INTC - Intel

[mbharadwaj]

Google's post on the vulnerability

 

https://googleprojectzero.blogspot.com/2018/01/reading-privileged-memory-with-side.html

[rb]

From the posts it is apparent that it's not only intel that's affected but also AMD and ARM or maybe just certain chips from AMD and ARM?

[pau_]

According to the papers, Intel's are the only ones known to be affected by Meltdown, but the paper leaves open the possibility that other manufacturers' processors are affected. This is the one that has a mitigation, and the one that got the headlines about the reduced performance.

 

The other attack, Spectre, ("harder to exploit, harder to mitigate") has no single mitigation if I'm reading correctly, and it affects Intel, AMD (in a non-default configuration), and ARM.

[rb]

I don't even know what this actually means in real life.

 

Basically the OS vendors will roll out patches that will fix the security issues but will brick the machines. As a user I see two options:

 

1. Upgrade the systems to faster chips to make up for the difference in power. So pay a bunch of money for more powerful chips that will be just as fast as the old ones because the more powerful chips are also affected. So pay money for nothing.

 

2. Wait for a new generation of chips to arrive that don't have a this vulnerability. In the meantime you get to operate a sewing machine.

 

Both options really suck!

[mbharadwaj]

Very informative tweets from Nicole Periroth (NYT, Cyber Security):

 

It is not threaded. Do refer to comments 1 trough 9.

[tng]

I wonder how far down the pipeline this affects. They probably need to finalize their chips a very long time before they hit the market. Whatever chips Intel have under development would presumably be affected by this issue too. How long will it take before Intel can get a new chip to market that doesn't have this vulnerability? It could be disastrous if it is over a year.

[Liberty]

[mbharadwaj]

I wonder how far down the pipeline this affects. They probably need to finalize their chips a very long time before they hit the market. Whatever chips Intel have under development would presumably be affected by this issue too. How long will it take before Intel can get a new chip to market that doesn't have this vulnerability? It could be disastrous if it is over a year.

 

afaik, it will take them more than a year. the biggest impact seems to be for the performance of virtual machines in cloud infrastructure. Amazon just announced "dedicated instances" at the recent reInvent. that will be popular. and serverless function as a service architectures (like lamdbas in aws, "cloud functions" in google cloud, "azure functions" in azure), and serverless databases like AWS aurora will be very easy to be convinced about in enterprise architecture debates. these will reduce the probability of error creeping in from coding bugs.

[Liberty]

http://www.businessinsider.com/intel-ceo-krzanich-sold-shares-after-company-was-informed-of-chip-flaw-2018-1

[boilermaker75]

http://www.businessinsider.com/intel-ceo-krzanich-sold-shares-after-company-was-informed-of-chip-flaw-2018-1

 

As a former employee and a former long-term share holder (25 years), this disgusts me.

[mwtorock]

typical example of agent problem - these are so called professional managers. they don't have enough skin in the game, only trying to make money. that is the reason I prefer owner operators.

 

on a side note, AMD should be an apparent winner of this situation.

[Liberty]

https://security.googleblog.com/2018/01/more-details-about-mitigations-for-cpu_4.html

[JayGatsby]

Can anyone verify this as true?: "There aren't any processors available at the moment that can replace the vulnerable ones and still provide the same kind of functionality." From CNN.. http://money.cnn.com/2018/01/04/technology/business/apple-macs-ios-spectre-meltdown/index.html

 

Looks like Spectre impacts AMD as well: "These attacks represent a serious threat to actual systems, since vulnerable speculative execution capabilities are found in microprocessors from Intel, AMD, and ARM that are used in billions of devices." https://spectreattack.com/spectre.pdf

 

Meltdown seems more specific to Intel. That's the one Snowden was tweeting about today and seems to be the bigger issue.

 

PSA of the day is everyone should have two-factor authentication for their accounts. IB and Schwab both have a simple little app that goes on your phone.

 

[boilermaker75]

Can anyone verify this as true?: "There aren't any processors available at the moment that can replace the vulnerable ones and still provide the same kind of functionality." From CNN.. http://money.cnn.com/2018/01/04/technology/business/apple-macs-ios-spectre-meltdown/index.html

 

Looks like Spectre impacts AMD as well: "These attacks represent a serious threat to actual systems, since vulnerable speculative execution capabilities are found in microprocessors from Intel, AMD, and ARM that are used in billions of devices." https://spectreattack.com/spectre.pdf

 

Meltdown seems more specific to Intel. That's the one Snowden was tweeting about today and seems to be the bigger issue.

 

PSA of the day is everyone should have two-factor authentication for their accounts. IB and Schwab both have a simple little app that goes on your phone.

 

I have IB and Schwab accounts and have used the two-factor authentication for years. I also have a word with Schwab (I forgot what they call it maybe safe word?) that if I call them on the phone they will not talk to me unless I say the word. If I were to forget the word, I need to go into a branch to reset it.

[mikazo]

Really interesting thread on the timeline and effort of making wide-ranging changes to CPUs:

 

https://twitter.com/securelyfitz/status/949370010652196864

[Liberty]

Not good: https://www.reuters.com/article/us-cyber-intel/intel-did-not-tell-u-s-cyber-officials-about-chip-flaws-until-made-public-idUSKCN1G62PS

[Liberty]

If you're wondering what hit INTC today:

 

https://www.bloomberg.com/news/articles/2018-04-02/apple-is-said-to-plan-move-from-intel-to-own-mac-chips-from-2020

[DTEJD1997]

Hey all:

 

There have been rumors about Apple designing, building, using their own chips since the days of Motorola processors in the very early nineties.

 

I've heard reports that Apple has actually accomplished this in the past...but they did not want to "fork" the software...

 

So they've stuck with Intel.

 

This time might be different.  I've read reports that the processor in the iPad Pro is really worldclass and can give "regular" Intel processors a run for their money.

 

I would think that AAPL could also work with AMD to get "beefed up" Ryzen processors at a "bargain price"...this could also be a significant threat to Intel.

 

We'll see, but I would be that AAPL is going to stick with Intel for the next few years.

[Liberty]

Apple's silicon team is world class, both for CPUs (SOCs, actually) and GPUs now. The A11 in the iPhone X is already basically almost laptop level. Remove the limitations that have to be there because it's running on a tiny battery (clocked higher, doesn't throttle, more aggressively use all cores, etc) and it would be a good foundation. In a couple years I'd imagine they could make something competitive with Intel in performance and probably much better on power consumption because of the ARM inherent efficiency (ie. laptops with much better battery life).

[rb]

It'll be hard to make ARM chips to compete with Xeons. They could try to imitate but won't quite be there. What happens after that?

 

Is Apple gonna build a foundry? That won't be very efficient. So what's the motivation behind this? Is it just we're Apple and we'll do it this way if we feel like it?

[DTEJD1997]

It'll be hard to make ARM chips to compete with Xeons. They could try to imitate but won't quite be there. What happens after that?

 

Is Apple gonna build a foundry? That won't be very efficient. So what's the motivation behind this? Is it just we're Apple and we'll do it this way if we feel like it?

RB:

 

Xeons and ARM chips are addressing 2 very different markets.

 

You are correct, it would probably be almost impossible to bring a ARM architecture chip up to AND surpass a current XEON. 

 

Xeons are for servers & workstations & the highest end PC's. 

 

Apple is going to be using ARM architecture stuff for iPads, consumer laptops, iMacs, low end & medium systems.  If Apple can improve their chips just a bit, they will have all the capabilities they need.

 

I've read reports that say the iPad pro is almost as powerful as a mobile 7th generation Intel i5 processor.  This level of computational power is close to being sufficient for any consumer application/need.

 

The only systems from Apple using Xeons are the Mac Pro's & iMac Pro's.  The iMac Pro is about a $5k system.  The Mac pro has a similar price.  These are a fraction of Apples sales currently....maybe 1 or 2% if even that?  There is supposed to be a update/refresh on the mac Pro systems pretty soon.

 

Heck, most of primary systems are Mac Pro's that are at least 6+ years old.

 

Interestingly, the Mac Pro is probably going to be manufactured in the USA.

 

We'll see.  It should be exciting!  I might even upgrade to the new models!

 

[rb]

Well that's not true. MacBook airs and pros use i7s which is a mid range xeon. Some use i5s - lower range xeons and probably not worth talking about here.

 

The fact is that one of the things that made macs so successful is that they've overpowered their systems. That's one of the main reasons why they were so stable and popular. They idea that x and y chip would be good enough for what the user needs is a radical departure from this strategy.

 

I don't doubt that Apple can design some chips to put in their macs. What eludes me is what they have to gain from it.

[rkbabang]

Well that's not true. MacBook airs and pros use i7s which is a mid range xeon. Some use i5s - lower range xeons and probably not worth talking about here.

 

The fact is that one of the things that made macs so successful is that they've overpowered their systems. That's one of the main reasons why they were so stable and popular. They idea that x and y chip would be good enough for what the user needs is a radical departure from this strategy.

 

I don't doubt that Apple can design some chips to put in their macs. What eludes me is what they have to gain from it.

 

I agree with you.  I have an i7 in my iMac (it is not an iMac Pro).  I think they should segregate their macs into 2 categories.  1) low power/ lower cost IOS only laptops and desktops which only run IOS and have Apple A<x> ARM chip.  And 2) The higher power laptop/desktop machines that have both an Apple A<x> ARM chip as well as a Xeon/i5/i7 intel chip.  These should run IOS concurrently on top of macOS.  In other words you are running macOS, but can run IOS apps in a window.

 

This way if someone wants a super light laptop which only runs IOS then they can get one, but if they wan't something more powerful they can get that too and still be able to run both native macOS and native IOS apps on it without emulation.

 

[Liberty]

It'll be hard to make ARM chips to compete with Xeons. They could try to imitate but won't quite be there. What happens after that?

 

Is Apple gonna build a foundry? That won't be very efficient. So what's the motivation behind this? Is it just we're Apple and we'll do it this way if we feel like it?

 

If they do this, they'll contract them out to the big foundries like TSMC, I think. Maybe Intel even opens up its own fabs to third parties someday, who knows (they have to weigh the benefits of keeping the best lithographic process for themselves vs spreading out the huge fixed costs over more volume).

 

As for competing with Xeons/iX, I think it's more complex than that, but first, let's rewind the clock a bit and wonder if 5 years ago people would've bet on Apple being the best ARM chip maker out there and giving a run for its money to Qualcomm and others? And now they've just begun making their own mobile GPUs and are already kicking ass.

 

If they make their own PC chips, they'll do it to get exactly the right balance of tradeoffs for their computers and to include customer silicon that the competition doesn't have (the stuff in the iPhone X that makes the facial recognition much faster, for example, or the secure enclave for security, etc).

 

Apple would love to have a laptop with 20+ hours of battery life that its Intel-using competitors can't match along with FaceID to unlock and a secure enclave to store all biometrics data, for example. And since they control the OS, they can more tightly integrate the two (ie. have specialized cores that get used by the OS for very specific functions maybe -- perhaps an hardware X86 emulator optimized for the transition period so that old software can run on the new ARM OS).

 

And since Apple's A11 already competes with not-too-old Intel laptop chips while using a fraction of the power, there's nothing that tells me that they can't make competitive laptop and desktop chips if they try.

 

They've truly built a world-class silicon design team over the past decade.

 

Ben Thompson wrote today:

 

The A11 chip inside the iPhone X and iPhone 8, for example, outperforms the Intel Core i5 7267U processor in the base model MacBook Pro 13″ on the synthetic Geekbench benchmark in both single-core and multi-core operations; the Intel chip, though, costs $300 while IHS Markit estimated that Apple pays $27.50 for each A11 chip.

 

These prices are not at all comparable: $27.50 is the price Apple allegedly pays TSMC; it does not include the massive amounts of R&D the company spent on actually designing it. Given that nearly half of an iPhone’s costs go towards these sorts of intangibles (along with huge amounts of patent licensing fees), the real cost is much higher. At the same time, though, it certainly isn’t $300; Intel’s price also includes huge amounts of R&D and the capital costs to make its own chips, but a big premium come from Intel’s near-monopoly on PC CPUs.

 

So another argument would be that it's cheaper, and also leaves Apple more in control of its roadmaps (Intel has delayed and missed targets a lot recently, screwing up with Apple's release cycles).

[DTEJD1997]

Well that's not true. MacBook airs and pros use i7s which is a mid range xeon. Some use i5s - lower range xeons and probably not worth talking about here.

 

The fact is that one of the things that made macs so successful is that they've overpowered their systems. That's one of the main reasons why they were so stable and popular. They idea that x and y chip would be good enough for what the user needs is a radical departure from this strategy.

 

I don't doubt that Apple can design some chips to put in their macs. What eludes me is what they have to gain from it.

I am not going to argue this at length... but an i3, i5, i7 are NOT Xeons. 

 

Xeons frequently use different motherboards, frequently have different LGA architecture, Xeons don't have integrated graphics, Xeons can use ECC memory...Xeons can be used in multiprocessor configurations (ever see a dual processor i7 system?), and on and on.  Xeons are typically a LOT more expensive than i3's, i5's, i7's.  The highest end Xeon is about $12,500....contrast that with the highest end i7, which is about $800.

 

Xeons are used primarily in SERVER configurations...but sometimes are used in HIGH end workstations.  There are some laptops that use Xeons (Dell Precision series), but I have yet to see any Apple laptop using a Xeon processor.  PC's Xeon laptops are pretty rare & expensive.

 

How do I know this?  It is my primary business.  I use 4 different Macs and 1 PC.  I sell HUNDREDS of computers/servers every year.  Over the course of my career, I've sold TENS of thousands of computers, and hundreds of servers.