Intel Chips Have Memory Access Design Flaw and Fix Could Lead to Performance Drop

[cmaier]

Smaller market share, now pretty much every Mac in use is intel based across not just a select computer user base [IE remember that the moves mentioned where when Apple was really catering for a small amount of personal users and many professional users]...

However, if apple is trying to merge app stores..either some Xcode tool to convert or a possible arch change...

Apple still has a small market share (i.e. for mac). And more and more software is subscription based. Adobe and Microsoft would get on board, Apple’s own stuff would be fine, and most of the Indy developers would jump to ARM too (most stuff in the MAS, for sure). That would cover the vast majority of what most people use. Much of the rest would come over or would be replaced by new options.
[doublepost=1515008025][/doublepost]

Intel has a tool on their website to check if your CPU is vulnerable but there is no Mac OS version!

You can compile it yourself in Xcode. Here’s the source code:

func am_i_affected {
print(“yes”)
}

 

[matjamca]

If there is such a performance drop (30%) following a fix, do you think manufacturers will have to compensate customers as the product is no longer ‘as described’ and may ‘not be fit for purpose’?

I have a mid-2012 15” Retina MacBook Pro.

 

[Edsel]

Introducing the 2018 Mac Pro:


View attachment 745087

I still have my vintage 2004 liquid-cooled PowerPC. It's time to re-activate that old monster!

 

[PKoz]

No, but some of them are. Point is this wasn’t an easy, obvious flaw, and Apple might suffer something similar.

That's quite the leap. Just because it isn't obvious to the masses that simply use these chips, doesn't mean it is a widely utilized design trap that is prone to catching designers. We have not heard of this happening to AMD, or the GPU designers or the ARM designers, etc. So far you merely have an unsubstantiated opinion based on nothing it seems.

 

[cmaier]

That's quite the leap. Just because it isn't obvious to the masses that simply use these chips, doesn't mean it is a widely utilized design trap that is prone to catching designers. We have not heard of this happening to AMD, or the GPU designers or the ARM designers, etc. So far you merely have an unsubstantiated opinion based on nothing it seems.

It actually requires some extreme design laziness to speculatively execute instructions that will have results tossed due to memory isolation. This should have been caught, but probably the person who did the design worked on the same block for 10 years and no one else ever looked at it.

 

[SRLMJ23]

Apple still has a small market share (i.e. for mac). And more and more software is subscription based. Adobe and Microsoft would get on board, Apple’s own stuff would be fine, and most of the Indy developers would jump to ARM too (most stuff in the MAS, for sure). That would cover the vast majority of what most people use. Much of the rest would come over or would be replaced by new options.
[doublepost=1515008025][/doublepost]
You can compile it yourself in Xcode. Here’s the source code:

func am_i_affected {
print(“yes”)
}

I was under the impression that EVERY Intel CPU made over the past 10 years was affected? At least according to the article on this site: "All modern computers with Intel chips from the last 10 years appear to be affected, including those running Windows, Linux, and macOS. "

So are you saying that only some CPU's are affected or all?

 

[cmaier]

I was under the impression that EVERY Intel CPU made over the past 10 years was affected? At least according to the article on this site: "All modern computers with Intel chips from the last 10 years appear to be affected, including those running Windows, Linux, and macOS. "

So are you saying that only some CPU's are affected or all?

I'm saying they all are

 

[itguy06]

what In the hell?

It's Intel's crappy chips.
[doublepost=1515009756][/doublepost]

We have not heard of this happening to AMD, or the GPU designers or the ARM designers, etc.

AMD has publicly stated their chips do not and have not contained this flaw.

 

[Kra-Kow]

For all those worried about OSX... For what it is worth:

https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/

"Finally, macOS has been patched to counter the chip design blunder since version 10.13.2, according to operating system kernel expert Alex Ionescu."

I haven't noticed any difference. Have you?

 

[itguy06]

That said, for someone like me who doesn't grant remote access to other people to their machines, and who runs only a few programs from reputable developers, e.g., Adobe, Bare Bones Software, Bombich, etc., there seems to be no chance my security will be compromised.

Just curious if my understanding of all this is correct, or am I missing something?

You are missing one HUGE thing. That's assuming those softwares were not compromised when you downloaded them. There have been a few cases of malicious software getting into developers' official distributions.

 

[mwined]

Apparently PCID can ameliorate the performance penalty, if apple decides to go that route.
You can find out if your cpu is recent enough by using the command line.

Typhon:~ jeremy$ sysctl machdep.cpu.brand_string
machdep.cpu.brand_string: Intel(R) Core(TM) i5-4690 CPU @ 3.50GHz
Typhon:~ jeremy$ sysctl machdep.cpu.features
machdep.cpu.features: FPU VME DE PSE TSC MSR PAE MCE CX8 APIC SEP MTRR PGE MCA CMOV PAT PSE36 CLFSH DS ACPI MMX FXSR SSE SSE2 SS HTT TM PBE SSE3 PCLMULQDQ DTES64 MON DSCPL VMX SMX EST TM2 SSSE3 FMA CX16 TPR PDCM SSE4.1 SSE4.2 x2APIC MOVBE POPCNT AES PCID XSAVE OSXSAVE SEGLIM64 TSCTMR AVX1.0 RDRAND F16C

Or, see a bunch of stuff with: sysctl machdep

 

[SRLMJ23]

I'm saying they all are

Ahh, okay. So what is the point of Intel releasing a tool for Windows (NOT macOS) to see if your chip is affected or not? Dumb. This is going to be one of the biggest class action lawsuits in U.S. history! Think of how many computers over the past 10 years with Intel CPU's are out there! Also, with Microsoft, you have to take in to account their Surface lineup since they all use Intel CPU's.

Would be really awesome if we all got new CPU's, I know I am dreaming on that one but damn, I am not taking a 30% hit in performance!


[doublepost=1515010484][/doublepost]

For all those worried about OSX... For what it is worth:

https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/

"Finally, macOS has been patched to counter the chip design blunder since version 10.13.2, according to operating system kernel expert Alex Ionescu."

I haven't noticed any difference. Have you?

Zero. Hmmm.

 

[cmaier]

Ahh, okay. So what is the point of Intel releasing a tool for Windows (NOT macOS) to see if your chip is affected or not?

could be that there are (will shortly be) some fixed chips out there, or that you have an 11 year old chip and aren't sure how old it is.

 

[smoledman]

This is absolutely massive news.

How so?

 

[cmaier]

How so?

Having to constantly flush the TLB can be a pretty massive performance hit for most types of workloads.

 

[SRLMJ23]

I have a 2015 MacBook Pro, with a Intel Broadwell chip in it. I will assume that it is affected!

Intel better do something really cool to help us out. I am not buying a new MacBook Pro to get a new Intel CPU which could still be affected! Intel should send me an upgraded version on my chip, based on socket size, and make sure it is not vulnerable to any attacks! Wishful thinking. 5-30% decrease is just unacceptable!

 

[pastrychef]

For all those worried about OSX... For what it is worth:

https://www.theregister.co.uk/2018/01/02/intel_cpu_design_flaw/

"Finally, macOS has been patched to counter the chip design blunder since version 10.13.2, according to operating system kernel expert Alex Ionescu."

I haven't noticed any difference. Have you?

No, I have not noticed any performance impact.

 

[sterlingindigo]

Thirty effin percent?! What in the actual...?!
So the $3k 2017 MBP I bought a few months ago...30%?!?!
Apple will give me $900 back right?

 

[pastrychef]

Thirty effin percent?! What in the actual...?!
So the $3k 2017 MBP I bought a few months ago...30%?!?!
Apple will give me $900 back right?

Are you on macOS 10.13.2? Have you noticed any performance impact? The flaw has already been addressed in 10.13.2.

 

[Patrice Brousseau]

Wow it's absolutely hilarious to watch all the arm-chair computer experts on these forums jump to conclusions. Similarly poor form on MR to post this to front-page until more info is officially released. Based on the info floating around in the last few hours, there appears to be minimal impact (if none) to the majority of normal users, including gamers. Albeit on linux, gaming results pre/post are basically identical with no impact:

https://www.phoronix.com/scan.php?page=news_item&px=x86-PTI-Initial-Gaming-Tests

But go ahead everyone jump to conclusions and complain before you even know what you are dealing with.

Perhaps those in the big boy cloud server world will have an issue on their hands, but at initial glance the majority of users will see zero impact so let's just wait and see how this thing plays out.

So far, after reading half of the replies, the first words of reason I read... thanks.

[sarcasm mode on]

- it's the end for Intel;
- everyone sees 30 percent hit, without much benchmarks, proofs or even patches at least on the Windows side;
- Apple will go ARM for sure...

[sarcasm mode off]

This site is named MacRumors for an obvious reason

Let's wait the official patches of Microsoft and Apple with real use benchmarks and then we'll see...

 

[DaveP]

So far, after reading half of the replies, the first words of reason I read... thanks.

[sarcasm mode on]

- it's the end for Intel;
- everyone sees 30 percent hit, without much benchmarks, proofs or even patches at least on the Windows side;
- Apple will go ARM for sure...

[sarcasm mode off]

This site is named MacRumors for an obvious reason

Let's wait the official patches of Microsoft and Apple with real use benchmarks and then we'll see...

Nonsense. It's much easier to have righteous indignation based on perceptions rather than reality. Immediate overreactions based on incomplete information are the best.

 

[thekev]

Thirty effin percent?! What in the actual...?!
So the $3k 2017 MBP I bought a few months ago...30%?!?!
Apple will give me $900 back right?

You don't need to freak out about this, because you will probably never notice the difference. That comment was made as a worst case scenario with regards to Linux under assumptions that do not match real world use. It was just a long stream of system calls. Does that sound normal to you? If not, calm down.

Of course if the impact really was 30% on initial patch attempts, they would spend a lot more time working on ways to mitigate the problem.

 

[Glideslope]

motherboard with local bus or PCI ?

PCI is nicer, so you can use S3 Trio or Virge for graphics card

Local Bus. LOL, I’ve been having flashbacks all day from this. It was my last PC.

 

[Kabeyun]

You won’t give it up because of a subsequent design flaw you didn’t know about until just now?

Personally I’ll take the soon-to-be-addressed-in-software-update problem over a 15-year-old microprocessor design, but to each his own.

Validated. Knew it would be soon. Didn’t know it’d be that soon.

 

[PickUrPoison]

Thirty effin percent?! What in the actual...?!
So the $3k 2017 MBP I bought a few months ago...30%?!?!
Apple will give me $900 back right?

Why would you look to Apple to compensate you for Intel’s screwup?