Intel Ice Lake CPUs Have a System Crashing Bug

[AleksandarK]

Intel CPUs have been rather notorious for system bugs recently. Starting from 2018's Spectre and Meltdown which used speculative execution to exploit systems, the string of new vulnerabilities just continued to this day. Recently we had CrossTalk exploit which represents a threat to cloud providers, where one user could compromise another just by using the same CPU from which the virtual instances are powered. These types of exploits are even more dangerous than ones that require local access, as that is already dangerous by itself. A lot of these issues are said to be ironed out by Intel's new microarchitecture designs like Ice Lake, Tiger Lake, and future revisions.

However, it seems like Intel is encountering some problems with even the latest Ice Lake CPUs when it comes to system bugs. JetBrains, a Czech provider of software development tools has a Java programming language development environment called IntelliJ integrated development environment. It was recently reported that on MacBook Air 2020 and Microsoft Surface Pro models equipped with 10th generation Intel Ice Lake CPUs, IntelliJ IDE causes system restart or a complete OS crash. In the report, the CPU ran in a Linux VM that isolates itself from MacOS so the macOS XNU kernel is not to blame. In the report thread, another user running Windows on Microsoft Surface Pro experienced the crash as well.


Thanks to community testing, we have found out that these issues are not just a software bug, however, it is a rather CPU specific bug that only occurs on Intel Ice Lake processors. Intel recently updated the CPU microcode and there is no improvement. It seems like the IntelliJ IDE has a specific sequence of instructions that trigger Ice Lake CPUs to crash OS. This behavior is concerning as this could be used for a possible exploit. Again, cloud providers are at risk here as if you crash the system the whole instance could crash. Of course server Ice Lake parts are yet to arrive, but the bug could be hidden in the core of the CPU design. Even with the latest microcode update Ice Lake CPUs are still crashing with this software, so we have to see how Intel responds to this.

View at TechPowerUp Main Site

 

[Space Lynx]

That Ryzen 4800x upgrade is looking more and more solid for me. Now if they could just get it together and give me proper drivers for Big Navi I could finally have an all red team build again, come on AMD!!! Stop letting me down!!!

 

[cucker tarlson]

This behavior is concerning as this could be used for a possible exploit.

no,it's concerning because it crashes the system

 

[laszlo]

is not a bug; seems more like the new power saving feature

 

[Gmr_Chick]

That Ryzen 4800x upgrade is looking more and more solid for me. Now if they could just get it together and give me proper drivers for Big Navi I could finally have an all red team build again, come on AMD!!! Stop letting me down!!!

My dude, I hate to break this to you but, AMD can't even seem to get all the bugs sorted out of the current Navi drivers, nearly a year into its existence...what makes you think Big Navi will be any different?

 

[Space Lynx]

My dude, I hate to break this to you but, AMD can't even seem to get all the bugs sorted out of the current Navi drivers, nearly a year into its existence...what makes you think Big Navi will be any different?

vast majority of navi driver bugs have been solved, as is apparent by the guy you sold your 5700xt nitro too and he said he is still loving that card as you told me? so he must not be having too many issues to say that. i think you just have had very bad luck with motherboards on your amd journey unfortunately. AMD has shown improvement, but not enough for me to buy big navi yet, i am leaning towards 4800x ryzen and rtx 3080 still but we will see. it all depends on gamersnexus review of big navi

 

[Gmr_Chick]

vast majority of navi driver bugs have been solved, as is apparent by the guy you sold your 5700xt nitro too and he said he is still loving that card as you told me? so he must not be having too many issues to say that. i think you just have had very bad luck with motherboards on your amd journey unfortunately. AMD has shown improvement, but not enough for me to buy big navi yet, i am leaning towards 4800x ryzen and rtx 3080 still but we will see. it all depends on gamersnexus review of big navi

Have they REALLY though? Because as far as I can tell, I still read about owners having issues with theirs... Perhaps most puzzling is the fact that even though people's issues seem to be consistent, AMD can't actually replicate them. At any rate, at least for me, Navi can go screw itself.

 

[Space Lynx]

Have they REALLY though? Because as far as I can tell, I still read about owners having issues with theirs... Perhaps most puzzling is the fact that even though people's issues seem to be consistent, AMD can't actually replicate them. At any rate, at least for me, Navi can go screw itself.

it does seem 50/50. i hear good i hear bad. i never had any issues with my gtx 1070. not one. so i am leaning towards ampere that is certain

 

[cucker tarlson]

it does seem 50/50. i hear good i hear bad. i never had any issues with my gtx 1070. not one. so i am leaning towards ampere that is certain

what if ampere is a bug festival and rdna2 turns out to be problem free ?

 

[Space Lynx]

what if ampere is a bug festival and rdna2 turns out to be problem free ?

like I said gamersnexus is the only youtuber i trust anymore, so we will see what he has to say, then i will decide

 

[Assimilator]

This is pretty bad. Like really, really bad. As in Pentium F00F bug levels of bad.

Intel is going to have to issue a hardware recall if they can't patch this in microcode. Even if they can, it's almost certainly going to cause significant slowdown a la Bulldozer TLB. I smell a class action lawsuit...

 

[stimpy88]

No wonder Apple is ditching Intel soon. Between Intel's bad performance, power and thermal issues, they also have to contend with the weekly microarchitecture flaws, firmware bugs and security issues.

 

[john_]

Intel bugs? So what? Who cares. OEMs don't care any way. And that's the point.

 

[CandymanGR]

At last, intel's true colors have shown.

 

[Shou Miko]

Have they REALLY though? Because as far as I can tell, I still read about owners having issues with theirs... Perhaps most puzzling is the fact that even though people's issues seem to be consistent, AMD can't actually replicate them. At any rate, at least for me, Navi can go screw itself.

The problem and majority of issues AMD has with their NAVI GPU's ain't just replicating them but when it happens on different hardware all around it doesn't make it easier to replicate in their test environment.

Same goes for reviewers like some had massive problems before, on and a good a mount of time after release date of the Navi cards and others had no issues at all.

Even some users with the same hardware setup can doesn't need to have the same issues and this is what causes the headache for AMD.

I know that AMD takes every NAVI issue reported seriously but if they as said before cannot replicate it in their test environment they cannot fix them as easy you change a setting your card crashes AMD changes the same setting and the card crashes then it's easier then they cannot get the same settings to crash.

It's not like every motherboard vendor send AMD a copy of every board that's made to test with.

 

[Bruno Vieira]

Have they REALLY though? Because as far as I can tell, I still read about owners having issues with theirs... Perhaps most puzzling is the fact that even though people's issues seem to be consistent, AMD can't actually replicate them. At any rate, at least for me, Navi can go screw itself.

Since the 20.3.1 I havent experienced any issues, with 5 games that Ive been playing. Its very stable 20.1.x and 20.2.x were a big problem. I have a 5700xt

LLT made a video recently and there is at least one Ice lake bug related to after fx.

 

[yeeeeman]

Since the 20.3.1 I havent experienced any issues, with 5 games that Ive been playing. Its very stable 20.1.x and 20.2.x were a big problem. I have a 5700xt

LLT made a video recently and there is at least one Ice lake bug related to after fx.

Linus has an in development platform, crashing during a game. That is totally expected.
This article, if you took the time to read it and understand it, explains a totally different scenario where using a Java tool will in some specific case give you a crash.

 

[1d10t]

Now I understand why they back ported to 14nm again

 

[Bruno Vieira]

Linus has an in development platform, crashing during a game. That is totally expected.
This article, if you took the time to read it and understand it, explains a totally different scenario where using a Java tool will in some specific case give you a crash.

In the first paragraph I spoke about my own experience with an AMD gpu, since people are talking about unstable drivers.

The second paragraph I cited an example about another known ice lake bug, probably not related to the bug reported by the article but affects many users.

 

[Caring1]

At last, intel's true colors have shown.

I bet they are feeling blue

 

[thesmokingman]

I wish I could say I was surprised...

 

[John Naylor]

Still waiting for a real world impact associated with any one of the varied vulnerabilities laid at AMds or Intel's feet. Who was hurt ?

 

[amdog]

That Ryzen 4800x upgrade is looking more and more solid for me. Now if they could just get it together and give me proper drivers for Big Navi I could finally have an all red team build again, come on AMD!!! Stop letting me down!!!

You want good drivers for a new architecture? are you mad boi? i just recently got B A R E L Y decent drivers on my Radeon VII. Now, please sit and wait a year and half.

 

[Punkenjoy]

It's always easier to be stable when you lead your market.

When you have to rush product asap on the market to avoid loosing too much, you can end up with issues.

 

[RJARRRPCGP]

Is this going to be a repeat of the mid-1990s? Where you have to chuck the processor, because of the Pentium 60 to Pentium 120 having a basic math problem, where IIRC, it can become obvious in the Windows calculator.