New AMD CEO readies strategy shift

[senseamp]

LOL. Don't be silly.

It would be game over for AMD if they walk away from x86, that is the only thing they have going for them.

What do they have going for them in the x86 market, aside from huge losses?

http://ycharts.com/companies/AMD/profit_margin

 

[LOL_Wut_Axel]

What do they have going for them in the x86 market, aside from huge losses?

http://ycharts.com/companies/AMD/profit_margin

Fusion has been doing great for them, last I looked.

 

[CHADBOGA]

What do they have going for them in the x86 market, aside from huge losses?

http://ycharts.com/companies/AMD/profit_margin

I never said that AMD have a lot going for them, just what little they do have, is intricately tied to x86.

The x86 processor market has very high margins and thus the potential for profitability is quite high if you can get your act together and there are really only two players.

The margins in the ARM market are far lower and there are many times more players than is the case in the x86 market.

 

[senseamp]

I never said that AMD have a lot going for them, just what little they do have, is intricately tied to x86.

The x86 processor market has very high margins and thus the potential for profitability is quite high if you can get your act together and there are really only two players.

The margins in the ARM market are far lower and there are many times more players than is the case in the x86 market.

Even if you include the time when AMD got its act together with the Athlon, and the high margins, it has destroyed an enormous amount of shareholder value over the years. It may have a lucky streak here and there, but in the long run, the house always wins, and on x86 turf, the house is Intel. On ARM turf, there is no house. AMD would be competing with companies of comparable size and capabilities, instead of being beaten to a bloody pulp by the 800lb gorilla. ARM is the future of the consumer space, x86 is going to be a CPU for workstations and servers, before eventually getting out-commoditized from below by ARM, just like x86 out-commoditized RISC workstation and server CPUs.

 

[lol123]

Even if you include the time when AMD got its act together with the Athlon, and the high margins, it has destroyed an enormous amount of shareholder value over the years. It may have a lucky streak here and there, but in the long run, the house always wins, and on x86 turf, the house is Intel. On ARM turf, there is no house. AMD would be competing with companies of comparable size and capabilities, instead of being beaten to a bloody pulp by the 800lb gorilla. ARM is the future of the consumer space, x86 is going to be a CPU for workstations and servers, before eventually getting out-commoditized from below by ARM, just like x86 out-commoditized RISC workstation and server CPUs.

There is no player in the world that is ever going to "out-commoditize" Intel the way Intel out-commoditized comparatively tiny companies like Sun, DEC and MIPS (IBM never was out-commoditized) in the workstation and server markets. If you believe that then I don't think you understand the fundamental facts of this industry.

 

[senseamp]

There is no player in the world that is ever going to "out-commoditize" Intel the way Intel out-commoditized comparatively tiny companies like Sun, DEC and MIPS (IBM never was out-commoditized) in the workstation and server markets. If you believe that then I don't think you understand the fundamental facts of this industry.

A revolution is impossible, until it's inevitable.

ARM ecosystem as a whole could out-commoditize Intel, no need for any one player to do it. ARM is the commodity CPU core, anyone can make it.

 

[lol123]

A revolution is impossible, until it's inevitable.

ARM ecosystem as a whole could out-commoditize Intel, no need for any one player to do it. ARM is the commodity CPU core, anyone can make it.

There are three major reasons why that isn't possible:

1. Different ARM designs are not necessarily platform compatible. This is why Microsoft will have to maintain 4 (!) different binaries for Windows 8 for ARM.
2. R&D budgets are spread thin across different companies in the ARM ecosystem, enabling no one player to gain process technology or architectural sophistication comparable to those of Intel.
3. Even the ARM ecosystem as a whole is nowhere near Intel in terms of R&D budget, margins or revenue.

 

[senseamp]

There are three major reasons why that isn't possible:

1. Different ARM designs are not necessarily platform compatible. This is why Microsoft will have to maintain 4 (!) different binaries for Windows 8 for ARM.

That is a bogus claim by Intel

2. R&D budgets are spread thin across different companies in the ARM ecosystem, enabling no one player to gain process technology or architectural sophistication comparable to those of Intel.

Process technology R&D budget is not spread thin at all. It's at a handful of foundries like TSMC.
Architectural R&D budgets are not spread that thin either, it's primarily ARM licensing the implementation, with a handful of big and capable companies doing their own off the architectural license.

3. Even the ARM ecosystem as a whole is nowhere near Intel in terms of R&D budget, margins or revenue.

Which is good, because that keeps the prices low. Commodity segments are not characterized by high margins, but fierce competition and low prices.

 

[lol123]

Which is good, because that keeps the prices low. Commodity segments are not characterized by high margins, but fierce competition and low prices.

Actually, Intel's success with establishing x86 as the dominant architecture in both the PC and server markets (despite lots of people believing in the 90's that RISC architectures are inherently superior and would win out in the long run) stems partly from economics of scale (which correspond to its sales volumes and revenue) and partly from its massive R&D and process tech advantage (which comes from its high profit margins). These same basic economic laws apply to ARM as well, although by all means you ARM fanatics can keep dreaming all you like about how they won't and about how Qualcomm, Samsung and Nvidia will join together as a happy family and pool their resources in their quest to destroy Intel's evil monopolistic empire. :whiste:

The fact however is that just 8 months ago, we heard some very cocky statements from the ARM camp and the ARM hype reached an apex. After Intel rolled out its shock and awe campaign with the 22nm tri-gate announcement, accelerated Atom roadmap and recently even "InGaAs quantum well transistors" and an experimental x86 chip idling at 10 mV, the ARM stock has stalled in the markets and both the ARM CEO and Jensen have become awfully quiet. Wonder why that is?

 

[senseamp]

Nope, I don't wonder about it at all. ARM is cleaning the floor with Intel in the fastest growing market segments of smartphones and tablets. Wonder why that is? Microsoft obviously is not buying Intel's hype, they wouldn't be making Windows for ARM if they thought x86 would be the dominant architecture for future consumer devices. They see where the wind is blowing and are allocating their R&D money accordingly.

 

[lol123]

You do know that Windows used to support Alpha and Itanium (the latter until very recently) too right? Do you think that meant they were betting for those architectures against x86?

 

[senseamp]

You do know that Windows used to support Alpha and Itanium (the latter until very recently) too right? Do you think that meant they were betting for those architectures against x86?

NT was developed before even Pentium Pro. x86 was not a scalable server architecture at that time. But Itanic is a good example of Intel's arrogance and telling the market segments what they need instead of listening. If they listened, they sure wouldn't hear any phone makers asking for an x86 phone chip

 

[Lonbjerg]

NT was developed before even Pentium Pro. x86 was not a scalable server architecture at that time. But Itanic is a good example of Intel's arrogance and telling the market segments what they need instead of listening. If they listened, they sure wouldn't hear any phone makers asking for an x86 phone chip

The server crowd didn't ask for x86 either...

 

[Blitzvogel]

I know there are difficulties in adapting x86 to compete directly against ARM, and that is certainly an issue, but the tablet market is in reach at least.

AMD needs to go after the ARM market but needs to simplify their line of CPU products, which includes getting Bulldozer out and readying it's successor as soon as possible (IE 2 years).

Brazos Successor + Derivatives ---> Phones, Tablets, and very low end laptops/netbooks. Dual cores/80 SPs for mobile, quad-core/160 SPs/dual channel DDR3 for laptops

Bulldozer + New Radeon architecture based Fusion ---> For mainstream x86 computing for both laptops and desktop. Chipset needs DDR3 capabilities pushed including tri-channel for higher end SKUs

Bulldozer CPUs ---> For servers and high end desktop computing.

 

[cbn]

Intel has leading edge process tech to differentiate their products in the mobile space from ARM and lagging foundry process tech...AMD does not have this option.

If you are going to compete while using less stellar process tech then you need a more stellar processor design.

That makes sense to me.

But what ARM CPU or SOC design would AMD use? (It seems like there are so many players in this area)

Will they go for Cortex A7?

 

[cbn]

More Information on the Cortex A7 (codename: "Kingfisher")

http://www.anandtech.com/show/4991/...dualcore-more-power-efficient-highend-devices

 

[Idontcare]

That makes sense to me.

But what ARM CPU or SOC design would AMD use? (It seems like there are so many players in this area)

Will they go for Cortex A7?

I think it also made sense to AMD execs a decade ago which is why the strategic wheels were set in motion for them to pursue acquiring Nvidia, and then ATI when the NV deal fell through, so they could develop APU's.

The market is crowded, you said it right. AMD could do a me-to approach to Nvidia's Tegra, but that's not a winning strategy to always be one-step behind Nvidia on one hand and one step behind Intel on the other.

They need to jump out there in some direction that will enable them to lead in that marketspace no matter how small it is. Nobody really survives long in this industry by being second-best in a lot of different things but never first in anything.

They need something for which they can hang their hat on, even if it is just a $100m product lineup in some otherwise obscure device market (like medical or something).

Honestly I think they had the right idea with fusion. Nvidia can't pursue them because they don't have x86, and Intel can't really pursue them because their GPU tech isn't even remotely competitive from the non-2D stuff.

 

[cbn]

The market is crowded, you said it right. AMD could do a me-to approach to Nvidia's Tegra, but that's not a winning strategy to always be one-step behind Nvidia on one hand and one step behind Intel on the other.

They need to jump out there in some direction that will enable them to lead in that marketspace no matter how small it is. Nobody really survives long in this industry by being second-best in a lot of different things but never first in anything.

They need something for which they can hang their hat on, even if it is just a $100m product lineup in some otherwise obscure device market (like medical or something).

It would be great to see Mr. Rory Read take some creative and niche direction with AMD. I particularly like the small product line-up idea you have mentioned.

Hmm....I wonder what plans they have hashed out at this point? Will his ideas be truly inspired? Or simply refined "Me-too" ideas like we expect from most other people?

P.S. I do think it is interesting that he is from Lenovo in light of the industry predictably moving away from expensive custom designed laptop models towards Smartphones and re-useable "Lap docks". My suspicion therefore is that he strongly believes in mobile....... but exactly how he will implement this strategy remains to be seen!

 

[wahdangun]

I think it also made sense to AMD execs a decade ago which is why the strategic wheels were set in motion for them to pursue acquiring Nvidia, and then ATI when the NV deal fell through, so they could develop APU's.

The market is crowded, you said it right. AMD could do a me-to approach to Nvidia's Tegra, but that's not a winning strategy to always be one-step behind Nvidia on one hand and one step behind Intel on the other.

They need to jump out there in some direction that will enable them to lead in that marketspace no matter how small it is. Nobody really survives long in this industry by being second-best in a lot of different things but never first in anything.

They need something for which they can hang their hat on, even if it is just a $100m product lineup in some otherwise obscure device market (like medical or something).

Honestly I think they had the right idea with fusion. Nvidia can't pursue them because they don't have x86, and Intel can't really pursue them because their GPU tech isn't even remotely competitive from the non-2D stuff.

spot on,

and its very stupid if AMD abandon X86, and we don't know if application in win8 arm ver. will be compatible with x86

 

[Dribble]

Don't see how AMD can get into ARM like nvidia have? It cost nvidia $2 billion, and say 5 years of hard work. AMD have neither the money or the time.

 

[blckgrffn]

The most interesting aspect of this to me is that AMD is in a relatively unique position to bolt an x86 interpreter on an ARM based CPU, are they not? As I recall, Intel bolted one on to EPIC/Itanium.

It shouldn't be that foreign of a concept, given that their CPUs are basically RISC (mircro-ops FTW) machines behind a x86 interpreter already, right?

Wouldn't a chip that could run both ARM and x86 applications be very desirable in the market place?

Damn the performance and all that, it would be a great marketing point I would think.

 

[StrangerGuy]

It would be great to see Mr. Rory Read take some creative and niche direction with AMD. I particularly like the small product line-up idea you have mentioned.

Hmm....I wonder what plans they have hashed out at this point? Will his ideas be truly inspired? Or simply refined "Me-too" ideas like we expect from most other people?

P.S. I do think it is interesting that he is from Lenovo in light of the industry predictably moving away from expensive custom designed laptop models towards Smartphones and re-useable "Lap docks". My suspicion therefore is that he strongly believes in mobile....... but exactly how he will implement this strategy remains to be seen!

Didn't AMD gave away their smartphone GPU tech on a silver platter under Meyer? It's too lateeee...

 

[Munky]

ARM is better suited to low power, mainly because it's instruction decoding is very simple, which basically saves millions of transistors and lots of power on the front end.

It also limits you to software written and compiled for ARM, as opposed to the much higher number of apps available for x86. I personally would rather have x86-compatible CPU in a tablet.

 

[cbn]

and its very stupid if AMD abandon X86, and we don't know if application in win8 arm ver. will be compatible with x86

http://en.wikipedia.org/wiki/Windows_8

According to that link old x86 apps will need to be for ported for ARM.

I'd imagine the same will be true for ARM apps on x86.

 

[cbn]

Didn't AMD gave away their smartphone GPU tech on a silver platter under Meyer? It's too lateeee...

I'm not sure how much this will really matter.

According to the Anandteh Silvermont Articlehere we are likely to see Silvermont Atoms equipped with Sandy Bridge level graphics. That is pretty strong Graphicsevolution for a handheld device!

So wouldn't this require only minimal changes from AMD on the GPU side of things? Maybe a switch to low power process tech?