Hardware bug in Skylake and Kaby Lake : Unsafe Hyper-Threading

[tsk]

Unless you're working with critical data, this is nothing to worry about for the average user, which is 99,8% of this forum.

[NvidiaIntelAMDLoveTriangle]

Oh Debian, just because your OS is broken, doesn't mean the hardware is too.

[mr moose]

6 minutes ago, Cool Guy said:

TBH I don't care.  

Alright, not my business.  I didn't need some whole explanation.  Not trying to be rude, just not my business so tbh I don't really care. 

 

You made it your business when you started giving advice and making claims about it.   Please refrain from giving advice about things if you don't really care enough to know how it will effect other users.

[ivan134]

12 minutes ago, Cool Guy said:

TBH I don't care.  

Alright, not my business.  I didn't need some whole explanation.  Not trying to be rude, just not my business so tbh I don't really care. 

It's a universal bug with the Intel processor microcode.  You can be running Solaris that magically supports Intel x86 and still be vulnerable to this bug. 

I would agree with that.  I'd expect some article to be out by now using a quote from a 12 year old saying how his brand new PC died because he ran too many applications, and highlighting on Intel's bug. 

GCC can still be used on Windows.  I guess just be careful what you run, or look into it and check what C compiler they used to compile the application.  I'm not knowledgeable in this area. 

Wow, kid. Edgy as fuck. Next time don't say dumb shit so confidently.

[Misanthrope]

While the high frame rate case scenario is becoming more popular, it's still a somewhat minority meaning for anything above 1080p or above a 1060 or so, it really might as well not matter.

 

So hyperthreading does matter...a bit..and only on very specific high framerate cases.

[ivan134]

8 minutes ago, Misanthrope said:

While the high frame rate case scenario is becoming more popular, it's still a somewhat minority meaning for anything above 1080p or above a 1060 or so, it really might as well not matter.

 

So hyperthreading does matter...a bit..and only on very specific high framerate cases.

Average framerate isn't everything. Getting 300 fps (redundant, but that's not the point) in some areas and dipping down the 110 fps (this makes hitscan heroes hard to play) in others in Overwatch, with all settings on low, is the reason I'm switching to an i7. Hyperthreaded CPUs give better minimums in CPU intensive games. Then there is the issue of stuttering. Fallout 4 and Battlefield 1 are not fun to play with an i5. Fallout 4 made @App4that buy a 4790k.

 

Spoiler

 

 

[sazrocks]

1 hour ago, NvidiaIntelAMDLoveTriangle said:

Oh Debian, just because your OS is broken, doesn't mean the hardware is too.

This is honestly one of the most ignorant statements I have seen a in a long time. Just because you irrationally hate Debian (a distro known for its stability), doesn't mean that you know far more than all the people who work on it.

[MageTank]

Honestly, I don't know how much of an issue this is for the average consumer. I've done copious amounts of stress testing, various degrees of which are far beyond brutal. I am talking hardcore Linpack MKL just to test my masochism. Have not had any issues with any of my software, be it stress test related, or general consumer (games, office, etc). I've owned various Skylake CPU's, from pentiums to the 6700k, and I currently have a lovely 7700k (Thanks @done12many2 for the best IMC i've ever seen) and still have not had any issues with HT.

 

I do recall a couple of games not playing nicely with it, such as an older Final Fantasy title, and some ancient MMO's, but that's it. Stability has never been an issue. The fact that these reports themselves can't quite pinpoint the exact cause of the instability, only makes it seem that more of a non-issue for the general consumer. If you are doing mission critical work, then yes, anything that can potentially cause down-time is a concern, but I wouldn't be standing on the edge waiting for Intel to rush out a Microcode update if your system has been completely fine thus far. 

 

While I know this will draw criticism from the software junkies (and those who pretend to be so), it's not like we have not seen errata on this scale before. I mean, have they even made TSX functional yet? lol...

[porina]

I guess the original report was written for a very different audience than here as it is hard to relate the info in it. In short, they say Intel have released a microcode update for some Skylake CPUs including "model 94 stepping 3". With CPU-z the extended model of my 6700k is 5E = 94, also stepping 3.

 

The Intel errata pdf for Skylake simply lists this issue as "no fix", however the document is dated April. The original article claims a microcode update released in April seemed to fix the problem, so there is a lack of clarity here.

 

Assumption: a bios update dated in or after April might contain the microcode update to fix this. Taking my Asus Maximus VIII Hero as example, the latest bios was released in April. Great? No, as the reported date inside the bios is from January. Keep an eye open for further bios updates which might fix this issue.

 

Should you change anything you're doing? If not mission critical, I'd suggest not. I've been running my 6700k daily since I built my system nearly 2 years ago, including a LOT of 24/7 compute tasks. It has not been error free, but the errors I have seen are due to optimistic OC of either CPU or ram at a given time, even when HT was off. Other Skylake systems, such as i5 so no HT, have not given any detected problems.

[tlink]

4 hours ago, NvidiaIntelAMDLoveTriangle said:

Oh Debian, just because your OS is broken, doesn't mean the hardware is too.

do you have eyes? if its fixed by a microcode update than its not an OS issue since that's an update that doesn't involve changing the OS at any level.

[Dabombinable]

Well..apart from the need to use a better cooler than I would with a 6700K, I'm glad I went with a 4790K instead.

[stephena]

49 minutes ago, NvidiaIntelAMDLoveTriangle said:

Oh Debian, just because your OS is broken, doesn't mean the hardware is too.

Quote

Please note that the defect can potentially affect any operating system
(it is not restricted to Debian, and it is not restricted to Linux-based
systems).

It's a CPU microcode bug. As has been said, it could affect any OS, be it Debian, Windows, Mac, or (as Cool Guy suggested) even Solaris. Don't assume that because an OCaml dev found it that you won't come across it. I believe there are some open-source program installers that actually contain source code and compile it at install time, and potentially other loads could trigger it too. More corporates will now be looking at moving to SkyLake and Kaby Lake, so they might find other ways to trigger this flaw.

That being said...

Quote

Apparently, Intel had indeed found the issue, *documented it* (see
below) and *fixed it*.  There was no direct feedback to the OCaml
people, so they only found about it later.

Quote

Fixes for processors with signatures[1] 0x406E3 and 0x506E3 are
available in the Intel public Linux microcode release 20170511.

Intel may have pushed an update for Windows already, though I wouldn't relax yet, Patch Tuesdays being what they are.

 

[Theorak]

OK, will investigate the steps for my Skylake CPU, I definitly noticed wierdness when using streaming software, but wrote it down to Win10.

[zMeul]

6 hours ago, Alaradia said:

has intel replied to these allegations?

intel has issued microcode updates, but it's unclear if mobo manufacturers actually deployed them into BIOS updates 

 

Quote

Errata: SKZ7/SKW144/SKL150/SKX150/SKZ7/KBL095/KBW095
"Short Loops Which Use AH/BH/CH/DH Registers May Cause Unpredictable System Behavior."

 

Problem:  "Under complex micro-architectural conditions, short loops of less than 64 instructions that use AH, BH, CH or DH registers as well as their corresponding wider register (e.g. RAX, EAX or AX for AH) may cause unpredictable system behavior. This can only happen when both logical processors on the same physical processor are active."

 

Implication: "Due to this erratum, the system may experience unpredictable system behavior."

 

[Dabombinable]

24 minutes ago, zMeul said:

intel has issued microcode updates, but it's unclear if mobo manufacturers actually deployed them into BIOS updates 

 

 

Asus seems to have done an update to their motherboards to fix the Intel ME issues so they should have. And I found out about that particular fix for my H87M Pro an hour ago despite it being a 2013 motherboard (god Haswell is getting old).

[zMeul]

15 minutes ago, Dabombinable said:

Asus seems to have done an update to their motherboards to fix the Intel ME issues so they should have. And I found out about that particular fix for my H87M Pro an hour ago despite it being a 2013 motherboard (god Haswell is getting old).

the ME issues and these HT discovered aren't similar

 

ps: Haswells aren't affected

[Dabombinable]

9 minutes ago, zMeul said:

the ME issues and these HT discovered aren't similar

 

ps: Haswells aren't affected

I know. Point is if manufacturers released updates to fix a problem on motherboards 4 years old, they more than likely will update the bios on their newer motherboards to fix this hyperthreading bug.

[SCHISCHKA]

8 hours ago, mariushm said:

This defect can, when triggered, cause unpredictable system behavior: it could cause spurious errors, such as application and system misbehavior, data corruption, and data loss.

People are jumping to conclusions without reading the post properly.

Quote

On 2017-05-29, Mark Shinwell, a core OCaml toolchain developer,
contacted the Debian developer responsible for the intel-microcode
package with key information about a Intel processor issue that could be
easily triggered by the OCaml compiler.

The mailing list specifies errata KBL095, KBW095. If you google intel.com you will get a pdf that TLDR says CPU spazziness can ocur when an application or OS executes a very specific set of instructions in a very specific way.

 

As far as i can tell Intel published this a few months ago because someone in an opensource project has discovered some of their software is causing this bug to trigger and has released a full public disclosure.

 

I don't know if you would see the same disclosure from Microsoft or other proprietary publishers; i dont care, you can invent your own theories on this.

I don't know if any other software has the exact same instructions that would trigger this bug. You can make up your own theories on this.

 

Are you running the OCaml comiler or running applications that execute those very specific CPU instructions?

 

I wouldnt give a shit unless a software publisher that I am a customer of informs me to upgrade my microcode.

[Morgan MLGman]

Only 2 pages of the thread? If it was about Ryzen and not Skylake/Kaby Lake it'd be at least 10 pages long

As for the thread itself, how does this affect users such as me running a 6700K under Windows 10? What are the risks and what triggers the bug?

[Aphala]

7 hours ago, Sierra Fox said:

unless Intel acknowledges this then I wouldn't worry. unverified claims. and given how long skylake has been out and this is now just an issue, seems a bit off.

After Intels i9 mid life crisis I'd hold my breath for anything coming from Intel. 

[SCHISCHKA]

3 minutes ago, Morgan MLGman said:

Only 2 pages of the thread? If it was about Ryzen and not Skylake/Kaby Lake it'd be at least 10 pages long

I don't think Intel having a software bug is as ground breaking news than Ryzen was.

Its like comparing something that happens everyday to the moon landing.

[Morgan MLGman]

7 minutes ago, SCHISCHKA said:

I don't think Intel having a software bug is as ground breaking news than Ryzen was.

Its like comparing something that happens everyday to the moon landing.

I didn't mean it like that. And by the way, bugs that have been out for a long time undetected are actually much more dangerous than those associated with a new platform launch as those tend to be found and fixed sooner.

 

I meant that with every thread describing an issue with Ryzen there was a horde of people saying Ryzen is the next "Zendozer" etc, I somehow never manage to catch those same people comment on issues with Intel CPUs or if they do, those issues are "suddenly" irrelevant while they may actually be more severe Hypocrisy at its finest.

 

I only worry because I still don't entirely know what that bug means for someone like me running an OC'd 6700K under Windows 10.

[Aphala]

2 minutes ago, Morgan MLGman said:

I didn't mean it like that. And by the way, bugs that have been out for a long time undetected are actually much more dangerous than those associated with a new platform launch as those tend to be found and fixed sooner.

 

I meant that with every thread describing an issue with Ryzen there was a horde of people saying Ryzen is the next "Zendozer" etc, I somehow never manage to catch those same people comment on issues with Intel CPUs or if they do, those issues are "suddenly" irrelevant while they may actually be more severe Hypocrisy at its finest.

 

I only worry because I still don't entirely know what that bug means for someone like me running an OC'd 6700K under Windows 10.

Remember! The Intel fan boys can and will fight to the death! 

[SCHISCHKA]

1 minute ago, Morgan MLGman said:

I only worry because I still don't entirely know what that bug means for someone like me running an OC'd 6700K under Windows 10.

that is why i mentioned in my other post this is a message for Debian users. You should be signed up to mailing lists for the software you really depend on. If it is not mentioned on any of those lists then dont worry about it. If you look into what has been published by Intel, it is a very specific set of instructions that triggers the bug. The bug was discovered by people who are working on the OCaml compiler so I doubt it will affect you. I wouldnt be surprised if Windows updates the microcode without telling you

[Morgan MLGman]

Just now, SCHISCHKA said:

that is why i mentioned in my other post this is a message for Debian users. You should be signed up to mailing lists for the software you really depend on. If it is not mentioned on any of those lists then dont worry about it. If you look into what has been published by Intel, it is a very specific set of instructions that triggers the bug. The bug was discovered by people who are working on the OCaml compiler so I doubt it will affect you. I wouldnt be surprised if Windows updates the microcode without telling you

Thank you for the explanation :3