Intel's smartphone platforms

What your saying is understandable, but what he is saying is basically what is already known.
If I had to make a list of things that were "already known" in this space and were not exactly accurate, if not completely false..

For instance when you license an ISA, design&validate a CPU from scratch based on it, have another company manufacture it for you, etc your product is not magically cheaper than other products that are "already known" to be expensive. Truth is only a few know the complicated cost structure of these products (and I am not among those :smile:) and simply because a lot of people think they do..well, that means nothing :)

Wasn't already "known" by many that x86 couldn't even compete in the low power department with ARM and suddenly a year ago people started to realize that..umh..wait a secont, that wasn't so true.
 
Wasn't already "known" by many that x86 couldn't even compete in the low power department with ARM and suddenly a year ago people started to realize that..umh..wait a secont, that wasn't so true.
Will Intel let companies design their own SoC with Silvermont?

Will Intel guarantee they will never prioritirize a company for chips delivery? If not will they let a company bring Silvermont to another foundry?

Will Intel let a company makes its own x86 CPU to increase differentiation against competition?

Will Intel design ultra low cost x86 CPU for lower end smartphones so that a phone maker can have the same software for all of its phones?

I think Intel will have to heavily change its business model to make a significant inroad in the phone market. Being the best technically is very far from enough.
 
All pricing info I can find on Intel Atom is this:
http://ark.intel.com/products/70105
That's $41 for their tablet chip.


Here is a list of some of their older Atom chips: http://ark.intel.com/products/family/29035

Which is fairly meaningless except as a jumping off point for negotiations between the Intel sales team and the customer's sales team.

A company with a good sales team buying in large enough quantities can likely significantly undercut that. Smaller quantities and an inept sales team will lead to a company paying what is listed on that page.

Just as the page says, it's the "recommended customer price." It's what Intel would like to get, but isn't necessarily what a customer would pay.

Without being privy to the confidential sales talks between Intel and it's customers, it's impossible to say what those go for on average or even what the lowest price being paid for them is. And those talks are confidential for a reason. Not only does Intel not want customers to know what other customer's are actually paying, other customers don't want their competition to know. Especially if one customer feels they are getting a better deal than their competition. And companies always feel like they are getting at least a somewhat good deal if they aren't paying the "recommended customer price."

Regards,
SB
 
Anand strongly hinted that Intel will have some high volume design wins with Silvermont, and Intel strongly hinted that they will have low cost Atom-based tablets available near the end of the year. I wonder if Amazon will be a partner...?
 
Which is fairly meaningless except as a jumping off point for negotiations between the Intel sales team and the customer's sales team.

A company with a good sales team buying in large enough quantities can likely significantly undercut that. Smaller quantities and an inept sales team will lead to a company paying what is listed on that page.

Just as the page says, it's the "recommended customer price." It's what Intel would like to get, but isn't necessarily what a customer would pay.

Without being privy to the confidential sales talks between Intel and it's customers, it's impossible to say what those go for on average or even what the lowest price being paid for them is. And those talks are confidential for a reason. Not only does Intel not want customers to know what other customer's are actually paying, other customers don't want their competition to know. Especially if one customer feels they are getting a better deal than their competition. And companies always feel like they are getting at least a somewhat good deal if they aren't paying the "recommended customer price."

Regards,
SB
http://www.theatlantic.com/technolo...but-it-could-still-win-the-mobile-war/275825/

As a interesting slightly related historical note, Otellini is revealing that after Apple switched their Macs to Intel x86 CPUs, Intel was presented with the opportunity to provide the CPU for what would become the iPhone. Apparently, Apple wanted a price per chip that was too low for Intel based on what Intel estimated the development/production costs would be and the volume they thought Apple would buy. As it turned out they over-estimated the forecasted costs and underestimated how well the iPhone would sell and Otellini now regrets not going with his gut and saying yes to Apple.

In terms of overall semiconductor market size, he mentions that economies of scale works out that 100 million units results in a $1000 average chip price while 1 billion units yields a $100 price and since 2005 they've been working on figuring out how to align the company to face a multi-billion unit market with a target $10 chip price.

Will Intel let companies design their own SoC with Silvermont?

Will Intel guarantee they will never prioritirize a company for chips delivery? If not will they let a company bring Silvermont to another foundry?

Will Intel let a company makes its own x86 CPU to increase differentiation against competition?

Will Intel design ultra low cost x86 CPU for lower end smartphones so that a phone maker can have the same software for all of its phones?

I think Intel will have to heavily change its business model to make a significant inroad in the phone market. Being the best technically is very far from enough.
On this issue Otellini says Intel isn't interested in directly competing against ARM, because they aren't interested in licensing their architecture. Instead they will focus building a better chip than the individual ARM implementations that Qualcomm, nVidia, and TI offer. I suppose that means they will continue to push technical superiority as their major selling point rather than address the flexibility that the ARM ecosystem offers. He mentions wanting to focus on building the best chip for Apple so Apple would want to buy it, but it's unclear whether he actually means he's willing to have the company design custom chips for specific customers or just have their standard chip be so good in general that customers can't refuse.
 
IMHO as much as there are chances that Intel might have CPUs on its roadmap someone like Apple would have a hard time to refuse (for iOS), as much could Apple's engineering have the chance to design a custom CPU that wouldn't have that much to envy against Intel's proposition. Apple's has its own OS and they only really know what their sw ecosystem roadmap includes for the future.
 
As good as Silvermont may be I don't think Intel is really going to provide a chip so good that Apple can't refuse it. I know people like to bring up the old PowerPC transition but that was totally different. Apple was a very different company back then, and from their point of view they had to get rid of a dead end supplier that wasn't at all equipped to cater to their market depends. The move was to another supplier which could already leverage a huge volume advantage.

Apple didn't spend hundreds of millions of dollars to gain the ability to develop world class CPU cores and no matter what anonymous CPUs were on Intel's marketing slides I doubt any of them was Swift.
 
By the way I recall a quite different story why Apple initially didn't adopt Atom:

http://www.xbitlabs.com/news/multim...Was_Supposed_to_Be_Intel_Atom_Based_Book.html

Apparently, Steve Jobs was pretty pleased with the result and even told his biographer that Intel was anyway too slow and not too flexible to offer something comparable.

"At the high-performance end, Intel is the best. They build the fastest chip, if you don't care about the power and cost. But they build just the processor on one chip, so it takes a lot of other parts. Our A4 has the processor and the graphics, mobile operating system, and memory control all in one chip. We tried to help Intel, but they do not listen much. We have been telling them for years that their graphics suck. Every quarter we schedule a meeting with me and our top three guys and Paul Otellini. At the beginning, we were doing wonderful things together. They wanted this big joint project to do chips for future iPhones. There were two reasons we did not go with them. One was that they are just really slow. They are like a steamship, not very flexible. We are used to going pretty fast. Second is that we just did not want to teach them everything, which they could go and sell to our competitors," Steve Jobs told his biographer.

There's more in that write-up to the topic, but Apple has so far set its priorities to the graphics part of their SoCs; if Apple wanted to mimic Tegras they could have done so a long time ago. And before anyone says it any of past Intel's proposition would again had included IMG GPU IP; just not as high end as Apple probably would had wanted.
 
By the way I recall a quite different story why Apple initially didn't adopt Atom:

http://www.xbitlabs.com/news/multim...Was_Supposed_to_Be_Intel_Atom_Based_Book.html

There's more in that write-up to the topic, but Apple has so far set its priorities to the graphics part of their SoCs; if Apple wanted to mimic Tegras they could have done so a long time ago. And before anyone says it any of past Intel's proposition would again had included IMG GPU IP; just not as high end as Apple probably would had wanted.
Well if the proposal was that Intel make Apple a custom chip, Otellini's contention that the deal fell through because of projected high costs and low volumes and Jobs' contention that it was Intel's inflexibility and potential technology leakage to the competition may not necessarily be at odds.
 
Well if the proposal was that Intel make Apple a custom chip, Otellini's contention that the deal fell through because of projected high costs and low volumes and Jobs' contention that it was Intel's inflexibility and potential technology leakage to the competition may not necessarily be at odds.

Both make sense as respective perspectives from each side; however IMHO Apple wanted and needed something unique. Apple developed the A4 and are evolving it step by step with each generation. The point here is - as Exophase also pointed out - why would Apple spent ungodly amounts of resources and time in order to design and develop their own SoCs and suddently dump it all for an Intel solution. Silverthorne is fine and dandy as a CPU - on paper at least; but what about SFF mobile GPUs? Want to bet that Apple will contiue to be in the forefront in terms of graphics performance compared to Intel's solutions with succeeding generations of SoCs?
 
Whatever the case, negotiations in the distant past of 2004 wouldn't have much bearing on negotiations that'd be made today. That was long before even the first iPhone. I wonder if Apple of that time even had ambitions of designing their own SoCs, let alone CPUs..
 
Before the iPhone launch I severely doubt it; but once the i-story started taking off definitely, otherwise they wouldn't had bought PA Semi f.e. that early.
 
I wonder if Apple of that time even had ambitions of designing their own SoCs, let alone CPUs..

I think the success of iPhone exceeded anybody's expectations, even Apple's.

When the success did materialize, they were very quick to shift to a vertical business structure for the product. Picking up PA Semi was instrumental, they got a world class CPU design team, one that is better at designing high performance CPUs with a mobile power envelope than ARM themselves.

Cheers
 
Before the iPhone launch I severely doubt it; but once the i-story started taking off definitely, otherwise they wouldn't had bought PA Semi f.e. that early.
They were hiring chip design engineers before the iPhone launch. I think it was always their intention to eventually go vertical.
 
So I don't know for sure what AnTuTu does but I doubt they wrote their own benchmark.
Found out that the mem/FP/int parts are using nbench.
Code:
00047d10 T DoNumSort
00047e80 T DoStringSort
00047fa8 T DoBitops
00048130 T DoEmFloat
00048308 T DoFourier
000484a0 T DoAssign
00048650 T DoIDEA
00048970 T DoHuffman
00048c18 T DoNNET
00048f48 T DoLU
 
Very interesting move, considering that not all Android Apps can run on x86. This is either Samsung's #1 or #2 tablet, so it's a significant risk.

My theory? Samsung wants a little experience making a high volume, low cost x86 tablet so that they get Silvermont right. Presumably they'll be able to share a lot of design work and parts between the Android and Win8 versions.
 
Back
Top