.
For SKL desktop SKUs yes, since Apr2017 (as we already know from the OP links in this thread).
For KBL desktop SKUs no, not yet. Intel is still dragging their feet. ASUS will get their chance to drag their feet once Intel passes the ball to them.
And CFL is, of course, an 8th gen refinement of 7th gen KBL which is itself a refinement of 6th gen SKL, they run off very much the same microarchitecture and microcode steppings (at least we know SKL and KBL do, CFL is not yet available for examination).

The latest info *from Intel* that I could find:
https://communities.intel.com/thread/115745
https://communities.intel.com/message/483358

Legolas (28Jun2017): "Does this issue effect Desktop Skylake processors and Kabylake processors? I hope Intel is looking into it."
Intel Corporation (28Jun2017): "Hi Legolas, This also affects the desktop processors; however, Intel is currently working on a solution, once available, it will be published. Best Regards, Juan V."
Intel Corporation (29Jun2017): "Hi Legolas, The likelihood of running into this issue is low. A [Skylake] fix for this issue is available and started rollout out to system manufacturers in April 2017. Intel recommends checking to make sure your BIOS is up to date. Best Regards, Juan V."

Do you have links to anything *from Intel* which state otherwise? Better yet, a link *from Intel* to a fix for my Kaby Lake CPUs?

It's worth noting that Intel discourages code complexes which create the specific conditions that cause this bug to manifest. NVIDIA is already working out other bugs in their drivers (which create high interrupt latencies when Hyperthreading) and might choose to also avert Intel's HT bug by simply compiling their binaries with different (or differently ordered) instructions, commands, and registers - so in this example, if NVIDIA chooses to implement this fix, the Intel HT bug wouldn't be invoked by running NVIDIA code (or by running game engines built onto it). I've installed patches for Autodesk, Solidworks, and SIMUL8 which also circumvent this Intel HT bug in those apps. Although the proper fix for Windows users is an Intel firmware update because no one really knows exactly how WinOS components really behave at the system hardware level, lol, otherwise there's still always a small chance for a stray unfixed app to "seriously compromise" the entire system.
"All opinions are not equal. Some are a very great deal more robust, sophisticated and well supported in logic and argument than others." - Douglas Adams

[/Korth]

Korth wrote:
@eldata,

No, not quite yet, lol.
For SKL desktop SKUs yes, since Apr2017 (as we already know from the OP links in this thread).
For KBL desktop SKUs no, not yet. Intel is still dragging their feet. ASUS will get their chance to drag their feet once Intel passes the ball to them.
And CFL is, of course, an 8th gen refinement of 7th gen KBL which is itself a refinement of 6th gen SKL, they run off very much the same microarchitecture and microcode steppings (at least we know SKL and KBL do, CFL is not yet available for examination).

The latest info *from Intel* that I could find:
https://communities.intel.com/thread/115745
https://communities.intel.com/message/483358

Legolas (28Jun2017): "Does this issue effect Desktop Skylake processors and Kabylake processors? I hope Intel is looking into it."
Intel Corporation (28Jun2017): "Hi Legolas, This also affects the desktop processors; however, Intel is currently working on a solution, once available, it will be published. Best Regards, Juan V."
Intel Corporation (29Jun2017): "Hi Legolas, The likelihood of running into this issue is low. A [Skylake] fix for this issue is available and started rollout out to system manufacturers in April 2017. Intel recommends checking to make sure your BIOS is up to date. Best Regards, Juan V."

Do you have links to anything *from Intel* which state otherwise? Better yet, a link *from Intel* to a fix for my Kaby Lake CPUs?

It's worth noting that Intel discourages code complexes which create the specific conditions that cause this bug to manifest. NVIDIA is already working out other bugs in their drivers (which create high interrupt latencies when Hyperthreading) and might choose to also avert Intel's HT bug by simply compiling their binaries with different (or differently ordered) instructions, commands, and registers - so in this example, if NVIDIA chooses to implement this fix, the Intel HT bug wouldn't be invoked by running NVIDIA code (or by running game engines built onto it). I've installed patches for Autodesk, Solidworks, and SIMUL8 which also circumvent this Intel HT bug in those apps. Although the proper fix for Windows users is an Intel firmware update because no one really knows exactly how WinOS components really behave at the system hardware level, lol, otherwise there's still always a small chance for a stray unfixed app to "seriously compromise" the entire system.


65725

eldata
Level 8
I read it when it was in its original location. Didn't make sense there either.

eldata
Level 8
UBU 1.69.4 now has the fixed uCode for KBL.

https://cloud.mail.ru/public/9SSs/YJbsWyC2V

Select Microcode for CPU Kabylake (LGA1151)

34 Version 34 Date 10-07-2016
3A Version 3A Date 22-08-2016
3C Version 3C Date 05-09-2016
3E Version 3E Date 16-09-2016
42 Version 42 Date 02-10-2016
48 Version 48 Date 15-11-2016
58 Version 58 Date 09-03-2017
5E Version 5E Date 06-04-2017 - Bug fix HT


Fixes for both SKYL and KBL are now in the public domain.