Will x86 be here "forever"?

[Arkaign]

It's as fast as the CPU cores in the next gen consoles. :awe:

What is a "full fledged x86 processors"? Atom?! Jaguar?! Pildedriver?! IB?! Haswell?!

So a future product will be better than a present ARM core? Wow... :$

What a luck for ARM that nobody is selling standalone CPUs. We talking here about SoCs. So the CPU design is only one part of the product.


Nobody will use Intel in the future because they do not want to be tied to one manufacture. That's the reason why ARM is winning this war: If you don't like your partner go to another one or design an own chip.

Jaguar is sort of pathetic, but in fact it's still faster than the processor in the Shield when comparing raw CPU performance. If you want to talk only about a single core, I guess you could consider them roughly equal in some ways? But that's a horrible target in terms of performance. 28nm is old garbage at this time for mobile consideration, and Jaguar looks like roadkill compared to full desktop x86 stuff even on the level of Phenom II, let alone Sandy Bridge from how many years ago? That's not a complete slam against Jaguar, it wasn't intended to be a competitor to even midrange processors. It's a budget, low performance, moderately efficient part. The best thing I can say about it is that it's far better than the previous Atoms, which most can agree were even worse.

SoC / CPU is semantics at best, everything we're talking about in the context of ultra mobile CPU performance is clearly talking about SoC, but in particular, the CPU portion of that device.

No, ARM has historically won this market because the big guys simply ignored it for many years. And people will definitely use Intel IF the performance is staggeringly better as it almost certainly will be, combined with power efficiency impossible with the low-end fabs used by the competition.

It's very similar to what's happening to AMD on the desktop. They don't have the R&D budget, nor access to competitive fabs in order to produce a processor that has the necessary raw performance or power/heat characteristics in any competitive manner on the midrange to high end markets. You can argue that they present a certain value in the entry level markets, largely due to superior iGPU (or the graphics portion of the so-called APU).

Now that Intel is all guns blazing on mobile, we're going to see mass destruction in that market for the next 5-8 years. It's not smart money to bet against Intel when their entire mindset and vast resources are utterly focused on a single target.

 

[sontin]

Jaguar is sort of pathetic, but in fact it's still faster than the processor in the Shield when comparing raw CPU performance.

Benchmarks? I only find some which are showing A15 and Jaguar on par per Core.

If you want to talk only about a single core, I guess you could consider them roughly equal in some ways?

Kabini has 4 cores, like Tegra4 or S800. For the same performance Kabini needs much more power.

SoC / CPU is semantics at best, everything we're talking about in the context of ultra mobile CPU performance is clearly talking about SoC, but in particular, the CPU portion of that device.

And that is the reason why x86 can't compete in the end. With only one real manufacture there is no choice. There is only Intel and what they will offer.

No, ARM has historically won this market because the big guys simply ignored it for many years. And people will definitely use Intel IF the performance is staggeringly better as it almost certainly will be, combined with power efficiency impossible with the low-end fabs used by the competition.

Sure. Intel will provide faster and cheaper Socs "combined with power efficiency" than any other ARM partner. :awe:

Now that Intel is all guns blazing on mobile, we're going to see mass destruction in that market for the next 5-8 years. It's not smart money to bet against Intel when their entire mindset and vast resources are utterly focused on a single target.

OEMs will not buy Intel. They are not stupid. Now it's their chance to get independent from the big brother. With Android/RT and ARM there is a huge market with so many difference companies that it will be suicide to go back to Intel.

 

[cmdrdredd]

OEMs will not buy Intel. They are not stupid. Now it's their chance to get independent from the big brother. With Android/RT and ARM there is a huge market with so many difference companies that it will be suicide to go back to Intel.

Actually suicide will be ignoring Intel and allowing a competitor to release a product using their chip and destroying everything else out there.

 

[sontin]

Who is buying Intel anyway?
Asus and Acer will still support ARM with their tablets. Both using Tegra 4 for the flagship product.
Samsung does not really care. Their "tablet" line is a throw away product right now.
HP builds a much better and cheaper product with Tegra 4 than with Clovertrail+.
Toshiba has no Atom tablet and is only selling i3 and i5.

All these OEMs know that they have the chance to bring Intel down to their level now.

 

[SiliconWars]

Actually suicide will be ignoring Intel and allowing a competitor to release a product using their chip and destroying everything else out there.

What is this intel chip that is destroying everything else out there?

If you're still holding out hope that Bay Trail is it you're going to be really disappointed. Just read the language coming from both camps and it's clear than Intel has better power draw but worse performance.

 

[sm625]

There is no reason for x86 to ever go away. The portion of a cpu die occupied by x86 specific transistors is very small and shrinking. New instructions will be added over time and the portion that is x86 will continue to shrink. I'm not sure why Intel hasnt tried this yet, but it is possible to have multiple front ends for the same core, ie x86, ARM, etc all sharing a common core, with each front end only consuming <5% of total cpu die. Remember that a gpu has absolutely nothing to do with x86, and neither do caches. And those are the two sections of the die that have expanded the most in the past 10 years.

 

[pelov]

It'll flourish for a long time in traditional server and workstation segments, where backwards compatibility and raw horsepower are paramount. I think on the consumer side it's already showed it's weaknesses, and it's got nothing to do with die space utilized or even power consumption but rather the appeal of products and pricing.

 

[Arkaign]

What is this intel chip that is destroying everything else out there?

If you're still holding out hope that Bay Trail is it you're going to be really disappointed. Just read the language coming from both camps and it's clear than Intel has better power draw but worse performance.

No, Bay Trail is basically irrelevant to what we're talking about. It's more of an experiment/stepping stone than anything else.

It's 14nm and 10nm that should have everyone scared to death. Nobody else will be able to keep up unless they can get the same level of unified investment going. We're going to be looking at massive leaps forward in every aspect : power usage, performance, integration (LTE onboard at 14nm forward), and it will allow for thinner products, less heat concerns, and this is exactly the kind of product that will drive premium sales of flagship devices : performance, efficiency, and miniaturization that is impossible with competition stuck on fab tech 18-36 months behind them.

We're only just beginning to see the results of Intel shifting gears fully to mobile.

 

[SiliconWars]

No, Bay Trail is basically irrelevant to what we're talking about. It's more of an experiment/stepping stone than anything else.

It's 14nm and 10nm that should have everyone scared to death. Nobody else will be able to keep up unless they can get the same level of unified investment going.

This has been getting said for years now. They are always just one or two steps away. The problem is the competition makes bigger strides.

We're going to be looking at massive leaps forward in every aspect : power usage, performance, integration (LTE onboard at 14nm forward), and it will allow for thinner products, less heat concerns, and this is exactly the kind of product that will drive premium sales of flagship devices : performance, efficiency, and miniaturization that is impossible with competition stuck on fab tech 18-36 months behind them.

Intel has always had this advantage and has never been able to put it to good use in mobile. The performance advantage is shrinking with each node and TSMC will be ready for volume production by early 2014.

We're only just beginning to see the results of Intel shifting gears fully to mobile.

No that's what Anand said about Clover Trail and see where that got them. The new Atom is what Intel will be stuck with for years. It can't beat 28nm ARM chips on 22nm so it won't beat 20nm ARM chips at 14nm.

 

[Arkaign]

This has been getting said for years now. They are always just one or two steps away. The problem is the competition makes bigger strides.

Intel has always had this advantage and has never been able to put it to good use in mobile. The performance advantage is shrinking with each node and TSMC will be ready for volume production by early 2014.

No that's what Anand said about Clover Trail and see where that got them. The new Atom is what Intel will be stuck with for years. It can't beat 28nm ARM chips on 22nm so it won't beat 20nm ARM chips at 14nm.

Huh? They've barely attempted anything resembling mobile in any meaningful way yet. Previous Atoms were clearly an afterthought. The 2013 product line is more of a test run than anything else.

The fab gap is going to grow, not shrink, unless gargantuan investments are made to match.

'Stuck with for years'? Are you high? Airmont at 14nm is probably what's going to be the hammer. Besides, on what metric do you claim that 22nm CT is inferior to 28nm ARM products? For such an early product, it seems to trade blows pretty well, it's main shortcoming in my mind is that it's missing the superior integration offered by Qualcomm's LTE SoC. Intel has plainly viewed it as merely a stepping stone though, while they get all of the various aspects in line for their ideal mobile vision.

You can dismiss Intel if you wish, but that's probably not the most logical conclusion when you're dealing with a corporation of that size and experience that has now shifted the bulk of it's entire emphasis into territory that directly covers mobile. I can understand the need for fanboys and haters with emotional reasons to hate Intel, but facts are facts, and things are about to get really really crazy in mobile.

Now Microsoft on the other hand, they really do seem to be in serious trouble. It will be a weird world seeing Intel-powered Android and iPhones/iPads while Microsoft begins to wind down into a smaller and smaller sphere of influence. Windows phones and tablets just don't get the numbers they need to be relevant.

 

[pelov]

Huh? They've barely attempted anything resembling mobile in any meaningful way yet. Previous Atoms were clearly an afterthought. The 2013 product line is more of a test run than anything else.

The fab gap is going to grow, not shrink, unless gargantuan investments are made to match.

'Stuck with for years'? Are you high? Airmont at 14nm is probably what's going to be the hammer. Besides, on what metric do you claim that 22nm CT is inferior to 28nm ARM products? For such an early product, it seems to trade blows pretty well, it's main shortcoming in my mind is that it's missing the superior integration offered by Qualcomm's LTE SoC. Intel has plainly viewed it as merely a stepping stone though, while they get all of the various aspects in line for their ideal mobile vision.

You can dismiss Intel if you wish, but that's probably not the most logical conclusion when you're dealing with a corporation of that size and experience that has now shifted the bulk of it's entire emphasis into territory that directly covers mobile. I can understand the need for fanboys and haters with emotional reasons to hate Intel, but facts are facts, and things are about to get really really crazy in mobile.

Now Microsoft on the other hand, they really do seem to be in serious trouble. It will be a weird world seeing Intel-powered Android and iPhones/iPads while Microsoft begins to wind down into a smaller and smaller sphere of influence. Windows phones and tablets just don't get the numbers they need to be relevant.

I was with you until that last sentence, where you essentially made the same criticisms of MS that people have of Intel: nowhere to be seen, lackluster products available, too expensive, and a history of being late.

I don't see the fab gap increasing, btw. In fact, it's likely to decrease in the coming years as expenses drastically increase and the 450mm/EUV bottleneck could put everyone on a level playing field.

 

[Arkaign]

I was with you until that last sentence, where you essentially made the same criticisms of MS that people have of Intel: nowhere to be seen, lackluster products available, too expensive, and a history of being late.

I don't see the fab gap increasing, btw. In fact, it's likely to decrease in the coming years as expenses drastically increase and the 450mm/EUV bottleneck could put everyone on a level playing field.

I understand your reaction, but hear me out :

With Intel v mobile : you will have Intel products appear seamlessly to the end user. iOS will work, Android will work, which is what the customer is used to. What they will get is higher capabilities, performance, and efficiency, but otherwise it's going to be a transparent movement.

With Microsoft v mobile : they really needed to do this earlier. Getting the mindshare of selling a Windows-based phone or tablet needed to be done in a meaningful way long, long, long ago, before the customer base became entirely used to Apple and Google platforms. In a way, it reminds me of the exact situation that alternatives to Windows faced once that juggernaut was underway : it didn't matter that the alternatives were often better in many ways, it mattered that everyone's software and experience was set in stone with using Windows.

With Microsoft's mobile products, and even to some degree Windows 8/8.1, Microsoft has tried to actually leave behind the legacy windows familiarity, which is necessary for a viable touchscreen experience on a phone or tablet, but at the same time also abandons that seamless jump. Being able to use Windows 3, 95, 98, ME, 2K, XP, Vista, 7 and then going to use a Windows Phone isn't even remotely similar.

The biggest proof is probably that we already have very very good Windows phones on the market, phones that arguably are better than many Android phones. But they have an incredibly hard time selling because the market has been sewn up by iOS/Android.

I could be totally wrong, and they will break that wall, but for now I have a hard time envisioning them breaking past niche marketshare in mobile.

At the same time, rumors of the desktop's demise are vastly premature (try doing real work on a tablet, even with a keyboard, hah), and server, well although *nix shares heaps of server marketspace, all of those smartphones and tablets relying on cloud/web assets need more servers to operate on. Microsoft will continue to do well in that area for sure.

I hope that clarifies what I meant. It's my perspective, and it could turn out to be utterly wrong. Someday a Windows phone could outsell the iPhone, or even be a sales competitor, but it doesn't seem likely to me at this point.

 

[SiliconWars]

The fab gap is going to grow, not shrink, unless gargantuan investments are made to match.

How do you figure? TSMC has working 20nm silicon and that's a fact. It will be ramping in Q1 2014, fact. Each new node gives less performance improvements. The gap is *shrinking*.

'Stuck with for years'? Are you high? Airmont at 14nm is probably what's going to be the hammer.

Probably huh. Do you know anything about chip design? Airmont is a shrink of 22nm Atom, it'll have ~25% improvement in power or performance. This is not going to "hammer" anything.

Besides, on what metric do you claim that 22nm CT is inferior to 28nm ARM products? For such an early product, it seems to trade blows pretty well, it's main shortcoming in my mind is that it's missing the superior integration offered by Qualcomm's LTE SoC. Intel has plainly viewed it as merely a stepping stone though, while they get all of the various aspects in line for their ideal mobile vision.

Early product? This is it for Intel on mobile. What changes did you see from Nehalem to Haswell? Mostly all on graphics, which is another place Intel is going to be sadly lacking. If Airmont does anything it'll be improved graphics that just get closer to Qualcomm's.

And like you said, it's still lacks much of Qualcomm's integration. Node advances long since ceased to give miracle performance increases that you mistakenly believe they still give.

You can dismiss Intel if you wish, but that's probably not the most logical conclusion when you're dealing with a corporation of that size and experience that has now shifted the bulk of it's entire emphasis into territory that directly covers mobile. I can understand the need for fanboys and haters with emotional reasons to hate Intel, but facts are facts, and things are about to get really really crazy in mobile.

What "facts" are these? Spell them out so we can see where we're all clearly getting it wrong because of being "haters with emotional reasons".

 

[pelov]

At the same time, rumors of the desktop's demise are vastly premature (try doing real work on a tablet, even with a keyboard, hah), and server, well although *nix shares heaps of server marketspace, all of those smartphones and tablets relying on cloud/web assets need more servers to operate on. Microsoft will continue to do well in that area for sure.

I hope that clarifies what I meant. It's my perspective, and it could turn out to be utterly wrong. Someday a Windows phone could outsell the iPhone, or even be a sales competitor, but it doesn't seem likely to me at this point.

I think you're underestimating the ties that MS and Intel have, at least with regards to keeping the status quo.

Both corporations relied on high margins for their products because they were the only game in town (AMD is a competitor, but they've been lagging for years). While PCs were booming, they made money hand over fist.

In the current landscape, neither Intel nor MS can get by charging the same prices they did a mere 2 years ago. It just won't fly. Asus, Acer, Toshiba, HP, Dell, LG, Samsung, etc., don't care about x86 compatibility on an Android tablet. There's just nothing Intel has to offer them that's special and worth the added tax. The OEMs have alternatives that are cheaper and allow them more flexibility with respect to BoM as well as profit margins. In short, Intel holds little-to-no influence in this landscape and can't command higher prices.

Likewise, the OEMs and device makers don't care for Win8 and whatever MS has to offer, especially not at that inflated price tag. Win8 isn't selling in the mobile world and 8.1 looks to be more of the same. OEMs and users alike are unsure whether MS is just going to drop it like they did WP7 and the Zune or buy into it and hope for long term support.

Now that tax I'm referring to is incredibly important for Intel and MS because that's how they make their living. They're used to OEMs lining up to pay $40-$100 for a Windows+Office license and $100-$300 for costly CPUs, then operating on very slim margins. Due to a high volume and a lack of competitors, it's viable.

But what happens when these devices stop selling like they did before? Do you really think Intel can pay for those costly fabs that are going to chew away at their profits each quarter by selling Atoms and competing with ARM? Do you really think that OEMs are going to say, "Yes, we'll pay you the extra $20+ because you're Intel" ? They now have the option of going elsewhere.

What the mobile race has showed, above all else, is that it's not an x86/Win32 world anymore. And that's a very scary scenario for both Intel and MS. In the traditional PC space, they'll do fine. They have nobody to challenge them just yet, but consumers are quickly losing interest and don't see a reason to pay $500-750 for a new PC when they can have a tablet for half the cost that can complement their now aging PC.

 

[Nothingness]

Yeah Intel never wanted to get into smartphone until now: http://www.ft.com/intl/cms/s/0/2182d7a8-05cf-11dd-a9e0-0000779fd2ac.html#axzz2ZnbR9ddO

Well they said they wanted, but I guess Otellini was not serious he was just kidding, but now Intel is serious!

 

[NTMBK]

Yeah Intel never wanted to get into smartphone until now: http://www.ft.com/intl/cms/s/0/2182d7a8-05cf-11dd-a9e0-0000779fd2ac.html#axzz2ZnbR9ddO

Well they said they wanted, but I guess Otellini was not serious he was just kidding, but now Intel is serious!

If you look at what they actual did to enter the smartphone market, yeah, you can argue they weren't serious

 

[jhu]

Age of universe: 14 billion years
Years until the sun engulfs the Earth: 4 billion years
Number of years x86 in existence: 35 years

Given this data, yes x86 will last forever.

 

[StrangerGuy]

I think you're underestimating the ties that MS and Intel have, at least with regards to keeping the status quo.

Both corporations relied on high margins for their products because they were the only game in town (AMD is a competitor, but they've been lagging for years). While PCs were booming, they made money hand over fist.

In the current landscape, neither Intel nor MS can get by charging the same prices they did a mere 2 years ago. It just won't fly. Asus, Acer, Toshiba, HP, Dell, LG, Samsung, etc., don't care about x86 compatibility on an Android tablet. There's just nothing Intel has to offer them that's special and worth the added tax. The OEMs have alternatives that are cheaper and allow them more flexibility with respect to BoM as well as profit margins. In short, Intel holds little-to-no influence in this landscape and can't command higher prices.

Likewise, the OEMs and device makers don't care for Win8 and whatever MS has to offer, especially not at that inflated price tag. Win8 isn't selling in the mobile world and 8.1 looks to be more of the same. OEMs and users alike are unsure whether MS is just going to drop it like they did WP7 and the Zune or buy into it and hope for long term support.

Now that tax I'm referring to is incredibly important for Intel and MS because that's how they make their living. They're used to OEMs lining up to pay $40-$100 for a Windows+Office license and $100-$300 for costly CPUs, then operating on very slim margins. Due to a high volume and a lack of competitors, it's viable.

But what happens when these devices stop selling like they did before? Do you really think Intel can pay for those costly fabs that are going to chew away at their profits each quarter by selling Atoms and competing with ARM? Do you really think that OEMs are going to say, "Yes, we'll pay you the extra $20+ because you're Intel" ? They now have the option of going elsewhere.

What the mobile race has showed, above all else, is that it's not an x86/Win32 world anymore. And that's a very scary scenario for both Intel and MS. In the traditional PC space, they'll do fine. They have nobody to challenge them just yet, but consumers are quickly losing interest and don't see a reason to pay $500-750 for a new PC when they can have a tablet for half the cost that can complement their now aging PC.

"WTF is business strategy and economics? We will win because we have 14nm and you don't, lawls."

Their traditional PC market is going to experience a vicious cycle, with continued dropping shipments the economics of scale that once allowed dirt cheap PCs won't be there anymore. Suppliers are going to cut back production and component prices will increase, and OEMs either to use cheaper CPUs or raise ASPs. Both of which are unhealthy to Intel.

BTW The next time someone here makes AMD vs Intel in the past is analogous to ARM vs Intel only proves they know shit about the current situation, period.

 

[ultimatebob]

Then why did Intel cut earnings and sales forecasts, along with capital spending? The trend is very very clear, the market is moving away from x86.

Away from x86 for tablets, anyway... not that it was a huge market for them to begin with. Most cell phones never used x86 processors to begin with, and very few manufacturers are making ARM laptops/desktops/servers.

 

[SiliconWars]

Away from x86 for tablets, anyway... not that it was a huge market for them to begin with. Most cell phones never used x86 processors to begin with, and very few manufacturers are making ARM laptops/desktops/servers.

That's because ARM hasn't been designing chips for those markets until now, being somewhat busy gobbling up the vast majority of the mobile market while the x86 guys slept.

 

[pyjujiop]

Nobody will use Intel in the future because they do not want to be tied to one manufacture. That's the reason why ARM is winning this war: If you don't like your partner go to another one or design an own chip.

If Intel offers superior computing power in the same power envelope as ARM, which is going to happen, then software developers will find ways to use that power in smartphones and tablets, and ARM will either have to match it or they will lose.

The fact Intel will be the only x86 vendor in that market space isn't going to cut much ice. Intel can only be so aggressive in the desktop/laptop space because they really need AMD to survive. As long as AMD exists, they can successfully argue they're not a monopoly. However, in the smartphone/tablet space, they can compete any way they want to, because Samsung, Nvidia, Qualcomm, among others, are already established in that market. Intel will simply offer a superior chip at a better price and challenge everyone else to match it.

The only reason ARM is "winning" right now is that they're the only game in town. In a few years, they're going to get pushed to the margins just like AMD in the desktop PC space.

 

[R0H1T]

Age of universe: 14 billion years
Years until the sun engulfs the Earth: 4 billion years
Number of years x86 in existence: 35 years

Given this data, yes x86 will last forever.

Over extrapolation^infinity that is if it makes any sense at all

 

[psyq321]

Up to this moment I can only see that Intel managed to reach and even exceed in some points ARM when it comes to efficiency in the laptop/compact laptop segment with Haswell - see latest Anand's review on Haswell ULT vs. ARM in terms of power consumption.

I'd also like to see ARM achieving the same in terms of entering the server market. But at this stage that is marketing only. Haswell ULT, on the other hand, is real - and it looks to me as a sneak-preview of things to come with Broadwell and 14 nm.

Then we will have some real competition at least, which should be very good for the consumer (both private and also institutional)