You do realize that the 3GS used a Cortex A8 right? And it was what? The 2nd phone to use it? Apple doesn't lag much.
The Cortex A9 wasn't ready for early 2010/late 2009 so the A4 had to be a Cortex A8. Do you REALLY think Apple has been ignoring the Cortex A9 for a year? WTF do you think the Intrinsity folks have been doing since April? The PA Semi folks have been with Apple even longer. Do you seriously believe Apple hasn't been the first to sample the Orion?
The timing is just right for the Cortex A9 based A5 (or whatever) to appear in time for the iPad 2 much less for the iPhone 5. Everyone has been sampling for months and getting set to start production early 2011.
That's what I am saying --- Samsung produced a cortex a8 chip a year ahead of Apple.
The dual core cortex a9 isn't even ready yet ---- RIGHT NOW --- from the big silicon companies. You are assigning mythical abilities to the PA Semi and Intrinsity teams.
I'm sorry but that line of argument is nonsense. What magically different things are people doing on an iPhone vs. an iPad? They're still playing games, consuming media, browsing the web, etc.
Barely more functional than a phone? A 7" tablet has roughly 4.5 times as much screen real estate than the iPhone. Maybe you lack the imagination to figure out to make proper use of that space, but others probably can. Browsing the web, watching videos, editing documents and more are all going to be better overall experiences on a tablet rather than a phone.
And having read your paranoid ramblings about Google I really don't thing you should be complaining to anyone about their "lack of sound reasoning" and their "emotional bias."
Size doesn't matter because you're doing the same stuff, but size makes all the difference. It's stupid to imagine that more space makes a qualitative difference in the experience, but lack imagination if you can't see how that extra space can be used.
Here, let me help:
Quote:
Barely more functional than a7" tablet? The iPad has more than twice as much screen real estate than a 7" tablet. Maybe you lack the imagination to figure out to make proper use of that space, but others probably can. Browsing the web, watching videos, editing documents and more are all going to be better overall experiences on an iPad rather than a 7" tablet
That's what I am saying --- Samsung produced a cortex a8 chip a year ahead of Apple.
The dual core cortex a9 isn't even ready yet ---- RIGHT NOW --- from the big silicon companies. You are assigning mythical abilities to the PA Semi and Intrinsity teams.
Yes, it's not in production RIGHT NOW. The iPhone 5 isn't launching RIGHT NOW but in 8 months.
The iPad 2 is likely to be March or April. 4-5 months from now.
This isn't mythical abilities for Apple but fairly reasonable timeframes given where the A9s are now...in prototype units like the LG phone, Playbook tablet, etc and very close to full rate production.
There are prototype dual core Cortex A9 iPads in Cupertino RIGHT NOW. There are prototype dual core Cortex A9 iPhone 5s in Cupertino RIGHT NOW. There are dual core Cortex A9 based A5 samples in Cupertino right now.
Will Apple decide to go with single core A5s over duals for battery life? Maybe, but strikes me as doubtful.
Yes, it's not in production RIGHT NOW. The iPhone 5 isn't launching RIGHT NOW but in 8 months.
The iPad 2 is likely to be March or April. 4-5 months from now.
This isn't mythical abilities for Apple but fairly reasonable timeframes given where the A9s are now...in prototype units like the LG phone, Playbook tablet, etc and very close to full rate production.
There are prototype dual core Cortex A9 iPads in Cupertino RIGHT NOW. There are prototype dual core Cortex A9 iPhone 5s in Cupertino RIGHT NOW. There are dual core Cortex A9 based A5 samples in Cupertino right now.
Will Apple decide to go with single core A5s over duals for battery life? Maybe, but strikes me as doubtful.
So you are telling me that the time lag between a big silicon company like TI vs. Apple is less than 2 months --- because the earliest time RIM launches the Playbook seems to be in February.
If Apple launches a dual core cortex A9 ipad 2 in April 2011, they are buying a stock CPU from Samsung because I don't think that Apple can do it within 2 months of other big silicon companies.
So you are telling me that the time lag between a big silicon company like TI vs. Apple is less than 2 months --- because the earliest time RIM launches the Playbook seems to be in February.
Yes, that's exactly what I am saying.
Apple's ability to execute is not tied to RIM's ability to execute. Thank god.
There is no reason they cannot have a dual core Cortex A9 based A5 design fabbed by Samsung and sampling internally as we speak (type). The S5PC100 in the 3GS was custom for Apple. The A8 in Hummingbird and the A4 is a custom version of that designed by Intrinsity and fabbed by Samsung.
Samsung and Apple have a very close relationship and there's no reason that Apple can't have been working on the A5 in parallel with Orion development. There's no need for things to have to happen sequentially.
So there's no reason to doubt that Apple couldn't lag TI by 2 months so long as Samsung doesn't lag TI by more than 2 months.
If I'm reading ARM's product page correctly it looks like there will also be significant space savings with the single-core A9.
Given Orion is 45nm this is a distinct possibility. Folks had expected 32nm but not so much so yah...2 core bigger than 1 core.
As an aside, TI isn't going past the 45nm process so effectively they become a fabless designer for next generation ARMs just like Apple. Probably end up using TSMC.
Apple's ability to execute is not tied to RIM's ability to execute. Thank god.
There is no reason they cannot have a dual core Cortex A9 based A5 design fabbed by Samsung and sampling internally as we speak (type). The S5PC100 in the 3GS was custom for Apple. The A8 in Hummingbird and the A4 is a custom version of that designed by Intrinsity and fabbed by Samsung.
Samsung and Apple have a very close relationship and there's no reason that Apple can't have been working on the A5 in parallel with Orion development. There's no need for things to have to happen sequentially.
So there's no reason to doubt that Apple couldn't lag TI by 2 months so long as Samsung doesn't lag TI by more than 2 months.
I am not talking about RIM, I am talking about TI and Samsung.
There is no reason to suspect that Samsung is going to do a custom Apple job ahead of Samsung's own chip --- and we haven't seen Samsung announcing that they are anything close to shipping their own dual core Cortex A9 chip yet.
And there is no reason to believe that big silicon companies like TI are shipping massive amounts of dual core cortex A9 chips either. Wall Street ain't estimating RIM to ship millions and millions of Playbook in Q1 --- they are estimating the figure to be in hundreds of thousands.
So you are telling me that somehow Samsung is going to ship a custom Apple job for millions and millions units --- ahead of their own Samsung chip (which won't be available in massive quantity until first half of next year).
I am not talking about RIM, I am talking about TI and Samsung.
There is no reason to suspect that Samsung is going to do a custom Apple job ahead of Samsung's own chip --- and we haven't seen Samsung announcing that they are anything close to shipping their own dual core Cortex A9 chip yet.
And there is no reason to believe that big silicon companies like TI are shipping massive amounts of dual core cortex A9 chips either. Wall Street ain't estimating RIM to ship millions and millions of Playbook in Q1 --- they are estimating the figure to be in hundreds of thousands.
So you are telling me that somehow Samsung is going to ship a custom Apple job for millions and millions units --- ahead of their own Samsung chip (which won't be available in massive quantity until first half of next year).
Think about it!
Say you are Samsung.
You have an internal demand for, say, 100 K A9s per month (0% profit).
You have an external demand for, say, 300 K A9s per month (20% profit).
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance.
Just who do you want to give priority -- especially since Apple can go to any other Fabs and make the same deal?
There's an old (supplier) saying from the days when IBM dominated the mainframe computer market with 97%:
When you make love with an 800 lb gorilla -- you stop when the Gorilla gets tired!
I'm sorry but that line of argument is nonsense. What magically different things are people doing on an iPhone vs. an iPad? They're still playing games, consuming media, browsing the web, etc.
Okay so you claim that the iPad isn't better because its bigger.
Quote:
Barely more functional than a phone? A 7" tablet has roughly 4.5 times as much screen real estate than the iPhone. Maybe you lack the imagination to figure out to make proper use of that space, but others probably can. Browsing the web, watching videos, editing documents and more are all going to be better overall experiences on a tablet rather than a phone.
But then you claim a 7" tablet is better than a phone because its bigger.
Oops!
Fact is, the size differential between iPhone and iPad is significant meaning that the capability differential is also significant. The size different between iPhone and Playbook on the other hand is minimal meaning the expected capability differential is also minimal.
Of course, if you consider the capabilities of a current Blackberry as the baseline then yes of course the Playbook will be exponentially more capable simply because Blackberries are so incapable.
You have an internal demand for, say, 100 K A9s per month (0% profit).
You have an external demand for, say, 300 K A9s per month (20% profit).
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance.
Just who do you want to give priority -- especially since Apple can go to any other Fabs and make the same deal?
You reached the same conclusion as me --- that's why I said that if Apple is coming out with a dual core a9 ipad2 in Q2, it's going to be a Samsung stock CPU.
What nht was talking about was a custom Apple dual core A5 chip coming out before a Samsung stock dual core A9 chip.
You reached the same conclusion as me --- that's why I said that if Apple is coming out with a dual core a9 ipad2 in Q2, it's going to be a Samsung stock CPU.
What nht was talking about was a custom Apple dual core A5 chip coming out before a Samsung stock dual core A9 chip.
No, I believe the A5 design is done. Samsung will give priority to manufacture the A5.
Here's some questions about state of the art SOC manufacturing:
Ages ago, I worked in the semiconductor industry.
In those days they would build chips to mil-qual specs for the military.
After manufacturing, the chips went through rigorous testing to assure that they performed to specs -- things like performing within a certain range under varying power, temperature, etc.
Those that passed testing went to fill orders.
Those that failed testing were either total rejects (scrapped) or marginal.
The marginal chips were often put in inventory where they could be used as less- critical spec'd parts (lower power, narrower temperature range, etc.)
So, with something like the A5:
1) Can they target manufacturing of say, a 1 GHz Dual-Core A9?
2) Can chips that don't meet those specs be used as a lesser part. e.g. a single core variant and/or a slower-clocked variant?
3) If so, do they design extra capacity into the chips to increase their yield?
Say Apple decides that for 2011 they want the following chips:
iPad - 1 GHz Dual-Core A9
iPhone - 1 GHz Single-Core A9
iPod Touch - 800 MHz Dual-Core A9
AppleTV - 800 MHz Single-Core A9
4) Would they manufacture a single chip targeted at the highest spec, then fill the lower-speced by under-clocking and/or disabling a core. etc."
5) If so, what other things can they over sped, then re-spec?
No, I believe the A5 design is done. Samsung will give priority to manufacture the A5.
Why should Samsung give priority to Apple --- when Samsung also manufactures smartphones and tablets themselves, which has a higher profit margin because it's a finished product. Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple.
You have an internal demand for, say, 100 K A9s per month (0% profit).
You have an external demand for, say, 300 K A9s per month (20% profit).
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance.
Just who do you want to give priority -- especially since Apple can go to any other Fabs and make the same deal?
There's an old (supplier) saying from the days when IBM dominated the mainframe computer market with 97%:
When you make love with an 800 lb gorilla -- you stop when the Gorilla gets tired!
.
Quote:
Originally Posted by samab
Why should Samsung give priority to Apple --- when Samsung also manufactures smartphones and tablets themselves, which has a higher profit margin because it's a finished product. Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple.
Again, lets assume that you are Samsung -- I am pulling numbers out of the air:
1) You plan to set up a manufacturing line/process for ARM Dual-Core A9
2) You Budget, say, $250 Million to set up and test the line over a period of 3 months.
3) You amortize these costs over, say, a 3-year expected life of the ARM Dual-Core A9
4) Over those 3 years you plan to sell 300 Million units at a profit of $10 per unit -- a very attractive 30% gross profit margin .
5) Just for ease of calculation, the production starts in Jan 2011
Now, using the numbers from my earlier post:
You have an internal demand for, say, 100 K A9s per month (0% profit). These are the units that Samusing sells to itself for use in Galaxy Tab follow-ons. There is no profit on these chips until they are included in the manufacture of the tablets, and subsequently sold at a profit. But for sake of argument lets say Samsung can sell 100 K Samsung Galaxy Tab2 at full retail, realizing the $10 profit on each ARM Dual-Core A9 included.
So, Samsung will make $1 million per month on Samsung chips in Samsung Tablets
You have an external demand for, say, 300 K A9s per month (@20% profit) or $6.70. Here Samsung sells the chips to others at 2/3 the profit realized by selling to Samsung. There is immediate (30 days later) profit.
So, Samsung will make 300 K x $6.70 ~= $2 million per month on Samsung chips sold to others.
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance. Sales of chips to Apple are at 15% margin vs internal 30% margin, or $5 profit per chip.
So, Samsung will make 2 million x $5 == $10 Million per month on Samsung chips sold to Apple.
Putting these together:
Samsung will make 100 K x $10.00 == $1 million per month on Samsung chips in Samsung Tablets
Samsung will make 300 K x $06.70 ~= $2 million per month on Samsung chips sold to others.
Samsung will make 2 mill x $05.00 == $10 Million per month on Samsung chips sold to Apple.
Noteworthy:
There is a thing called the time/value of money
Apple pays Samsung in advance.
Apple guarantees sales for 6 months
Apple reserves certain rights to increase quantities/orders at these prices
Apple, essentially, pays for Samsung to setup the $250 Million manufacturing line/process (the A5 is a superset of the Samsung chip)
Apple can go to another Fab (a Samsung competitor) and make the same offer -- reducing Samsung's economy of scale
A bird in the hand is better than a turd in the bush!
To answer your question:
"Why should Samsung give priority to Apple -- Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple."
If Samsung can make $100 (including $10 on the A9) each and sell 1 million Tabs in 6 months thats $100 million profit ($10 Million on the A9) possible
Samsung can make $5 each and sell 12 million A5s in 6 months that's $60 million profit guaranteed
More profit, Less risk, better use of $ (payment in advance vs manufacturing and inventory costs).
That's what I am saying --- Samsung produced a cortex a8 chip a year ahead of Apple.
The dual core cortex a9 isn't even ready yet ---- RIGHT NOW --- from the big silicon companies. You are assigning mythical abilities to the PA Semi and Intrinsity teams.
No, you are just arguing shiite for the sake of trying to cover your ass. CortexA9 has been licensable as IP from ARM since early 2009. Tegra2 dual core shipped in 2010, although not power optimized as it came from Nvidia, it shows the maturity of IP to tape-out is well past question.
I wouldn't state my reputation in TI's ability to churn out other folks CPU IP, that's not their strength. But you just go ahead, because your technical reputation is already damaged beyond repair. Do you actually know facts or check them at all before you post? Because it sure doesn't read that way.
And Samsung didn't announce an A8 phone until Feb 2009, which is only a couple months before iPhone 3GS shipped. And Samsung produced that version of the A8 as a Samsung part number. Apple later made the A4 based on further CortexA8 improvements.
There has been plenty of time for Apple to do an CortexA9 version if they thought it was economically smart to ship. That's what Apple will use as criteria for inclusion, not spec-whoring.
Why should Samsung give priority to Apple --- when Samsung also manufactures smartphones and tablets themselves, which has a higher profit margin because it's a finished product. Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple.
Because every A4/A5 sold is finance risk-free. Samsung has no R&D skin in the A4/A5 game, and it keeps their contract fab busy. The Samsung contract fab CAN'T produce Galaxy Tabs, it can only provide chips, and Samsung uses the contracts to pay off the capital expenditure of building and upgrading the fab. That way they get their own Cortex CPUs without having to bear all the manufacturing financial risk themselves on top of the R&D risk inherent in implementing the Cortex IP on silicon. Flat out Samsung needs Apple contract business more than Apple needs Samsung specifically because TSMC could do the job nicely too.
The profit on the contract run chips is many times higher compared to in-house taped and used ARM Cortex chips because of all of the above. I'll even sat Dick's numbers on internal cost are skewed in your favor because he missed accounting for the fact Samsung has no R&D investment to recoup for A4/A5 before they can recognize any profit. And that skewing still didn't look good for you, your position is even worse than you were led to believe the the first thrashing.
So you have it exactly backwards as far as fabbing chips go. It's obvious you have no idea how the industry works, at all.
Because every A4/A5 sold is finance risk-free. Samsung has no R&D skin in the A4/A5 game, and it keeps their contract fab busy. The Samsung contract fab CAN'T produce Galaxy Tabs, it can only provide chips, and Samsung uses the contracts to pay off the capital expenditure of building and upgrading the fab. That way they get their own Cortex CPUs without having to bear all the manufacturing financial risk themselves on top of the R&D risk inherent in implementing the Cortex IP on silicon. Flat out Samsung needs Apple contract business more than Apple needs Samsung specifically because TSMC could do the job nicely too.
The profit on the contract run chips is many times higher compared to in-house taped and used ARM Cortex chips because of all of the above. I'll even sat Dick's numbers on internal cost are skewed in your favor because he missed accounting for the fact Samsung has no R&D investment to recoup for A4/A5 before they can recognize any profit. And that skewing still didn't look good for you, your position is even worse than you were led to believe the the first thrashing.
So you have it exactly backwards as far as fabbing chips go. It's obvious you have no idea how the industry works, at all.
Rightly or wrongly --- Korean and Japanese conglomerates don't care about short term profitability --- between cross shareholdings and government pushing technology agendas, satisfying Apple's needs are going to be a low priority.
I am not saying that's the right thing for Samsung to do financially, I am saying that's what they would do --- given how the Korean and Japanese conglomerates operate. Hell, after 20 decades of economic mess, the Japanese conglomerates would still do that.
Rightly or wrongly --- Korean and Japanese conglomerates don't care about short term profitability --- between cross shareholdings and government pushing technology agendas, satisfying Apple's needs are going to be a low priority.
This is called saying even more stupid things in the hopes that folks forget the stupid things already stated. This is why Hiro says your technical reputation is beyond repair.
Quote:
I am not saying that's the right thing for Samsung to do financially, I am saying that's what they would do --- given how the Korean and Japanese conglomerates operate. Hell, after 20 decades of economic mess, the Japanese conglomerates would still do that.
Given that Samsung gave Apple A4 production equal if not higher priority over Hummingbird used in their own flagship products this is demonstrably false.
Apple had millions of A4s produced for them and launched the iPad in April where Samsung's flagship Galaxy S using Hummingbird didn't launch until June. Apple announced in Jan, took pre-orders on March 12 before Samsung even announced the S (March 23).
Given which product came out first and the qty of iPads built it's fairly easy to make the case that Samsung gave Apple A4 production priority over its own.
This is called saying even more stupid things in the hopes that folks forget the stupid things already stated. This is why Hiro says your technical reputation is beyond repair.
Given that Samsung gave Apple A4 production equal if not higher priority over Hummingbird used in their own flagship products this is demonstrably false.
Apple had millions of A4s produced for them and launched the iPad in April where Samsung's flagship Galaxy S using Hummingbird didn't launch until June. Apple announced in Jan, took pre-orders on March 12 before Samsung even announced the S (March 23).
Given which product came out first and the qty of iPads built it's fairly easy to make the case that Samsung gave Apple A4 production priority over its own.
Given that stock cortex a8 chips were available one year before both Samsung and Apple released their version, you are saying that Samsung and Apple have caught up with the big silicon companies in one year's time.
Comments
You do realize that the 3GS used a Cortex A8 right? And it was what? The 2nd phone to use it? Apple doesn't lag much.
The Cortex A9 wasn't ready for early 2010/late 2009 so the A4 had to be a Cortex A8. Do you REALLY think Apple has been ignoring the Cortex A9 for a year? WTF do you think the Intrinsity folks have been doing since April? The PA Semi folks have been with Apple even longer. Do you seriously believe Apple hasn't been the first to sample the Orion?
The timing is just right for the Cortex A9 based A5 (or whatever) to appear in time for the iPad 2 much less for the iPhone 5. Everyone has been sampling for months and getting set to start production early 2011.
That's what I am saying --- Samsung produced a cortex a8 chip a year ahead of Apple.
The dual core cortex a9 isn't even ready yet ---- RIGHT NOW --- from the big silicon companies. You are assigning mythical abilities to the PA Semi and Intrinsity teams.
I'm sorry but that line of argument is nonsense. What magically different things are people doing on an iPhone vs. an iPad? They're still playing games, consuming media, browsing the web, etc.
Barely more functional than a phone? A 7" tablet has roughly 4.5 times as much screen real estate than the iPhone. Maybe you lack the imagination to figure out to make proper use of that space, but others probably can. Browsing the web, watching videos, editing documents and more are all going to be better overall experiences on a tablet rather than a phone.
And having read your paranoid ramblings about Google I really don't thing you should be complaining to anyone about their "lack of sound reasoning" and their "emotional bias."
Size doesn't matter because you're doing the same stuff, but size makes all the difference. It's stupid to imagine that more space makes a qualitative difference in the experience, but lack imagination if you can't see how that extra space can be used.
Here, let me help:
Barely more functional than a7" tablet? The iPad has more than twice as much screen real estate than a 7" tablet. Maybe you lack the imagination to figure out to make proper use of that space, but others probably can. Browsing the web, watching videos, editing documents and more are all going to be better overall experiences on an iPad rather than a 7" tablet
That's what I am saying --- Samsung produced a cortex a8 chip a year ahead of Apple.
The dual core cortex a9 isn't even ready yet ---- RIGHT NOW --- from the big silicon companies. You are assigning mythical abilities to the PA Semi and Intrinsity teams.
Yes, it's not in production RIGHT NOW. The iPhone 5 isn't launching RIGHT NOW but in 8 months.
The iPad 2 is likely to be March or April. 4-5 months from now.
This isn't mythical abilities for Apple but fairly reasonable timeframes given where the A9s are now...in prototype units like the LG phone, Playbook tablet, etc and very close to full rate production.
There are prototype dual core Cortex A9 iPads in Cupertino RIGHT NOW. There are prototype dual core Cortex A9 iPhone 5s in Cupertino RIGHT NOW. There are dual core Cortex A9 based A5 samples in Cupertino right now.
Will Apple decide to go with single core A5s over duals for battery life? Maybe, but strikes me as doubtful.
Yes, it's not in production RIGHT NOW. The iPhone 5 isn't launching RIGHT NOW but in 8 months.
The iPad 2 is likely to be March or April. 4-5 months from now.
This isn't mythical abilities for Apple but fairly reasonable timeframes given where the A9s are now...in prototype units like the LG phone, Playbook tablet, etc and very close to full rate production.
There are prototype dual core Cortex A9 iPads in Cupertino RIGHT NOW. There are prototype dual core Cortex A9 iPhone 5s in Cupertino RIGHT NOW. There are dual core Cortex A9 based A5 samples in Cupertino right now.
Will Apple decide to go with single core A5s over duals for battery life? Maybe, but strikes me as doubtful.
So you are telling me that the time lag between a big silicon company like TI vs. Apple is less than 2 months --- because the earliest time RIM launches the Playbook seems to be in February.
If Apple launches a dual core cortex A9 ipad 2 in April 2011, they are buying a stock CPU from Samsung because I don't think that Apple can do it within 2 months of other big silicon companies.
Will Apple decide to go with single core A5s over duals for battery life? Maybe, but strikes me as doubtful.
If I'm reading ARM's product page correctly it looks like there will also be significant space savings with the single-core A9.
So you are telling me that the time lag between a big silicon company like TI vs. Apple is less than 2 months --- because the earliest time RIM launches the Playbook seems to be in February.
Yes, that's exactly what I am saying.
Apple's ability to execute is not tied to RIM's ability to execute. Thank god.
There is no reason they cannot have a dual core Cortex A9 based A5 design fabbed by Samsung and sampling internally as we speak (type). The S5PC100 in the 3GS was custom for Apple. The A8 in Hummingbird and the A4 is a custom version of that designed by Intrinsity and fabbed by Samsung.
Samsung and Apple have a very close relationship and there's no reason that Apple can't have been working on the A5 in parallel with Orion development. There's no need for things to have to happen sequentially.
So there's no reason to doubt that Apple couldn't lag TI by 2 months so long as Samsung doesn't lag TI by more than 2 months.
If I'm reading ARM's product page correctly it looks like there will also be significant space savings with the single-core A9.
Given Orion is 45nm this is a distinct possibility. Folks had expected 32nm but not so much so yah...2 core bigger than 1 core.
As an aside, TI isn't going past the 45nm process so effectively they become a fabless designer for next generation ARMs just like Apple. Probably end up using TSMC.
Yes, that's exactly what I am saying.
Apple's ability to execute is not tied to RIM's ability to execute. Thank god.
There is no reason they cannot have a dual core Cortex A9 based A5 design fabbed by Samsung and sampling internally as we speak (type). The S5PC100 in the 3GS was custom for Apple. The A8 in Hummingbird and the A4 is a custom version of that designed by Intrinsity and fabbed by Samsung.
Samsung and Apple have a very close relationship and there's no reason that Apple can't have been working on the A5 in parallel with Orion development. There's no need for things to have to happen sequentially.
So there's no reason to doubt that Apple couldn't lag TI by 2 months so long as Samsung doesn't lag TI by more than 2 months.
I am not talking about RIM, I am talking about TI and Samsung.
There is no reason to suspect that Samsung is going to do a custom Apple job ahead of Samsung's own chip --- and we haven't seen Samsung announcing that they are anything close to shipping their own dual core Cortex A9 chip yet.
And there is no reason to believe that big silicon companies like TI are shipping massive amounts of dual core cortex A9 chips either. Wall Street ain't estimating RIM to ship millions and millions of Playbook in Q1 --- they are estimating the figure to be in hundreds of thousands.
So you are telling me that somehow Samsung is going to ship a custom Apple job for millions and millions units --- ahead of their own Samsung chip (which won't be available in massive quantity until first half of next year).
I am not talking about RIM, I am talking about TI and Samsung.
There is no reason to suspect that Samsung is going to do a custom Apple job ahead of Samsung's own chip --- and we haven't seen Samsung announcing that they are anything close to shipping their own dual core Cortex A9 chip yet.
And there is no reason to believe that big silicon companies like TI are shipping massive amounts of dual core cortex A9 chips either. Wall Street ain't estimating RIM to ship millions and millions of Playbook in Q1 --- they are estimating the figure to be in hundreds of thousands.
So you are telling me that somehow Samsung is going to ship a custom Apple job for millions and millions units --- ahead of their own Samsung chip (which won't be available in massive quantity until first half of next year).
Think about it!
Say you are Samsung.
You have an internal demand for, say, 100 K A9s per month (0% profit).
You have an external demand for, say, 300 K A9s per month (20% profit).
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance.
Just who do you want to give priority -- especially since Apple can go to any other Fabs and make the same deal?
There's an old (supplier) saying from the days when IBM dominated the mainframe computer market with 97%:
When you make love with an 800 lb gorilla -- you stop when the Gorilla gets tired!
.
I'm sorry but that line of argument is nonsense. What magically different things are people doing on an iPhone vs. an iPad? They're still playing games, consuming media, browsing the web, etc.
Okay so you claim that the iPad isn't better because its bigger.
Barely more functional than a phone? A 7" tablet has roughly 4.5 times as much screen real estate than the iPhone. Maybe you lack the imagination to figure out to make proper use of that space, but others probably can. Browsing the web, watching videos, editing documents and more are all going to be better overall experiences on a tablet rather than a phone.
But then you claim a 7" tablet is better than a phone because its bigger.
Oops!
Fact is, the size differential between iPhone and iPad is significant meaning that the capability differential is also significant. The size different between iPhone and Playbook on the other hand is minimal meaning the expected capability differential is also minimal.
Of course, if you consider the capabilities of a current Blackberry as the baseline then yes of course the Playbook will be exponentially more capable simply because Blackberries are so incapable.
Think about it!
Say you are Samsung.
You have an internal demand for, say, 100 K A9s per month (0% profit).
You have an external demand for, say, 300 K A9s per month (20% profit).
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance.
Just who do you want to give priority -- especially since Apple can go to any other Fabs and make the same deal?
You reached the same conclusion as me --- that's why I said that if Apple is coming out with a dual core a9 ipad2 in Q2, it's going to be a Samsung stock CPU.
What nht was talking about was a custom Apple dual core A5 chip coming out before a Samsung stock dual core A9 chip.
You reached the same conclusion as me --- that's why I said that if Apple is coming out with a dual core a9 ipad2 in Q2, it's going to be a Samsung stock CPU.
What nht was talking about was a custom Apple dual core A5 chip coming out before a Samsung stock dual core A9 chip.
No, I believe the A5 design is done. Samsung will give priority to manufacture the A5.
Ages ago, I worked in the semiconductor industry.
In those days they would build chips to mil-qual specs for the military.
After manufacturing, the chips went through rigorous testing to assure that they performed to specs -- things like performing within a certain range under varying power, temperature, etc.
Those that passed testing went to fill orders.
Those that failed testing were either total rejects (scrapped) or marginal.
The marginal chips were often put in inventory where they could be used as less- critical spec'd parts (lower power, narrower temperature range, etc.)
So, with something like the A5:
1) Can they target manufacturing of say, a 1 GHz Dual-Core A9?
2) Can chips that don't meet those specs be used as a lesser part. e.g. a single core variant and/or a slower-clocked variant?
3) If so, do they design extra capacity into the chips to increase their yield?
Say Apple decides that for 2011 they want the following chips:
iPad - 1 GHz Dual-Core A9
iPhone - 1 GHz Single-Core A9
iPod Touch - 800 MHz Dual-Core A9
AppleTV - 800 MHz Single-Core A9
4) Would they manufacture a single chip targeted at the highest spec, then fill the lower-speced by under-clocking and/or disabling a core. etc."
5) If so, what other things can they over sped, then re-spec?
No, I believe the A5 design is done. Samsung will give priority to manufacture the A5.
Why should Samsung give priority to Apple --- when Samsung also manufactures smartphones and tablets themselves, which has a higher profit margin because it's a finished product. Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple.
Think about it!
Say you are Samsung.
You have an internal demand for, say, 100 K A9s per month (0% profit).
You have an external demand for, say, 300 K A9s per month (20% profit).
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance.
Just who do you want to give priority -- especially since Apple can go to any other Fabs and make the same deal?
There's an old (supplier) saying from the days when IBM dominated the mainframe computer market with 97%:
When you make love with an 800 lb gorilla -- you stop when the Gorilla gets tired!
.
Why should Samsung give priority to Apple --- when Samsung also manufactures smartphones and tablets themselves, which has a higher profit margin because it's a finished product. Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple.
Again, lets assume that you are Samsung -- I am pulling numbers out of the air:
1) You plan to set up a manufacturing line/process for ARM Dual-Core A9
2) You Budget, say, $250 Million to set up and test the line over a period of 3 months.
3) You amortize these costs over, say, a 3-year expected life of the ARM Dual-Core A9
4) Over those 3 years you plan to sell 300 Million units at a profit of $10 per unit -- a very attractive 30% gross profit margin .
5) Just for ease of calculation, the production starts in Jan 2011
Now, using the numbers from my earlier post:
You have an internal demand for, say, 100 K A9s per month (0% profit). These are the units that Samusing sells to itself for use in Galaxy Tab follow-ons. There is no profit on these chips until they are included in the manufacture of the tablets, and subsequently sold at a profit. But for sake of argument lets say Samsung can sell 100 K Samsung Galaxy Tab2 at full retail, realizing the $10 profit on each ARM Dual-Core A9 included.
So, Samsung will make $1 million per month on Samsung chips in Samsung Tablets
You have an external demand for, say, 300 K A9s per month (@20% profit) or $6.70. Here Samsung sells the chips to others at 2/3 the profit realized by selling to Samsung. There is immediate (30 days later) profit.
So, Samsung will make 300 K x $6.70 ~= $2 million per month on Samsung chips sold to others.
Apple comes to you with an order for 2 Million A9s per month (15% profit), for 6 months, payable in advance. Sales of chips to Apple are at 15% margin vs internal 30% margin, or $5 profit per chip.
So, Samsung will make 2 million x $5 == $10 Million per month on Samsung chips sold to Apple.
Putting these together:
Samsung will make 100 K x $10.00 == $1 million per month on Samsung chips in Samsung Tablets
Samsung will make 300 K x $06.70 ~= $2 million per month on Samsung chips sold to others.
Samsung will make 2 mill x $05.00 == $10 Million per month on Samsung chips sold to Apple.
Noteworthy:
There is a thing called the time/value of money
Apple pays Samsung in advance.
Apple guarantees sales for 6 months
Apple reserves certain rights to increase quantities/orders at these prices
Apple, essentially, pays for Samsung to setup the $250 Million manufacturing line/process (the A5 is a superset of the Samsung chip)
Apple can go to another Fab (a Samsung competitor) and make the same offer -- reducing Samsung's economy of scale
A bird in the hand is better than a turd in the bush!
To answer your question:
"Why should Samsung give priority to Apple -- Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple."
If Samsung can make $100 (including $10 on the A9) each and sell 1 million Tabs in 6 months thats $100 million profit ($10 Million on the A9) possible
Samsung can make $5 each and sell 12 million A5s in 6 months that's $60 million profit guaranteed
More profit, Less risk, better use of $ (payment in advance vs manufacturing and inventory costs).
That's what I am saying --- Samsung produced a cortex a8 chip a year ahead of Apple.
The dual core cortex a9 isn't even ready yet ---- RIGHT NOW --- from the big silicon companies. You are assigning mythical abilities to the PA Semi and Intrinsity teams.
No, you are just arguing shiite for the sake of trying to cover your ass. CortexA9 has been licensable as IP from ARM since early 2009. Tegra2 dual core shipped in 2010, although not power optimized as it came from Nvidia, it shows the maturity of IP to tape-out is well past question.
I wouldn't state my reputation in TI's ability to churn out other folks CPU IP, that's not their strength. But you just go ahead, because your technical reputation is already damaged beyond repair. Do you actually know facts or check them at all before you post? Because it sure doesn't read that way.
And Samsung didn't announce an A8 phone until Feb 2009, which is only a couple months before iPhone 3GS shipped. And Samsung produced that version of the A8 as a Samsung part number. Apple later made the A4 based on further CortexA8 improvements.
There has been plenty of time for Apple to do an CortexA9 version if they thought it was economically smart to ship. That's what Apple will use as criteria for inclusion, not spec-whoring.
Why should Samsung give priority to Apple --- when Samsung also manufactures smartphones and tablets themselves, which has a higher profit margin because it's a finished product. Selling a million galaxy tab is more profitable to Samsung than selling 10 million custom A5 chip to Apple.
Because every A4/A5 sold is finance risk-free. Samsung has no R&D skin in the A4/A5 game, and it keeps their contract fab busy. The Samsung contract fab CAN'T produce Galaxy Tabs, it can only provide chips, and Samsung uses the contracts to pay off the capital expenditure of building and upgrading the fab. That way they get their own Cortex CPUs without having to bear all the manufacturing financial risk themselves on top of the R&D risk inherent in implementing the Cortex IP on silicon. Flat out Samsung needs Apple contract business more than Apple needs Samsung specifically because TSMC could do the job nicely too.
The profit on the contract run chips is many times higher compared to in-house taped and used ARM Cortex chips because of all of the above. I'll even sat Dick's numbers on internal cost are skewed in your favor because he missed accounting for the fact Samsung has no R&D investment to recoup for A4/A5 before they can recognize any profit. And that skewing still didn't look good for you, your position is even worse than you were led to believe the the first thrashing.
So you have it exactly backwards as far as fabbing chips go. It's obvious you have no idea how the industry works, at all.
Because every A4/A5 sold is finance risk-free. Samsung has no R&D skin in the A4/A5 game, and it keeps their contract fab busy. The Samsung contract fab CAN'T produce Galaxy Tabs, it can only provide chips, and Samsung uses the contracts to pay off the capital expenditure of building and upgrading the fab. That way they get their own Cortex CPUs without having to bear all the manufacturing financial risk themselves on top of the R&D risk inherent in implementing the Cortex IP on silicon. Flat out Samsung needs Apple contract business more than Apple needs Samsung specifically because TSMC could do the job nicely too.
The profit on the contract run chips is many times higher compared to in-house taped and used ARM Cortex chips because of all of the above. I'll even sat Dick's numbers on internal cost are skewed in your favor because he missed accounting for the fact Samsung has no R&D investment to recoup for A4/A5 before they can recognize any profit. And that skewing still didn't look good for you, your position is even worse than you were led to believe the the first thrashing.
So you have it exactly backwards as far as fabbing chips go. It's obvious you have no idea how the industry works, at all.
Rightly or wrongly --- Korean and Japanese conglomerates don't care about short term profitability --- between cross shareholdings and government pushing technology agendas, satisfying Apple's needs are going to be a low priority.
I am not saying that's the right thing for Samsung to do financially, I am saying that's what they would do --- given how the Korean and Japanese conglomerates operate. Hell, after 20 decades of economic mess, the Japanese conglomerates would still do that.
Rightly or wrongly --- Korean and Japanese conglomerates don't care about short term profitability --- between cross shareholdings and government pushing technology agendas, satisfying Apple's needs are going to be a low priority.
This is called saying even more stupid things in the hopes that folks forget the stupid things already stated. This is why Hiro says your technical reputation is beyond repair.
I am not saying that's the right thing for Samsung to do financially, I am saying that's what they would do --- given how the Korean and Japanese conglomerates operate. Hell, after 20 decades of economic mess, the Japanese conglomerates would still do that.
Given that Samsung gave Apple A4 production equal if not higher priority over Hummingbird used in their own flagship products this is demonstrably false.
Apple had millions of A4s produced for them and launched the iPad in April where Samsung's flagship Galaxy S using Hummingbird didn't launch until June. Apple announced in Jan, took pre-orders on March 12 before Samsung even announced the S (March 23).
Given which product came out first and the qty of iPads built it's fairly easy to make the case that Samsung gave Apple A4 production priority over its own.
This is called saying even more stupid things in the hopes that folks forget the stupid things already stated. This is why Hiro says your technical reputation is beyond repair.
Given that Samsung gave Apple A4 production equal if not higher priority over Hummingbird used in their own flagship products this is demonstrably false.
Apple had millions of A4s produced for them and launched the iPad in April where Samsung's flagship Galaxy S using Hummingbird didn't launch until June. Apple announced in Jan, took pre-orders on March 12 before Samsung even announced the S (March 23).
Given which product came out first and the qty of iPads built it's fairly easy to make the case that Samsung gave Apple A4 production priority over its own.
Given that stock cortex a8 chips were available one year before both Samsung and Apple released their version, you are saying that Samsung and Apple have caught up with the big silicon companies in one year's time.