Intel Hyperthreading Bug Unearthed In Kaby Lake And Skylake Processors

[Guest]

According to Henrique de Moraes Holschuh, a Debian Linux developer, Kaby Lake and Skylake processors have a serious flaw in their hyperthreading implementation.

Intel Hyperthreading Bug Unearthed In Kaby Lake And Skylake Processors : Read more

 

[leoscott]

There's a reason that buying the latest, greatest when it is first released is called bleeding edge.

 

[TJ Hooker]

There's a reason that buying the latest, greatest when it is first released is called bleeding edge.

Skylake has been out for close to two years. Even Kaby Lake is about half a year old. It really isn't "first released" behaviour at this point, and Skylake is in no way bleeding edge.

 

[JamesSneed]

You can look up all the erratas for Skylake this one is the last in the list. Its a rather long list. This was linked in the article the point is there is stuff in here that is nothing to do with bleeding edge, lots of old erratas with no fix.

https://www.intel.com/content/www/us/en/processors/core/desktop-6th-gen-core-family-spec-update.html

 

[DookieDraws]

Isn't Haswell also affected by this very same issue?

 

[Deleted member 1353997]

This may sound weird, but I'm actually happy about this news.

I've been having problems with Windows 10 hanging almost every day ever since I built a new PC with a Skylake i7 6700K CPU. Despite tons of researching, I never found an explanation for this weird problem, not to mention a solution.

This news is the first time since long that I've seen something that might be related to my issue. Just thinking that I may finally be able to solve my problem by disabling HT fills me with joy.
It sucks, but disabling HT is significantly better than buying a new mainboard and CPU.

 

[ammaross]

This may sound weird, but I'm actually happy about this news.

I've been having problems with Windows 10 hanging almost every day ever since I built a new PC with a Skylake i7 6700K CPU. Despite tons of researching, I never found an explanation for this weird problem, not to mention a solution.

This news is the first time since long that I've seen something that might be related to my issue. Just thinking that I may finally be able to solve my problem by disabling HT fills me with joy.
It sucks, but disabling HT is significantly better than buying a new mainboard and CPU.

No, this is very likely not your issue.

 

[ffleader1]

It's kinda funny to think that after 5 years without competitor and just need to copy/paste the same architecture over and over again to make profit, Intel still managed to let a bug happen, on two latest generations.

 

[nzalog]

You can look up all the erratas for Skylake this one is the last in the list. Its a rather long list. This was linked in the article the point is there is stuff in here that is nothing to do with bleeding edge, lots of old erratas with no fix.

https://www.intel.com/content/www/us/en/processors/core/desktop-6th-gen-core-family-spec-update.html

Literally all the erratas have no fix.

 

[bigpinkdragon286]

Any chance of Tom's running a few benchmarks before and after the updates to see if performance changes as a result?

 

[jaber2]

This explains why I'm getting such low frame rates

 

[TechyInAZ]

Good thing my entire family all runs haswell and skylake core i5s.

 

[Job___]

its " leading edge"
stupid ******

Watch the language. This is a family friendly site.

 

[nzalog]

its " leading edge"
stupid ******

Why exactly is that necessary?

Also many people do call it 'bleeding edge'...

 

[bit_user]

Isn't Haswell also affected by this very same issue?

Why do you think so? Neither the article nor the post on the mailing list say anything about Broadwell or Haswell.

The bug was probably introduced in the development of Skylake and carried over to Kaby Lake (there's probably some amount of code reuse between them).

 

[bit_user]

This explains why I'm getting such low frame rates

No, probably not.

 

[bit_user]

its " leading edge"
stupid ******

Why exactly is that necessary?

Also many people do call it 'bleeding edge'...

It does bug me when people use "leading edge" and "bleeding edge" interchangeably.

I think "bleeding edge" was originally created in reference to alpha/beta-quality stuff (usually software), where as "leading edge" traditionally just meant the latest release or the newest product. But I guess enough people decided they liked the sound of "bleeding edge" so much that they use it exclusively, and it now seems to be displacing the original term even where it would be more appropriate.

 

[nzalog]

its " leading edge"
stupid ******

Why exactly is that necessary?

Also many people do call it 'bleeding edge'...

It does bug me when people use "leading edge" and "bleeding edge" interchangeably.

I think "bleeding edge" was originally created in reference to alpha/beta-quality stuff (usually software), where as "leading edge" should probably be a bit more mature. But I guess enough people decided they liked the sound of "bleeding edge" that they use it exclusively to mean the latest of whatever.

I've always thought that the "bleeding edge" was just a play on "leading edge" since new stuff tends to have bugs that aren't worked out while leading edge just means the latest cool stuff. Hence why people use it interchangeably because the latest cool stuff tends to have a bunch of patches coming to it before it's considered 100% stable.

 

[nzalog]

Good thing my entire family all runs haswell and skylake core i5s.

The article said all Skylakes. So, you'll want to update the BIOS on those Skylake i5's, when it's available.

i5's don't have no hyperthreading, so no bug either.

 

[bit_user]

Good thing my entire family all runs haswell and skylake core i5s.

The article said all Skylakes. So, you'll want to update the BIOS on those Skylake i5's, when it's available.

i5's don't have no hyperthreading, so no bug either.

*sigh*

Thanks for catching that.

 

[bit_user]

There's a reason that buying the latest, greatest when it is first released is called bleeding edge.

I typically steer clear of the first few steppings of any new CPU. But after a few months, I figure most of the more significant errata have been fixed, and any show-stopper bugs probably would've surfaced by then.

Maybe that's not as sure as I assumed, but there's no way I'm buying CPUs from 2+ generations ago, just to be extra sure they're solid. Obviously not speaking about mission-critical systems, here.

 

[Jasjar]

It's kinda funny to think that after 5 years without competitor and just need to copy/paste the same architecture over and over again to make profit, Intel still managed to let a bug happen, on two latest generations.

Well a bug is going to keep happening if you don't make any changes to the chip. Intel literally just got Skylake and just made it better at overclocking. But you can't even overclock to the improvements, because Intel who wanted to save $1 on manufacturing on a $400 CPU, used bad thermal paste, instead of soldering the IHS. The response Intel gave was "Stop overclocking if you don't want high temps". Yes, don't overclock on a CPU that was sold to overclock.

 

[psiboy]

Ryzen for the win! lol 😛

 

[bit_user]

Ryzen for the win! lol 😛

It has bugs, too.

https://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Compiler-Issues

Kinda hard to imagine you could launch a big, totally new architecture without a few slipping through.

 

[lilyammy]

fml. Here I was getting ready to snuggle in for an evening of running sets of 63 instructions using BH and EAX registers on both logical cores of my same physical processor.