I am not sure I care. But what I WOULD rather see, is accelerated production of iPhones to release them prior to Independence Day, as was done in many years past. There is a benefit to Apple to release the newer iPhone prior to this holiday in the US because that is a time when many families get together and show and tell. Plus September just seems too late in the year for new iPhone reveals, plus there would be less end of the year shortages if production ramps earlier.
@leavingthebigG : TSMC never sued Samsung. TSMC sued a former TSMC employee, Liang Mong-song, who left the company almost 6 years ago for his breaching of noncompete agreement. If you believe that TSMC had a working 14nm processing back 2009, sure, Samsung did steal TSMC's tech. I'm, however, inclined to agree with ExtremeTech's Joel Hruska's take on the charges:
and, being that CommonWealth is a Taiwanese trade mag, I'm not too surprised by their spin on this. I suspect that it's all part of the campaign orchestrated by TSMC, Foxconn, and other competing Taiwanese interests.
@tooltalk thanks for the correction. TMSC did not sue Samsung as I had stated. After reading an article to remind myself of the "facts", I still hope TSMC wins the lawsuit in an effort to put fear into others who might choose to steal TSMC IP then sell it to Samsung or any other company.
Again, no matter what Ming Chi says, you can not make the same SOC on two difference processes. Apple will choose one or the other, not both.
Agree, also the two parts will run at different speed and when people benchmark these differences will show up and you know what the media will do with that.
Apple has more money than Croesus. They can buy forests, create dozens of server and solar farms, but cannot fab their own silicon. It can't be for lack of money or brains. Would rather forego my dividends than see Apple held hostage by the vicissitudes of this industry.
Quote:
Originally Posted by SpamSandwich
Only if Apple also sold to competitors would that remotely make sense.
SS I totally agree! Obviously and understandably RH doesn't have quite a grasp on how much it would cost to develop, build and maintain ongoing state-of-the -art fab facilities, especially for a single customer.
@tooltalk thanks for the correction. TMSC did not sue Samsung as I had stated. After reading an article to remind myself of the "facts", I still hope TSMC wins the lawsuit in an effort to put fear into others who might choose to steal TSMC IP then sell it to Samsung or any other company.
@leavingthebigG : this particular non-compete case has little to do with IP theft -- none accused or proven so far -- and, as many in the industry noted, it's somewhat laughable to claim the one single former employee, who left the company 6 years ago, accused of breaching his noncompete had all to do Samsung's 14nm process today. Further, I'm not generally too crazy about any restriction on hiring and in many places it's very difficult to enforce something like that.
I'm just pointing out that this is a sad example of the desperate PR campaign waged by Samsung's competitors in Taiwan like TSMC and Foxconn as explained in this article.
Again, no matter what Ming Chi says, you can not make the same SOC on two difference processes. Apple will choose one or the other, not both.
You realistically can't say this. First; it is very possible to deliver functional identical parts which is all that matters to Apple. Sure there will be physical difference due to the process so the chips aren't the "same" but from the users standpoint they won't be able to tell the difference.
Second; Apples volumes are huge! This means the economics that many companies have to live with don't impact Apple. Apple is approaching a 100 million chips per quarter, a number that would tax many companies foundries. Apples volume dictates sharing production across different facilities or even vendors.
Beyond all of that have we even seen a definitive references that tells us any of this is true? Apple could go with global foundries for all we know or even somebody else.
I agree. I actually believe that both had tapeouts prior to any production decision, and when you think about the volumes over two years of production, you might be looking at 300 million A9's, so a split makes sense from the start.
I really don't think people take volumes into consideration here. Even your numbers seem low. If you factor in all Apple devices using A series chips at one time, Apple is getting close to shipping 100 million A series chips per quarter. Of course that doesn't represent all A9's later this year but the mix could tip in the direction of more advanced chips later this year. That would be due to a new Apple TV, updated iPhones and updated iPad Mini, iPad and so forth. In other words by late 2016 I would expect many new Apple products running on 64 bit chips of recent design.
Samsung made the first 64bit A-series processor (iPhone 5S) and they still only have a half baked 64bit processor in their own phones (a reference design) two years later. I doubt anything they learned from making the 64 bit A-series processors would be able to help them without a complete redesign of the Arm reference designs as Apple id.
The A-series designs are so far ahead of the Arm reference designs that by the time Samsung could reverse engineered it and put something together Apple would be 2-3 generations beyond that... They just can't compete in this market so they stick with what all the other Droid's are doing.
The Exynos 7420 is anything but 'half baked'. There are rumours that Samsung are working on custom cores for their next effort and have already achieved a 45% improvement over the stock ARM A57.
The Exynos 7420 is anything but 'half baked'. There are rumours that Samsung are working on custom cores for their next effort and have already achieved a 45% improvement over the stock ARM A57.
You need to re-read the post you were responding to. The 7420 is half baked because it is still an ARM reference design, even though tweaked.
Perhaps the following generation will be "fully baked"; i.e., an ARM Architectural license and a full custom design, like Apple has been doing for a couple of generations of A Series.
Not impressed by "45% improvement over stock" until there is shipped product, but good for Samsung if they succeed, whatever the clock they need to run it at to get there.
You need to re-read the post you were responding to. The 7420 is half baked because it is still an ARM reference design, even though tweaked.
Perhaps the following generation will be "fully baked"; i.e., an ARM Architectural license and a full custom design, like Apple has been doing for a couple of generations of A Series.
Not impressed by "45% improvement over stock" until there is shipped product, but good for Samsung if they succeed, whatever the clock they need to run it at to get there.
The term half-baked implies something half finished. If the OP, meant what you state, then you are both applying an arbitrary and meaningless definition as to what constitutes an adequately functional processor. There seems to be an implication that the stock Arm Big Little design Samsung is using is somehow inadequate. That seems to be somewhat of an odd notion given that in Anandtech's benchmarking of the S6 Edge it bettered an iPhone 6 model in 16 out of 22 benchmarks.
This should be based on cost and performance, instead of Terry Guo being a coward he should instead encourage creativity and improve on better designs and performance, it is a global economy and compitiendo market place creates innovation and better products.Terry by telling and pursuing companies to only use Taiwanese services or products even if inferior only hurts Taiwan and consumers, capitalism works because of fair competition.
Court papers have proven that Samsung does in fact cross lines and Samsung publicly stating it will shield its subsidiaries from other divisions as a result proved they were doing it. Samsung has zero cred.
Not sure what to make of the TSMC quote about it pushing Apple to choose them. Apple doesn't get "pushed" by anyone. It does the pushing. Sounds more like them doing a self-patting on the back.
Maybe 'pushing' was really 'asked nicely' and the distinction lost lost in translation ...
Comments
@leavingthebigG : TSMC never sued Samsung. TSMC sued a former TSMC employee, Liang Mong-song, who left the company almost 6 years ago for his breaching of noncompete agreement. If you believe that TSMC had a working 14nm processing back 2009, sure, Samsung did steal TSMC's tech. I'm, however, inclined to agree with ExtremeTech's Joel Hruska's take on the charges:
and, being that CommonWealth is a Taiwanese trade mag, I'm not too surprised by their spin on this. I suspect that it's all part of the campaign orchestrated by TSMC, Foxconn, and other competing Taiwanese interests.
@tooltalk thanks for the correction. TMSC did not sue Samsung as I had stated. After reading an article to remind myself of the "facts", I still hope TSMC wins the lawsuit in an effort to put fear into others who might choose to steal TSMC IP then sell it to Samsung or any other company.
Agree, also the two parts will run at different speed and when people benchmark these differences will show up and you know what the media will do with that.
Apple has more money than Croesus. They can buy forests, create dozens of server and solar farms, but cannot fab their own silicon. It can't be for lack of money or brains. Would rather forego my dividends than see Apple held hostage by the vicissitudes of this industry.
Only if Apple also sold to competitors would that remotely make sense.
SS I totally agree! Obviously and understandably RH doesn't have quite a grasp on how much it would cost to develop, build and maintain ongoing state-of-the -art fab facilities, especially for a single customer.
@tooltalk thanks for the correction. TMSC did not sue Samsung as I had stated. After reading an article to remind myself of the "facts", I still hope TSMC wins the lawsuit in an effort to put fear into others who might choose to steal TSMC IP then sell it to Samsung or any other company.
@leavingthebigG : this particular non-compete case has little to do with IP theft -- none accused or proven so far -- and, as many in the industry noted, it's somewhat laughable to claim the one single former employee, who left the company 6 years ago, accused of breaching his noncompete had all to do Samsung's 14nm process today. Further, I'm not generally too crazy about any restriction on hiring and in many places it's very difficult to enforce something like that.
I'm just pointing out that this is a sad example of the desperate PR campaign waged by Samsung's competitors in Taiwan like TSMC and Foxconn as explained in this article.
You realistically can't say this. First; it is very possible to deliver functional identical parts which is all that matters to Apple. Sure there will be physical difference due to the process so the chips aren't the "same" but from the users standpoint they won't be able to tell the difference.
Second; Apples volumes are huge! This means the economics that many companies have to live with don't impact Apple. Apple is approaching a 100 million chips per quarter, a number that would tax many companies foundries. Apples volume dictates sharing production across different facilities or even vendors.
Beyond all of that have we even seen a definitive references that tells us any of this is true? Apple could go with global foundries for all we know or even somebody else.
I really don't think people take volumes into consideration here. Even your numbers seem low. If you factor in all Apple devices using A series chips at one time, Apple is getting close to shipping 100 million A series chips per quarter. Of course that doesn't represent all A9's later this year but the mix could tip in the direction of more advanced chips later this year. That would be due to a new Apple TV, updated iPhones and updated iPad Mini, iPad and so forth. In other words by late 2016 I would expect many new Apple products running on 64 bit chips of recent design.
Samsung made the first 64bit A-series processor (iPhone 5S) and they still only have a half baked 64bit processor in their own phones (a reference design) two years later. I doubt anything they learned from making the 64 bit A-series processors would be able to help them without a complete redesign of the Arm reference designs as Apple id.
The A-series designs are so far ahead of the Arm reference designs that by the time Samsung could reverse engineered it and put something together Apple would be 2-3 generations beyond that... They just can't compete in this market so they stick with what all the other Droid's are doing.
The Exynos 7420 is anything but 'half baked'. There are rumours that Samsung are working on custom cores for their next effort and have already achieved a 45% improvement over the stock ARM A57.
The Exynos 7420 is anything but 'half baked'. There are rumours that Samsung are working on custom cores for their next effort and have already achieved a 45% improvement over the stock ARM A57.
You need to re-read the post you were responding to. The 7420 is half baked because it is still an ARM reference design, even though tweaked.
Perhaps the following generation will be "fully baked"; i.e., an ARM Architectural license and a full custom design, like Apple has been doing for a couple of generations of A Series.
Not impressed by "45% improvement over stock" until there is shipped product, but good for Samsung if they succeed, whatever the clock they need to run it at to get there.
You need to re-read the post you were responding to. The 7420 is half baked because it is still an ARM reference design, even though tweaked.
Perhaps the following generation will be "fully baked"; i.e., an ARM Architectural license and a full custom design, like Apple has been doing for a couple of generations of A Series.
Not impressed by "45% improvement over stock" until there is shipped product, but good for Samsung if they succeed, whatever the clock they need to run it at to get there.
The term half-baked implies something half finished. If the OP, meant what you state, then you are both applying an arbitrary and meaningless definition as to what constitutes an adequately functional processor. There seems to be an implication that the stock Arm Big Little design Samsung is using is somehow inadequate. That seems to be somewhat of an odd notion given that in Anandtech's benchmarking of the S6 Edge it bettered an iPhone 6 model in 16 out of 22 benchmarks.
Maybe 'pushing' was really 'asked nicely' and the distinction lost lost in translation ...