ambidextrous computing; AMD project skyla..skybridge!

[shady28]

Problem is they are not going anywhere. Custom ARM development cost is already reaching x86 fast. Yet the performance is nowhere near and they havent expanded into x86 segments. And x86 is hammering down into ARM segments instead.

And if you look on growth, you will notice most of it comes from ARM companies canibalizing one another.

I think you're (clearly) wrong about that.

ARM has invaded Intel's major cash cow. Mobile and Tablet systems are punching big gaping holes into the PC/Desktop ecosystem.

What you seem to be saying is that ARM isn't ready to invade Intel's last main stronghold, the server market.

But think about where ARM processors were 10 years ago.. 15 years ago.. even 5 years ago... 5 years ago would have been when a few people had a slight inkling of what might be about to happen.

If it were just Intel vs ARM then of course, Intel would probably "win" just like Intel vs AMD. But it isn't that simple. ARM licenses a core tech, and others customize and extend it.

This gives the licensees the ability to differentiate themselves. Apple's A7 is one example, but Motorola's X8 may be an even better one because it isn't a drastic redesign - it's just innovative. The X8 uses ARM licensing via Qualcomm SnapDragon Pro, but adds processors to allow their phones to do voice recognition while in standby mode, detect motion which is used to activate the camera, or bring up an alert without unlocking the phone when it detects its been moved, to improve its voice recognition, etc etc.

Point being that ARMs licensing gives huge flexibility to a licensee to differentiate themselves. That's key. You can't do that with Intel. You can just... do what everyone else is doing.

So it isn't Intel vs ARM, it's Intel vs ARM, QualComm, Apple, MediaTek, RockChip, Motorola, Nvidia, Samsung, TSMC, GloFlo, etc.

 

[jdubs03]

Very interesting time, the ARM whole model has a uniqueness to it; the amount of cooperation between the licensees is impressive; same thing with the openPOWER alliance.

Intel has a tough challenge ahead, and its dependent on IA doing a great job (Broadwell/CT->Skylake/Broxton...of which I think they can), and then switching to a iii-v/450mm manufacturing process before anyone else to have an material/gate width advantage along with larger wafers. but it's also dependent on Microsoft and high praise for windows 9 (same with AMD).

Don't forget AMD, they stand to benefit a lot from the move to 16nm. Despite their losses, poor management, and attrition, they still have put out some quality designs; though 6 months after Intel's release. If they can get 16nm to in early 2016 (with both arm/x86), I think they will have a fighting chance.

I think the future war between intel and ARM will be won in the next 3 years. Bay-Trail was a nice introduction to 22nm and a proper Atom to challenge ARM, cherry trail (superior GPU, CPU=?) will take on the new cortex-a57's (some at 20nm), and broxton will take on the 16FF+ iterations. in terms of processor design for low-power, right now ARM looks like they have a shot, but x86 isn't going anywhere.

Be mindful too that Krzanich said the 2/1 performance ratio will still hold between Core and Atom, and now with Broadwell-Y scaling down to 4.5W TDP (end of year), and Bay Trail breaking the x86-power hungry myth (last year-this year), it makes sense that Intel will be focusing adding performance more so than lowering power consumption.

 

[ShintaiDK]

I think you're (clearly) wrong about that.

ARM has invaded Intel's major cash cow. Mobile and Tablet systems are punching big gaping holes into the PC/Desktop ecosystem.

What you seem to be saying is that ARM isn't ready to invade Intel's last main stronghold, the server market.

But think about where ARM processors were 10 years ago.. 15 years ago.. even 5 years ago... 5 years ago would have been when a few people had a slight inkling of what might be about to happen.

If it were just Intel vs ARM then of course, Intel would probably "win" just like Intel vs AMD. But it isn't that simple. ARM licenses a core tech, and others customize and extend it.

This gives the licensees the ability to differentiate themselves. Apple's A7 is one example, but Motorola's X8 may be an even better one because it isn't a drastic redesign - it's just innovative. The X8 uses ARM licensing via Qualcomm SnapDragon Pro, but adds processors to allow their phones to do voice recognition while in standby mode, detect motion which is used to activate the camera, or bring up an alert without unlocking the phone when it detects its been moved, to improve its voice recognition, etc etc.

Point being that ARMs licensing gives huge flexibility to a licensee to differentiate themselves. That's key. You can't do that with Intel. You can just... do what everyone else is doing.

So it isn't Intel vs ARM, it's Intel vs ARM, QualComm, Apple, MediaTek, RockChip, Motorola, Nvidia, Samsung, TSMC, GloFlo, etc.

Lets summon up. We got 10s of millions of x86 phones and x86 tablets. How many ARM desktops, servers and laptops do we have?

 

[Idontcare]

I think you're (clearly) wrong about that.

ARM has invaded Intel's major cash cow. Mobile and Tablet systems are punching big gaping holes into the PC/Desktop ecosystem.

What you seem to be saying is that ARM isn't ready to invade Intel's last main stronghold, the server market.

But think about where ARM processors were 10 years ago.. 15 years ago.. even 5 years ago... 5 years ago would have been when a few people had a slight inkling of what might be about to happen.

If it were just Intel vs ARM then of course, Intel would probably "win" just like Intel vs AMD. But it isn't that simple. ARM licenses a core tech, and others customize and extend it.

This gives the licensees the ability to differentiate themselves. Apple's A7 is one example, but Motorola's X8 may be an even better one because it isn't a drastic redesign - it's just innovative. The X8 uses ARM licensing via Qualcomm SnapDragon Pro, but adds processors to allow their phones to do voice recognition while in standby mode, detect motion which is used to activate the camera, or bring up an alert without unlocking the phone when it detects its been moved, to improve its voice recognition, etc etc.

Point being that ARMs licensing gives huge flexibility to a licensee to differentiate themselves. That's key. You can't do that with Intel. You can just... do what everyone else is doing.

So it isn't Intel vs ARM, it's Intel vs ARM, QualComm, Apple, MediaTek, RockChip, Motorola, Nvidia, Samsung, TSMC, GloFlo, etc.

I feel compelled to respond with a simple "wow" type of awestruck compliment as your post so succinctly captures what is different about the present situation, and in all the right ways :thumbsup:

Intel is limiting their process node advantage to the innovations which their IC design teams can come up. That is only a win if your process folks AND your design folks are bar-none the world's best in every way possible...and the marketing/business/sales folks must be on top of their game as well.

But ARM is doing something that is very smart, leveraging diversification across IC design teams and companies to determine the path of least resistance in terms of defining for them what the market wants, and where next year's innovation is going to come from.

Intel must knock this out of the park in order to survive, whereas ARM merely needs just one of their multitude of customers to come up with the next iPhone and they have themselves at least another decade of revenue secured.

 

[witeken]

I don't think there's anything wrong with Intel's way of doing it. Intel still has the advantage of being an IDM, which apparently also allows them to do certain things fabless companies like Qualcomm can't:

There are advantages to having fabs. You'll see many things, especially with Broadwell, that you cannot do without owning a fab.

Also, I think your premise that Intel doesn't have a top-notch design team is questionable. Certainly one that fails to use their process advantage. Your theory of "path of least resistance" sounds nice, but in reality all R&D budget for designing SoCs is much bigger because Apple, AMD, Samsung, Nvidia, Qualcomm,... all do about the same, instead of a company like Intel that spends its money 1 time to create 1 good x86 SoC instead of a multitude of subpar designs.

 

[Vesku]

The way I read it, IDC isn't saying that Intel doesn't have a top notch staff just that they pretty much HAVE to deliver stellar products every tick of the product cycle since it is the sale of Intel products that keep the funding flowing. Where as with the way ARM licenses they just need a few of many R&D teams among their licensees to deliver some stellar products. Both Intel and AMD have made noises about becoming a bit more like ARM, Intel has offered to integrate IP onto their x86 designs and AMD has offered to make custom ICs ala console chips, in an attempt to gain some of the diversity of effort ARM leverages.

 

[jdubs03]

imagine an x-scale/a57 processor at 14nm. intel should've retained their arm license.

 

[Third_Eye]

Lets summon up. We got 10s of millions of x86 phones and x86 tablets. How many ARM desktops, servers and laptops do we have?

Yes, 10s of millions of x86 phones and tablets where in Intel is losing money in the multi-billion dollar range last 2 years every and the near future



As a note, Intel is able to subsidize these gargantuan (mobile)losses purely bcos of the insane profits provided by the x86-x64 PC+Laptop+Data center CPUs as well as continue the huge lead in manufacturing. All the various ARM vendors need to do is to attack piece by piece the profit generating sectors.
AMD + Applied Micro attack the server aspect of it. QCOM(already successful) and MediaTek the phone section. AllWinner, Rockchip, Mediatek the cheap tablets, Apple, Samsung the higher tablet section. That way Intel would really "compete".
Intel cannot attack ARM. It can win over 1 vendor, the other vendor may cause problems. Same thing like Microsoft and Linux whacamole

 

[simboss]

Your theory of "path of least resistance" sounds nice, but in reality all R&D budget for designing SoCs is much bigger because Apple, AMD, Samsung, Nvidia, Qualcomm,... all do about the same, instead of a company like Intel that spends its money 1 time to create 1 good x86 SoC instead of a multitude of subpar designs.

This is based one the assumption that there is such thing as a "good" SoC and a "bad" SoC, and that each player is going to aim for the same target.

Where there is not added value to design, you can simply license a well know piece of IP, so you end up sharing the cost for this, be it the CPU, the GPU or any other piece of the SoC.

Intel is obviously able to get a good "SoC" in some metrics, but they can't compete in all of them at the same time (performance, power, cost, integration, differentiation, speed of execution, ...) and the pain to switch to x86 outweighs the benefits for most OEMs for now.

 

[CHADBOGA]

imagine an x-scale/a57 processor at 14nm. intel should've retained their arm license.

They have.

They just have no plans to use it.

 

[witeken]

Yes, 10s of millions of x86 phones and tablets where in Intel is losing money in the multi-billion dollar range last 2 years every and the near future



As a note, Intel is able to subsidize these gargantuan (mobile)losses purely bcos of the insane profits provided by the x86-x64 PC+Laptop+Data center CPUs as well as continue the huge lead in manufacturing. All the various ARM vendors need to do is to attack piece by piece the profit generating sectors.
AMD + Applied Micro attack the server aspect of it. QCOM(already successful) and MediaTek the phone section. AllWinner, Rockchip, Mediatek the cheap tablets, Apple, Samsung the higher tablet section. That way Intel would really "compete".
Intel cannot attack ARM. It can win over 1 vendor, the other vendor may cause problems. Same thing like Microsoft and Linux whacamole

Your failed premise is that ARM is a collaborative effort to kill Intel. It isn't. Every company that uses ARM's ISA does what it thinks will gain them most money. If Samsung thinks they can better stop making SoCs because the RoI isn't high enough anymore or they can't compete with their off-the-shelf ARM IP, they'll quit making SoCs, because at the end of the day, what they care about is the amount of tablets and phones sold. Same goes for other companies and the markets they want to compete in. Every ARM SoC vendor is also a competitor of the other ARM users.

Intel can surely attack ARM, but ARM won't go away because ARM also serves markets that could never generate a satisfying amount of money for Intel. Phones and tablets do, and therefor they'll join Qualcomm, Nvidia, Samsung and MediaTek in this space. Today companies use Qualcomm because they have the fastest SoCs, but when Intel will have those, the SoC supplier will be changed in disfavor of Qualcomm.

 

[jdubs03]

Your failed premise is that ARM is a collaborative effort to kill Intel. It isn't. Every company that uses ARM's ISA does what it thinks will gain them most money. If Samsung thinks they can better stop making SoCs because the RoI isn't high enough anymore or they can't compete with their off-the-shelf ARM IP, they'll quit making SoCs, because at the end of the day, what they care about is the amount of tablets and phones sold. Same goes for other companies and the markets they want to compete in. Every ARM SoC vendor is also a competitor of the other ARM users.

Intel can surely attack ARM, but ARM won't go away because ARM also serves markets that could never generate a satisfying amount of money for Intel. Phones and tablets do, and therefor they'll join Qualcomm, Nvidia, Samsung and MediaTek in this space. Today companies use Qualcomm because they have the fastest SoCs, but when Intel will have those, the SoC supplier will be changed in disfavor of Qualcomm.

you were on to something about ARM, and what makes the most money wins; economic factors surely are critical, but it may not just be that; samsung and apple design their own soc's and use specific os's for their ecosystem (more so with apple). so its not just a chip design/process dynamic, but an os (proprietary) inclusion into that dynamic.

i don't expect apple/samsung to jump ship for x86 in mobile, there just isn't enough motivation to make a switch, even a solid intel advantage of for both cpu+gpu likely wouldn't encourage a change. there maybe more a sense of corporate pride for r&d, so a compelling intel solution perf/w and $ maybe ignored.

intel's hope relies on those white-box makers for volume, but also just better marketing, but that's a whole 'nother story.

 

[raghu78]

Your failed premise is that ARM is a collaborative effort to kill Intel. It isn't.

Not kill but to definitely compete for marketshare in every part of the computing spectrum. Till ARMv7 that portion was mobile and embedded . With ARMv8 its literally every market where Intel participates. Jim Keller was high praise for the ARMv8 ISA. As the lead designer of Apple Cyclone, Keller has shown what a high performance ARMv8 core looks like. It looks like a Intel big core like ivybridge. Intel is trying to get into ARM's stronghold (mobile) and ARM is trying to make inroads into Intel's stronghold (servers). Eventually the ARM ecosystem by its sheer size, resources and customizability will have a bigger impact on Intel than vice- versa.

Every company that uses ARM's ISA does what it thinks will gain them most money. If Samsung thinks they can better stop making SoCs because the RoI isn't high enough anymore or they can't compete with their off-the-shelf ARM IP, they'll quit making SoCs, because at the end of the day, what they care about is the amount of tablets and phones sold. Same goes for other companies and the markets they want to compete in. Every ARM SoC vendor is also a competitor of the other ARM users.

What a failed argument. Samsung is a IDM for their products and a foundry for others like Apple. There is no way Samsung will choose Intel across the product stack because Intel sells finished chips. that would mean loss of business for Samsung Semiconductor. thats not going to happen. The reason Samsung will buy Qualcomm chips is because they can manufacture the chips at their fabs. ARM vendors are going to be even more aggressive taking marketshare from x86.

Intel can surely attack ARM, but ARM won't go away because ARM also serves markets that could never generate a satisfying amount of money for Intel. Phones and tablets do, and therefor they'll join Qualcomm, Nvidia, Samsung and MediaTek in this space.

forget ARM going away. They are getting bigger and stronger at Intel's expense. AMD which got hurt badly due to tablets realized their best option is going with both platforms as both are going to be the dominant platforms. AMD's K12 will extend the performance range of ARM into the high end of servers .

Today companies use Qualcomm because they have the fastest SoCs, but when Intel will have those, the SoC supplier will be changed in disfavor of Qualcomm.

You think Qualcomm is going to sit twiddling their thumbs. Qualcomm is the No.1 mobile SOC vendor because of their leadership in LTE baseband integration and wide compatibility across a range of carrier networks. So its wireless tech is the reason for their dominance. Intel is not aiming at a static target but a dynamic one. Qualcomm continues to invest in high end custom ARM cores, wireless tech and graphics. Qualcomm also has superior graphics performance than Intel which is more important than CPU in mobile devices. The foundries are closing the gap at 16/14 FINFET. This helps the ARM partners like Apple, Qualcomm, AMD, Nvidia to better compete against Intel. If anything Intel is going to see more competition in traditional x86 markets like desktops, notebooks and servers going forward.

 

[teejee]

This is based one the assumption that there is such thing as a "good" SoC and a "bad" SoC, and that each player is going to aim for the same target.

Where there is not added value to design, you can simply license a well know piece of IP, so you end up sharing the cost for this, be it the CPU, the GPU or any other piece of the SoC.

Intel is obviously able to get a good "SoC" in some metrics, but they can't compete in all of them at the same time (performance, power, cost, integration, differentiation, speed of execution, ...) and the pain to switch to x86 outweighs the benefits for most OEMs for now.

I agree, this is very important. Intel has right now superior production technology but it is even more important to have the right product, i e the product the product companies want.
Intel is used to control this, they more or less owns the roadmap for the whole PC industry. But when it comes to phones/tablet it is the opposite, they have almost zero control. They have to design a SOC and then hope it is the right product at the right time.

Subsidizing tablet SOC's in china in order to reach their volume target is an example how Intel choose to handle not having the right product.

Another example where Intel failed with their product planning is that all their phone SOC's so far have had high-end CPU performance but only mid-range GPU performance. So they didn't fit well in any phone category.

 

[Third_Eye]

Your failed premise is that ARM is a collaborative effort to kill Intel. It isn't. Every company that uses ARM's ISA does what it thinks will gain them most money. If Samsung thinks they can better stop making SoCs because the RoI isn't high enough anymore or they can't compete with their off-the-shelf ARM IP, they'll quit making SoCs, because at the end of the day, what they care about is the amount of tablets and phones sold. Same goes for other companies and the markets they want to compete in. Every ARM SoC vendor is also a competitor of the other ARM users.

I think your failed premise is that you misunderstood what I meant to say.

ARM is not there to kill Intel. ARM is there to promote its ISA. ARM thrives when more companies use the ARM ISA. ARM provides choice to its clients.

a) Want nothing but ARM designed SOC. Processor-On-Package solutions (POP). ARM works with foundries TSMC and GF to provide that.

b) Want to license the Core and then do the integration to implement it whatever way you chose? Go ahead SMSG, Mediatek, Allwinner, NVIDIA et all.

c) Want to go on your own and create your own product with just compatibility to ARM ISA. You can do that like QCOM, Marvel and Apple today and NVIDIA, SMSG, AMD soon.

d) Is your new custom ISA compatible 64 bit core running late? No problem, I will license my Cortex A53/A57 core so that you can use that as a standby till your in-house core is ready... (QCOM's first 64bit S810 family)

e) Want to be the second, but implement your own way to mix and match high performance+low power instead of using the basic big-little approach? NVIDIA go ahead with T3 and T4

While ARM thrives by expanding the usage of its ISA, Intel thrives because by restricting x86/x64 ISA. Today AMD is the only other company that is left standing.

The reason x86 came up compared to far superior architectures is because it had
a) It grew bottom up. Slowly eating its RISC competition from below.
b) Volumes of scale and manufacturing.
c) The WINTEL eco-system is unmatched till date in the corporate & home computing world. It is totally by chance/accident. This fed INTEL with steady cash flow which allowed them to overtake their competition.
d) Intel had also greatly covered (invested) in other platforms that could have dethroned MS (like Linux and even Netware) but did not. The former did improve Intel's standing in servers.
e) Similarly MS after trying to get NT run on both MIPS as well as PowerPC closed and stuck with x86.

Intel's clients fear INTEL. It is Intel that makes most of the profits on its x86/x64 products. Their clients have to fight for scraps or make money of value-added services.
ARM's clients like ARM. The clients have more control of making money with the selected model. ARM just makes of licensing.

HUGE DIFFERENCE!


a) In the mobile arena, there is no WINTEL trump-card.
b) Intel is playing catch-up. For all goodness in Baytrail, it does not have a proper Android implementation that Intel x86 Android tablets are still using the crappy Clovertrail.
c) The software eco-systems triggered by Phones and tablets do not have a Intel ISA Inside (no pun intended).
d) ARM is biting (not eating) INTEL from the below.
e) Even MS has decided that it is best for its mobile adventures to concentrate on ARM architecture. x86/x64 is purely legacy at this point.

The only thing that INTC has in this new world is a lead in manufacturing, which is sustained by the profits its gets from its x86/x64 division. LOL! Anything that was non x86 has bled and died inside Intel (Intel RISC, Intel Itanium, Intel TV). Now even certain x86 segments are bleeding.
With other fab manufacturers reducing the gap, the problems INTC has filling up its fabs, 450nm wafer postponement, ARM eco-system being better, things are a little different now.

Intel can surely attack ARM, but ARM won't go away because ARM also serves markets that could never generate a satisfying amount of money for Intel. Phones and tablets do, and therefor they'll join Qualcomm, Nvidia, Samsung and MediaTek in this space. Today companies use Qualcomm because they have the fastest SoCs, but when Intel will have those, the SoC supplier will be changed in disfavor of Qualcomm.

The same thing INTEL did to the RISC heavy weights. They had their turf and did not want to come down to the bottom. Intel grew bottom up. Now ARM is growing bottom-up. More market share and mindshare. More Developer attention. Keep in mind that it was the developer Apple is making a good case for how a 64 bit ARMV8-ISA implementation is as good as a Baytrail x86 implementation in both power and performance at a lower clock speed and 1 manufacturing node below.

 

[witeken]

Not kill but to definitely compete for marketshare in every part of the computing spectrum. Till ARMv7 that portion was mobile and embedded . With ARMv8 its literally every market where Intel participates. Jim Keller was high praise for the ARMv8 ISA. As the lead designer of Apple Cyclone, Keller has shown what a high performance ARMv8 core looks like. It looks like a Intel big core like ivybridge. Intel is trying to get into ARM's stronghold (mobile) and ARM is trying to make inroads into Intel's stronghold (servers). Eventually the ARM ecosystem by its sheer size, resources and customizability will have a bigger impact on Intel than vice- versa.

As far as I know, Intel is still the biggest semiconductor company with the highest R&D budget and a multiple year process lead, while ARMv8 can only be found in 1 phone, so I think Intel is safe, but Intel on the other hand has the potential to disrupt the smartphone market with their immense process lead. No ISA can save you when you have costlier transistors that perform worse and consume more energy.

What a failed argument. Samsung is a IDM for their products and a foundry for others like Apple. There is no way Samsung will choose Intel across the product stack because Intel sells finished chips. that would mean loss of business for Samsung Semiconductor. thats not going to happen. The reason Samsung will buy Qualcomm chips is because they can manufacture the chips at their fabs. ARM vendors are going to be even more aggressive taking marketshare from x86.

I don't think it's failed. Samsung was just an example. Both Apple and Samsung don't rely (purely) on the silicon they design to generate money, because they do that with their tablets and phones. My point is that all companies follow the money. If they can't generate money or it isn't worth to stay in a market, they will stop making SoCs and go to someone else for their chips. Likewise, the argument was that Qualcomm could do the high-end of the market, MediaTek the low-end, etc. But it doesn't work that way. MediaTek can and will also take an A57 core from ARM and put it into their SoCs and directly compete with Qualcomm and Nvidia, that's how the market works.

forget ARM going away. They are getting bigger and stronger at Intel's expense. AMD which got hurt badly due to tablets realized their best option is going with both platforms as both are going to be the dominant platforms. AMD's K12 will extend the performance range of ARM into the high end of servers .

Intel's expense? I don't see any sign of that. But with Intel invading into the tablet market this year and phones next year, the story is different from ARM. BTW, without x86, you can't do much on the desktop market, which is Intel's biggest source of money. About the same goes for servers.

You think Qualcomm is going to sit twiddling their thumbs. Qualcomm is the No.1 mobile SOC vendor because of their leadership in LTE baseband integration and wide compatibility across a range of carrier networks. So its wireless tech is the reason for their dominance. Intel is not aiming at a static target but a dynamic one. Qualcomm continues to invest in high end custom ARM cores, wireless tech and graphics. Qualcomm also has superior graphics performance than Intel which is more important than CPU in mobile devices.

That's how technology works, everything always keeps improving. Intel's doing the same and soon Qualcomm's "leadership in LTE baseband ..." will be gone, just like their graphics lead.

The foundries are closing the gap at 16/14 FINFET. This helps the ARM partners like Apple, Qualcomm, AMD, Nvidia to better compete against Intel. If anything Intel is going to see more competition in traditional x86 markets like desktops, notebooks and servers going forward.

The gap is growing, not shrinking. TSMC's 16nm really is 20nm FinFET, and Intel will have a 10nm long before TSMC even has its real 14nm.

 

[ShintaiDK]

Yes, 10s of millions of x86 phones and tablets where in Intel is losing money in the multi-billion dollar range last 2 years every and the near future



As a note, Intel is able to subsidize these gargantuan (mobile)losses purely bcos of the insane profits provided by the x86-x64 PC+Laptop+Data center CPUs as well as continue the huge lead in manufacturing. All the various ARM vendors need to do is to attack piece by piece the profit generating sectors.
AMD + Applied Micro attack the server aspect of it. QCOM(already successful) and MediaTek the phone section. AllWinner, Rockchip, Mediatek the cheap tablets, Apple, Samsung the higher tablet section. That way Intel would really "compete".
Intel cannot attack ARM. It can win over 1 vendor, the other vendor may cause problems. Same thing like Microsoft and Linux whacamole

Remove modems from Qualcomm, and you got an entirely different company than you expect.

Samsung and Apple doesnt care about ARM, they both only care about device sales. Because thats the only place money is for them. The rest of the companies you list struggles to even sell for 300M$ in MPU revenue.

Again, ARM have moved nowhere since it started, x86 went into ARM segments.

 

[monstercameron]

Remove modems from Qualcomm, and you got an entirely different company than you expect.

Samsung and Apple doesnt care about ARM, they both only care about device sales. Because thats the only place money is for them. The rest of the companies you list struggles to even sell for 300M$ in MPU revenue.

Again, ARM have moved nowhere since it started, x86 went into ARM segments.

• remove x86 isa from intel
• remove graphics from nvidia/AMD
• etc.

your point?

 

[raghu78]

As far as I know, Intel is still the biggest semiconductor company with the highest R&D budget and a multiple year process lead, while ARMv8 can only be found in 1 phone, so I think Intel is safe, but Intel on the other hand has the potential to disrupt the smartphone market with their immense process lead. No ISA can save you when you have costlier transistors that perform worse and consume more energy.

Everybody knows that ARM is an inherently more efficient ISA than x86. Don't take my word. Microprocessor design veterans like Jim Keller are of that view

http://techreport.com/review/26418/amd-reveals-k12-new-arm-and-x86-cores-are-coming

Jim Keller was very complimentary about the ARMv8 ISA in his talk, saying it has more registers and "a proper three-operand instruction set." He noted that ARMv8 doesn't require the same instruction decoding hardware as an x86 processor, leaving more room to concentrate on performance. Keller even outright said that "the way we built ARM is a little different from x86" because it "has a bigger engine." I take that to mean AMD's ARM-compatible microarchitecture is somewhat wider than its sister, x86-compatible core. We'll have to see how that difference translates into performance in the long run.

http://www.youtube.com/watch?v=eTVnxaXCLg0

see video at 54:50 to 55:20

btw Intel's process lead has not helped them in the mobile market where baseband integration is key. The ARMv8 cores are being actively designed into mobile SOCs.

http://www.anandtech.com/show/7925/...bit-socs-with-lte-category-67-support-in-2015

Qualcomm has standard ARMv8 64 bit CortexA57/A53 cores based top to bottom product stack with integrated baseband in 2015. Qualcomm continues to work on their custom ARMv8 core and will launch in 2016 at 16/14 FINFET.

I don't think it's failed. Samsung was just an example. Both Apple and Samsung don't rely (purely) on the silicon they design to generate money, because they do that with their tablets and phones. My point is that all companies follow the money. If they can't generate money or it isn't worth to stay in a market, they will stop making SoCs and go to someone else for their chips. Likewise, the argument was that Qualcomm could do the high-end of the market, MediaTek the low-end, etc. But it doesn't work that way. MediaTek can and will also take an A57 core from ARM and put it into their SoCs and directly compete with Qualcomm and Nvidia, that's how the market works.

Competing with Qualcomm is easier said than done. Also you don't realize that the top ARM partners are licensees who design custom cores to differentiate their products - Apple Cyclone, Qualcomm, AMD (K12), Nvidia (Denver), Marvell. That kind of effort is not easy for companies like Mediatek who do not have the resources to design custom cores.

Intel's expense? I don't see any sign of that. But with Intel invading into the tablet market this year and phones next year, the story is different from ARM. BTW, without x86, you can't do much on the desktop market, which is Intel's biggest source of money. About the same goes for servers.

Is the PC market shrinking or not. Intel is shrinking in revenues but more significantly in profitability. AMD's computing division revenues got gutted by the explosive growth of the tablet market. Tablets cannibalized low end notebooks and AMD was overexposed in the low end with Brazos and did not have a presence in high end desktops and servers like Intel to offset those massive losses.

http://files.shareholder.com/downlo...-4079-b7f7-956c8821cd8c/Intel_ARand10K_13.pdf

page 35. Intel's PC Client Group lost 2.6 billion revenue from 2011 to 2013.

http://ir.amd.com/phoenix.zhtml?c=74093&p=quarterlyearnings

AMD 2013 CS revenues were down to 3.1 billion in 2013 from 4 billion in 2012. Who gained at the expense of the x86 PC market. the ARM ecosystem of tablets, smartphones and notebooks like ARM based chromebooks.

That's how technology works, everything always keeps improving. Intel's doing the same and soon Qualcomm's "leadership in LTE baseband ..." will be gone, just like their graphics lead.

vice-versa. ARMv8 cores like Cyclone are competing with Intel big cores. So Intel's CPU leadership also faces the same challenges as ARMV8 custom cores like K12 and Denver go for Intel' server market share.

The gap is growing, not shrinking. TSMC's 16nm really is 20nm FinFET, and Intel will have a 10nm long before TSMC even has its real 14nm.

Intel's node naming is misleading.

http://www.electronicsweekly.com/mannerisms/general/the-intel-nanometre-2013-02/

Intel 22nm has no significant transistor density advantage over the foundries 28nm though it does have performance advantage due to FINFET. Intel's 22nm and TSMC/Samsung/GF 28nm all use single pattern immersion lithography.

Intel's 14nm FINFET and TSMC 16FF+ and Samsung 14 FINFET are all of similar transistor density with a marginal lead to Intel. They use dual pattern immersion litho with a M1 metal pitch of 64 nm. TSMC 16FF+ has a 15% area scaling over TSMC 20nm planar. Samsung 14 FINFET has a similar 15% area scaling over 20nm planar designs.

https://markets.jpmorgan.com/research/email/-kjegkq4/GPS-1336259-0

page 4

http://globalfoundries.com/docs/def...dries-14nm-collaboration---final.pdf?sfvrsn=2

If Intel 22nm FINFET Baytrail is competing with 28nm planar tech based mobile SOCs from Qualcomm then you can figure out for yourself how things are going to be with no process node advantage in 2016.

 

[raghu78]

Remove modems from Qualcomm, and you got an entirely different company than you expect.

Samsung and Apple doesnt care about ARM, they both only care about device sales. Because thats the only place money is for them. The rest of the companies you list struggles to even sell for 300M$ in MPU revenue.

Again, ARM have moved nowhere since it started, x86 went into ARM segments.

remind me again which two companies lost close to 2 billion each in annual revenue over the last 2 years in their PC divisions. Intel and AMD. Intel was able to offset with server sales increase while AMD was gutted. :thumbsdown:

 

[witeken]

For those who are skeptical about Intel's ability to enter and dominate those markets where ARM is currently used, I'd like to give you the perspective of Intel's CEO. It's from a few quarters ago, but I think it summarizes the situation very well.

Brian Krzanich - Chief Executive Officer, Director
So, yes. From that perspective, they see a roadmap now. I think more importantly than just 14 nanometer. What they see is a roadmap from us around the Atom SoC. They see Bay Trail as a great first step and great product. You have seen some of the performance benchmarks out there. We just talked about the transistor density. Stacy has talked about our cost and ability to hit these lower prices points. It has got good graphics performance. As I said we are able to provide a 64-bit solution across all OS options as well.

So they look at that and then they look at our roadmap, and say okay, they have got LTE, they have got connectivity, they have got a standalone, they have plans to integrate those technologies in, in a basic SoC. They are starting to build confidence in our roadmap along with us. That's, I think, what they really look at this as they see our Atom roadmap that's highly competitive.

I don't want to search for it now, but I've also seen a quote which claims that multiple companies, who first were costumers of MediaTek, a company known for its cheap products, are now switching to IA. So, with that said, I personally don't think there's any substance to doubt Intel's or x86's ability to pursue those markets. The only place where I've seen those doubts, are from internet users (or ARM itself, which isn't surprising).

 

[raghu78]

For those who are skeptical about Intel's ability to enter and dominate those markets where ARM is currently used, I'd like to give you the perspective of Intel's CEO. It's from a few quarters ago, but I think it summarizes the situation very well.

I don't want to search for it now, but I've also seen a quote which claims that multiple companies, who first were costumers of MediaTek, a company known for its cheap products, are now switching to IA. So, with that said, I personally don't think there's any substance to doubt Intel's or x86's ability to pursue those markets. The only place where I've seen those doubts, are from internet users (or ARM itself, which isn't surprising).

Witeken
all said and done the companies which lost revenue in the last 2 years were x86 PC companies. They were cannibalized by ARM partners selling mobile devices- smartphones, tablets and notebooks(Chromebooks).

The same is now going to extend to servers with ARMv8. You can wait and see how this unfolds for x86 and ARM.

 

[cbn]

Both Intel and AMD have made noises about becoming a bit more like ARM, Intel has offered to integrate IP onto their x86 designs and AMD has offered to make custom ICs ala console chips, in an attempt to gain some of the diversity of effort ARM leverages.

Yes, specifically with the Goldmont atom core, the Intel CEO referred to it as having a chassis that allowed easy integration.

How Intel implements this will be very interesting. It would be great to see Intel be able to occupy more niches with atom. (to keep the market competitive and to benefit us as consumers).

With Cortex A57, I can already imagine the multitude of SOCs and niches that could be potentially filled due to the sheer number of partners ARM has (Even Rockchip has licensed Cortex A57).....so Intel with its atom needs to compete.

 

[Third_Eye]

For those who are skeptical about Intel's ability to enter and dominate those markets where ARM is currently used, I'd like to give you the perspective of Intel's CEO. It's from a few quarters ago, but I think it summarizes the situation very well.



I don't want to search for it now, but I've also seen a quote which claims that multiple companies, who first were costumers of MediaTek, a company known for its cheap products, are now switching to IA. So, with that said, I personally don't think there's any substance to doubt Intel's or x86's ability to pursue those markets. The only place where I've seen those doubts, are from internet users (or ARM itself, which isn't surprising).

That quote was from the great "Digitimes".
Important note.

MediaTek competes with Intel more in the phone areas. In tablets where baseband is not the main thing, Intel with its Baytrail is far better (if it can get Android right). So Intel has to compete with Allwinner and Rockchip which still targets 1 step lower. And Intel plans to lose money for the foreseeable future (i.e) contra revenue. And keep in mind the Contra revenue recipients are mainly some Chinese ODMs who will receive free money, bribe and promotion for utilizing the Intel platform. Sounds like a good deal to me ;-D

MediaTek will have its Global LTE based solution in Q2-2014. By the end of this year they will have a 64bit ARM core with LTE baseband.
The closest Intel will come to compete with Mediatek head-on would be in Q2-2015 with the SOFIA chipset which is its very low end at that time frame based on 28nm TSMC process. At that time MediaTek will be putting out 64 bit compatible ARMv8 + LTE baseband + Power VR 6+ based chips.

So I read BK's statement with a grain of salt. With respect to their 14nm plans, their 2 year lead is shrinking to most likely 1 year now. And Apple has shown that they can compete with Baytrail 1 node down. In end 2015 all their competitors will have 14nm/16nm FinFET hybrids producing chips closing on more.

 

[cbn]

In tablets where baseband is not the main thing, Intel with its Baytrail is far better (if it can get Android right). So Intel has to compete with Allwinner and Rockchip which still targets 1 step lower.

Intel has Merrifield though.

So Merrifield would be the 1 step lower chip to Bay Trail.