Intel Urged to Take 'Immediate Action' Amid Threats From Apple Silicon and AMD

[thecombatwombat]

I'm surprised by how dominant XP was and 10 is. I'm also surprised that 7 wasn't bigger.

I would have guessed that 7 was the version of Windows that did best, followed by XP then 10. But it's actually XP > 10 > 7. I guess something important in XP's dominance was the lack of competition. Linux hadn't emerged yet, and Apple was stuck with PPC and the iPod hadn't started to get everyone to pay attention to Apple.

I'm kind of skeptical of that data. Linux at 6%?

That aside though, I think the XP numbers have more to do with Microsoft than anything else. XP was the successor to Me, Me was basically a disaster. Those charts don't even bother showing it. I assume it's just lumpled into the Windows 98 numbers.

XP got very unhappy Windows users to move in droves. It was a massive leap forward. Windows users were never that unhappy again, and so MS never migrated their own users as quickly again.

 

[dogslobber]

We live in a Windows x86 world and it's not going anywhere.

 

[Falhófnir]

The issue with Intel is indeed the chip design - they just don't scale down which is why they remain "stuck" on 14nm for their most powerful chips and they are having such a problem getting to 10nm, much less anything smaller.

Just as Core replaced Netburst when the latter hit the wall, so now has Core hit the wall and Intel needs to find an entirely new x86 architecture to replace it - one that can scale down to sub-10nm.

I think that was Lakefield... and that's a disaster. They let Microsoft down big time, killing the Surface Neo and Windows 10X in the process. I think that has pretty much pushed Microsoft towards Arm, especially now with Apple Silicon they're sure to be spooked about Apple being in such a commanding position when they have no answer.

 

[axboi87]

Precisely. Compatibility is everything for some. Remember Commodore Amiga? It was a decade ahead of IBM + MS DOS. It was so nice, so user-friendly! And yet it failed, because there was no software for it, and it wasn't able to run existing business applications at a reasonable speed.

The market can be segmented like this:
- Consumers: They're very easy to transition, they only need a browser, a word processor, and whatever is hot at the moment. Gaming is important, though.
- Computer professionals: Big brand software publishers can do the transition very quickly. Everything is subscription-based anyway, few people are stuck with old software.
- Corporate: They mostly need word processing, accounting, communication, but also need the occasional obscure custom solution that must run super reliably. Emulation can take care of the legacy software. On the other hand, they are not going to transition all machines at the same time, but over a long period of 5-7 years. They need to support both old and new hardware equally well for a long period of time.
- Manufacturing, scientific: This is very tricky, they rely on $200,000+ special equipment, metalworking, medical devices, truly one-of-a-kind peripherals, such as mass spectrometers, CNC machines. Their investment is not for 10 but 25 years. They also run seriously expensive CAD software, industrial flow control, custom controller software. Consider power plants, Xray, irradiation, drug research. There's no transitioning of these applications for a very long time.
- Data warehouse: It tends to be always the latest and greatest (continuous integration and continuous delivery). Cloud software tends to be .NET or Java, which is machine-independent and platform-independent. Often the hardware itself is a service in the cloud, hosted and maintained by Amazon or Microsoft.

The corporate end is changing rapidly. I work in IT and absolutely everything we use, and 90% of what our customers use (with that 10% going away fast) is just some web app. Accounting, Single-Sign-On, File Sharing, Operations, CRM, Document Management, and even specialty apps like Property Management software (commercial real estate). With Enterprise SaaS and even IT infrastructure (SaaS Single Sign On replacing Active Directory for example), endpoint devices can now be totally OS agnostic. Due to this, windows marketshare continues to fall and things like Chromebook and Macs are becoming big in the Enterprise. Hell even my CPA is an all Mac shop

 

[redcaptrickster]

I'm surprised by how dominant XP was and 10 is. I'm also surprised that 7 wasn't bigger.

I'm not. The graph also explains the popularity of the Tiger King, K-Pop, and CEOs who let R2D2 name their children.

 

[Analog Kid]

They can buy AMD to begin with.

That would certainly bring a more speedy end to the x86 market, if that's what you're after...

And why some of the newer Ryzen designs are like hockey-pucks. I think the issue backward compatibility, intel can't start over like Apple did. Intel can't say, no more 32/16 bit apps. (I still decided a 10900K won out against a Ryzen in my own new build)

I wonder how true that is...

If Intel announced a new x86 architecture that removed all the ancient cruft and focused on keeping compatibility with what's most useful to todays applications, would it succeed? What if they trapped unimplemented instructions and emulated them in software-- old code would still run rather than crash, but the less common instructions would be slow encouraging compiler writers to stop using them.

Or provide a Rosetta2 like software preprocessor to convert old applications to the updated instruction set. Crufty x86 to less crufty x86.

Intel could still make fully backwards compatible CPUs and sell them at a very high margin for those applications that rely on them because presumably those customers would rather pay the premium than rebuild their code. Right now, everyone is paying the price of supporting those legacy apps. Take the burden off of the new customers and put it on the older customers where it belongs.

Basically what @amartinez1660 just suggested:

Maybe they could have two models? The backwards compatible one and the fresh, new, clean slate from here on? That would require TONS of cooperation for things like Apple does with Universal executables and whatnot on the PC side though.
And, also, couldn’t they just continue selling the current models for those that want to continue using legacy software? does legacy/non-updated decade old 2010 software really needs to be supported by 2020 CPUs? I mean, it already does and will continue to run it forever, so I guess they could sell today’s 2020 models for years on end which will chew on that legacy software.

I probably could have left it at that...

 

[Homme]

The problem is Intel's foundries are barely at 10nm. Apple would need to invest billions, if not tens of billions, to get them to sub-10nm and Apple would be dependent on hoping Intel's fab engineers could do it.

Much better to spend a few scores of millions to help TSMC get their new processes into production by agreeing to be "first mover" and buying chips while TSMC are ramping up production and yields are low and then getting exclusive access to that process for a time.

exactly. What’s the point of Apple buying Intel for that sole purpose when TSMC would be a better fit

 

[Yebubbleman]

It's my understanding that:

(a) Xeon still very much dominates in the server market over Epyc/Opteron and that there's no imminent danger of that changing drastically anytime soon

(b) Non-Xeon consumer/prosumer x86 CPUs are only one of MANY different businesses and business units that Intel has (for instance, it has a fairly decent flash storage business last I checked; it also owns McAfee which isn't a small fish by any means)

(c) Given (b), the fact that people are freaking out about Intel going downhill based on only one of its many business units (albeit, its most popular one) seems a bit rash.

 

[ipponrg]

These kinds of graphs always lump in Surface and similar products as 'laptops' but never do the same in Apple's favour with the iPad line. They should include the iPad.

The Surface does run a full desktop /laptop OS unlike the iPad though.

 

[I7guy]

That would certainly bring a more speedy end to the x86 market, if that's what you're after...



I wonder how true that is...

If Intel announced a new x86 architecture that removed all the ancient cruft and focused on keeping compatibility with what's most useful to todays applications, would it succeed? What if they trapped unimplemented instructions and emulated them in software-- old code would still run rather than crash, but the less common instructions would be slow encouraging compiler writers to stop using them.

Or provide a Rosetta2 like software preprocessor to convert old applications to the updated instruction set. Crufty x86 to less crufty x86.

Intel could still make fully backwards compatible CPUs and sell them at a very high margin for those applications that rely on them because presumably those customers would rather pay the premium than rebuild their code. Right now, everyone is paying the price of supporting those legacy apps. Take the burden off of the new customers and put it on the older customers where it belongs.

Theoretically I suppose anything is possible, however, maybe it's not feasible. Intel x86/x64 architecture runs deep and wide and companies know there is an upgrade path for full compatibility. And maybe that's intel's biggest asset.

 

[Analog Kid]

In any case, if what the article says is one of the biggest reason, employees fleeing because of mismanagement, then it’s doubly sad. We probably all have been on a mismanaged company: things run circles, asking for a pen takes a week to be resolved by procurement, meetings are created to discuss about the next meetings, people fighting for the credits of others, impossible deadlines still treated as if on target in an insane state of denial until it’s too late, etc... goddamn, writing that made me stress.

Intel has been badly managed for decades, from a personnel management perspective. People put up with it because it was freaking Intel and you’ll put up with a lot to work for the best.

When they’re no longer the best, however...

 

[SBlue1]

I'm pesonally just happy my Mac doesn't run as hot as a BBQ anymore.



Edit: The gracious @Freeangel1 pointed out I made a spelling mistake... I shall leave it as-is in honor of you

I am so surprised and kinda disappointed my M1 MacBook Air is so cold. I was used to having the old MacBook Pro keeping my knees warm.

 

[Falhófnir]

It's my understanding that:

(a) Xeon still very much dominates in the server market over Epyc/Opteron and that there's no imminent danger of that changing drastically anytime soon

(b) Non-Xeon consumer/prosumer x86 CPUs are only one of MANY different businesses and business units that Intel has (for instance, it has a fairly decent flash storage business last I checked; it also owns McAfee which isn't a small fish by any means)

(c) Given (b), the fact that people are freaking out about Intel going downhill based on only one of its many business units (albeit, its most popular one) seems a bit rash.

It does risk sending the company's chip business into a tailspin though; less money from selling chips means less money to invest in improving their chips, means it's even harder to catch up, leading to less money from selling chips... the company overall probably won't disappear, but Nokia is still around too, just not in the main business that made them famous (excl. HMD licensing the name).

 

[Analog Kid]

Theoretically I suppose anything is possible, however, maybe it's not feasible. Intel x86/x64 architecture runs deep and wide and companies know there is an upgrade path for full compatibility. And maybe that's intel's biggest asset.

I think it has been their biggest asset in the past— it’s why that hot mess of an architecture is still the top selling PC processor.

I also think Intel believes it’s their biggest asset now. I’m less sure if that’s actually true, or if Intel just fell victim to their own marketing.

The big question is whether that asset suddenly became a massive liability. It may be too early to tell, but the performance of AS hints that this might be true. The fact that AS can run translated code nearly as fast or faster than the original x86 on a native processor suggests this might be true.

How much 16bit code are you still running on your machine? How many memory models are you utilizing (because you need them, not just because Intel forces boot into legacy modes before you can switch to a modern one.). Certainly Mac didn’t need any of that, but Mac was burdened with it. Are we all running slower and burning more power because a couple banks and government agencies are trying to keep their 16bit COBOL apps alive?

Ordinarily I’d say Intel knows their business better than I can guess it, but I’m starting to wonder if backwards compatibility went from customer need to religion at some point and stopped being questioned.

 

[vmistery]

Well AMD spin off Global foundries so I’m sure Intel could spin off theirs. But we have to remember Intel is still making money unlike AMD in it’s bad years so I’d be surprised if they don’t pull something out over the next few years. Those ringing the death bell for AMD too are being premature, they’re ARM licensees already and I’d be shocked if they didn’t have an ARM design in the pipeline. For now though, I’m going g to enjoy watching Intel take a beating

 

[Analog Kid]

It's my understanding that:

(a) Xeon still very much dominates in the server market over Epyc/Opteron and that there's no imminent danger of that changing drastically anytime soon

(b) Non-Xeon consumer/prosumer x86 CPUs are only one of MANY different businesses and business units that Intel has (for instance, it has a fairly decent flash storage business last I checked; it also owns McAfee which isn't a small fish by any means)

(c) Given (b), the fact that people are freaking out about Intel going downhill based on only one of its many business units (albeit, its most popular one) seems a bit rash.

In 2019 the storage business reported a $1.2B operating loss. It’s there largely to help defray fixed costs on fabs, as far as I can tell. If you look at their businesses, just about all of their money is from processors or from devices that support their processors.

As far as not leaving Xeon any time soon, it’s not good when your customers are only buying your products until the consumer technology wave reaches their business too...

 

[Constable Odo]

Good riddance to x86 processors in Apple computers. Now Apple can finally make their products thin and not have thermal throttling issues. I'd still like to see Apple put large batteries in their laptops, so I'd gladly accept the additional weight and thicker case. Apple is definitely going to have the thinnest desktops available and they're going to be more powerful with Apple Silicon than any consumer-based Intel processor. Yay!

 

[Piggie]

I have zero sympathy for Intel, they sat on their butts for far far too many years.
However I don't want them to go.
Less companies competing with each other is never a good thing for consumers, and things always change.
Who knows, in another 20 years Apple may be so lazy and Intel have by them new things under totally new management and new teams.
Nothing ever last forever or stays the same.

 

[bodonnell202]

Intel’s arrogance will ultimately be their undoing.

 

[bodonnell202]

They've been resting on their laurels since about 2006 when core 2 dominated the CPU world. People love to accuse apple with little to no innovation but it really is Intel that is guilty of that.

Side note - I picked up an M1 MBP because my 2016 MBP is literally falling apart. This thing is incredible. Graphics are better than the discrete, no fans running all the time, it's amazing. Can't wait for the 16" version

Sandy Bridge (2nd gen Core) was a pretty leap, they’ve really just been iterating ever since.

 

[yurc]

Might be bad karma for serving us 4-core mainstream chips for almost decade with criminal price.

 

[Analog Kid]

Well AMD spin off Global foundries so I’m sure Intel could spin off theirs. But we have to remember Intel is still making money unlike AMD in it’s bad years so I’d be surprised if they don’t pull something out over the next few years. Those ringing the death bell for AMD too are being premature, they’re ARM licensees already and I’d be shocked if they didn’t have an ARM design in the pipeline. For now though, I’m going g to enjoy watching Intel take a beating

AMD has shown themselves to be much more agile. I would not be surprised if they led the PC market into a new age on a new architecture. I’m not sure Intel can be that company. The Itanium experience doesn’t bode well.

Unless x86 has a bunch of hidden potential we haven’t seen, I can’t imagine Intel will be the company to lead the next wave unless they buy a company who can and essentially transfer the name to a new company...

 

[cmaier]

so many intel haters
just wait when the apple arm does not satisfy digital content creators
you guys have a lot hope in apple that these m1 processors and all their apple proprietary GPU and other BS will meet your needs

I personally LOVE intel and building my own PC.
Every iMac I have ever owned has had a bad video card making it useless for graphic design or video editing.

“digital content creators” will do fine with apple silicon. Much better than x86.

 

[The Cappy]

Its too late

Hopefully not. Whether I am personally reliant on intel or not, I don't want intel to fail. I like a world where AMD is succeeding and intel is succeeding and both compete vigorously.

 

[cmaier]

Hopefully not. Whether I am personally reliant on intel or not, I don't want intel to fail. I like a world where AMD is succeeding and intel is succeeding and both compete vigorously.

When has intel ever competed fairly against AMD?