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

[CartoonCat]

All the new Macs including the new and upcoming MacPro should be using AMD Processors. They are faster than Intel's i7 use less power, and cost half as much. Take a look at the new AMD "Thread Stripper" it is great. Doubtful Tim is aware of this though, and will probably continue to use Intel. I hope I'm wrong.

 

[jerwin]

On the other hand, this will mean that iPhones will regain their precious geekbench scores, as those are calibrated against a Intel Core i7-6600U (4000 pts)--which will now be slower.

 

[cmaier]

All the new Macs including the new and upcoming MacPro should be using AMD Processors. They are faster than Intel's i7 use less power, and cost half as much. Take a look at the new AMD "Thread Stripper" it is great. Doubtful Tim is aware of this though, and will probably continue to use Intel. I hope I'm wrong.

The problem with AMD is they have always been a one hit wonder place. They do a good job then their politics destroys the team that did it and they fall behind again for a long time. I saw it close up after opteron.

 

[AdonisSMU]

Holy Molly!!! This is HUGE!

 

[adam1080]

Apple: We might see some backlash for making people’s phones run slower.

Intel: Hold my beer…

 

[Deacon-Blues]

Well I paid $2800 for a brand new MacBook Pro 15" Touch Bar in April. That is the most I've ever paid for a computer, over $800 more than I paid for my last MacBook Pro. Now I am told my high priced high performance computer is going to take a substantial performance hit. Deacon is unhappy. I don't blame Apple, I blame Intel. But I am very, very displeased.

 

[1k9]

Ok so does this mean if you just bought a new computer (iMac, MacBoo Pro) and your in your return period still you should return them and wait this out? I’ve got a lot of money on the line here.

 

[MacHiavelli]

If it is a hardware flaw that impacts on performance, the products aren't fit for purpose and we're mis-sold. Most customers will be able to ask for exchanges, especially those in Europe. Don't put up with Intel fobbing you off with a software fix.

 

[firewood]

ARM on the low end, maybe. (I don’t see why, unless upcoming Core-Y is terrible.) But across the lineup causes more issues than it solves.

According to the "Innovators Dilemma" a highly successful low-end will eventually eat the high-end. Just like 8 inch disk drives, much faster than the early smaller HDs, are now obsolete, even for enterprise servers. Multiple large corporations using large product volumes to pour money into competing for the fastest ARM processors will eventually overwhelm the x86 ecosystem. So far, Apple looks like the Top Dog in this competition.

 

[gwaizai]

Forget 2018 The Year of the Dog, how about 2018 The Year of the Slowdown?

 

[EightyTwenty]

If it is a hardware flaw that impacts on performance, the products aren't fit for purpose and we're mis-sold. Most customers will be able to ask for exchanges, especially those in Europe. Don't put up with Intel fobbing you off with a software fix.

What will you exchange them for exactly? The issue affects every Intel chip made in the last 10 years. Going to exchange for a Core2Duo from 2007?

 

[chucker23n1]

According to the "Innovators Dilemma" a highly successful low-end will eventually eat the high-end. Just like 8 inch disk drives, much faster than the early smaller HDs, are now obsolete, even for enterprise servers. Multiple large corporations using large product volumes to pour money into competing for the fastest ARM processors will eventually overwhelm the x86 ecosystem. So far, Apple looks like the Top Dog in this competition.

It never happened with PowerPC or MIPS or any of the other contenders in the 90s, though. And so far, the high-end ARM efforts are all about low-power systems, not high performance.

Yes, it could happen. But cmaier treats it as a foregone conclusion. I don’t.

And I’m still waiting on an answer on why anyone is supposed to trust that a bug that took ten years to discover and happened to a company with decades of experience in CPU design wouldn’t happen to a company that… doesn’t have that experience at all.

 

[Stella]

Arm still isn’t up to the performance of Intel. Until that happens, there’s no point and a complete over reaction.

time to go AMD or ARM

m

 

[Aldaris]

Ok so does this mean if you just bought a new computer (iMac, MacBoo Pro) and your in your return period still you should return them and wait this out? I’ve got a lot of money on the line here.

Possibly... (pure speculation here) chances are it’ll be months or years before intel chips are clean of this “bug” (at least coming out of the foundaries). I’m personally skeptical of an OS update and will wait to let others see what kind of real world hit they take. As others have said you practically need physical access to the machine, so I’ll be waiting this one out. I can’t imagine anything above a 10% hit to my workflow personally, 30% seems crippling to me. Really a nasty deal.

 

[Stella]

There’s a ten year old Mac OS vulnerability too. Are you asking the same questions to Apple? How could Apple not have discovered this bug years ago?

The answer is - software and cpu’s are complex beasts. I suspect there are issues with the Arm CPUs Apple use that haven’t been discovered.

It never happened with PowerPC or MIPS or any of the other contenders in the 90s, though. And so far, the high-end ARM efforts are all about low-power systems, not high performance.

Yes, it could happen. But cmaier treats it as a foregone conclusion. I don’t.

And I’m still waiting on an answer on why anyone is supposed to trust that a bug that took ten years to discover and happened to a company with decades of experience in CPU design wouldn’t happen to a company that… doesn’t have that experience at all.



Found the conservative snowflake.

 

[bsolar]

Probably won‘t because they‘re fixing it at the OS level for everyone. I doubt you will be able to choose whether or not to use kernel page table isolation.

As far as I understand at least in Linux the workaround is not activated per default on AMD processors. You can toggle it via boot parameters in any case.

 

[cmaier]

It never happened with PowerPC or MIPS or any of the other contenders in the 90s, though. And so far, the high-end ARM efforts are all about low-power systems, not high performance.

Yes, it could happen. But cmaier treats it as a foregone conclusion. I don’t.

And I’m still waiting on an answer on why anyone is supposed to trust that a bug that took ten years to discover and happened to a company with decades of experience in CPU design wouldn’t happen to a company that… doesn’t have that experience at all.



Found the conservative snowflake.

A company is just a collection of people. This collection of people has decades of experience. The fact that they now get their paychecks from Apple and not some other company doesn’t mean their experience doesn’t count.

 

[oldmacs]

Apple was a founder of the PowerPC consortium, which for a period was making faster desktop processors than Intel. In fact, the current IBM Power mainframe still appears to be faster than any Intel server.

Yeah PPC chips were faster for a period, but that was a long time ago.

That’s just crazy. Right now, as we speak, developers target Apple Watch and iPad in the same Xcode project. Yet iPad seems to do a lot more with its apps than Apple Watch. You don’t run the same software on both. You share code but target each platform individually in the same project. You keep writing these conclusions without any evidence or even logical support.

Look what happened to iWork when they brought the Mac and iOS versions into line. We lost a lot of features so that the iOS and Mac versions were aligned.

Maybe i’m looking at it wrong, but I can just see a lot of Mac software being rewritten from the ground up to be in line with an iOS version and thus loosing features.

Just like when Everyone said Apple couldn’t do a phone, and Apple couldn’t do a mobile processor, and Apple couldn’t do a watch.

They have tons of senior engineers who have designed top of class desktop processors, none of which suffer from this intel bug. They’ve shown that their design skills are superior to the competition by eking out significantly higher speed than the competition in the mobile space. They have even designed processors FOR MACS

They’re different scenarios. I can see you absolutely hate Intel.

You are also wrong about “emulating everything.” Most instructions executed by the cpu are from the OS. The OS will be native. A small percentage of the code would have to be recompiled or emulated.

In reality Apple won’t make it simple and easy like this. They screwed me around with yanking Rosetta support.

I get that you’re obviously very knowledgeable about processors, but If Apple jumps again it will be very painful, and yes we know this as they’ve done it twice before. As i’ve allready said, past experience has shown that those of us on the old platform will be screwed, software will be unnecessarily left behind, we’ll have to deal with years of stuffing around including a non optimised operating system and non optimised software, the new hardware won’t be worth buying until 2-3 years of revisions are made etc.

Your stated benefits aren’t exactly drawn from tangible evidence and even then aren’t even that attractive given the negatives.

I’ve made my point more than enough times and i’m done here for the moment.

 

[Schnegg]

First reaction: Shocker, Intel is gonna die!

Reality is, all computer systems have terrible bugs that are exploitable. All we can hope for is that they are not trivial like the recent root thing.

Only very simple systems are secure, complex ones connected to the web aren‘t. I sometimes wonder what experienced Cobol hackers could do to the world with alle the banking and insurance stuff still needing it.

 

[1k9]

All iMacs and MacBook Pros 50% off until further notice. After all they are now selling known defective products, correct?

 

[dampfnudel]

As much as I hate Intel, I’m sure the lions share of the settlement will go to some rich punk law firm and we’ll be stuck holding $5 rebates on our next Intel CPU purchase...

It will be Apple, HP, Dell, etc. who will have to compensate us (if the courts decide in our favor). Intel will have to make amends with them and their losses. It has the potential to be the biggest technology screwup in history and given the negative consequences if Intel had to shoulder the full burden, the US government will probably have to step in somehow. Maybe shielding Intel and telling us in the interest of economic prosperity and security globally that we all should accept our performance hit gracefully. At best, maybe $100 off our next computer.

 

[PickUrPoison]

I’m not speaking for anyone. But if you don’t think the future of Apple is ARM across the board you haven’t been paying attention.

I hear what you’re saying, and sure the A11 is an amazing chip, but how do you get single/dual/quad threaded performance equivalent to a 90W Intel processor out of the few Watts an A11 dissipates? Sure if I were virtualizing servers or otherwise had a bunch of VMs, or was rendering or transcoding or doing FEA or had similar tasks I might rather have 16 A11. But many workloads just aren’t amenable to throwing a hundred cores at it.

I guess maybe the instruction set gets more specialized/optimized (CISC-like?) and clock speeds go up, the CPU becomes more complex and the die size increases... but there is quite a performance gap at the moment. I suppose it’s just a matter of time though. Eventually I can see MacOS running on Apple parts though, I would think 7-10 years away though.

 

[getalot681]

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.

 

[frompast]

So after all..

Mac Pro 1,1 save the world

 

[Applebot1]

This could be the shake up the cpu industry needs. Shocking by Intel and refuse to believe they only known about this recently.