"We believe Apple is deeply involved in the manufacturing and supply chain process down to the factory floor," analyst Peter Misek said. "We believe Apple is creating deep partnerships with a handful of suppliers in industries where it feels there is a competitive advantage (e.g., secure supply, secure lower price, achieve a technological edge)."
"The firm believes Apple is leveraging its massive cash flow to fund suppliers' capex. Analysis suggests Apple is spending $3B+ in FY12 capex on these equipment purchases and $7B+ in FY13, but in return Apple is reducing its COGS by ~$1.4B (~84bp improvement to GM) in FY12 and ~$1.9B (~89-95bp improvement to GM) in FY13."
Unattributed. Published 14 February 2012. Jefferies Lifts Q1/FY12 Estimates on Apple (AAPL); Can Maintain Higher Margins Longer. Street Insider. Retrieved 6 July 2012.
In post 56, where I also get ticked off at a really dumb, insulting post about it, aimed at me.
In post 56, where I also get ticked off at a really dumb, insulting post about it, aimed at me.
It doesn't help to reply in kind tho, particularly as a moderator. You (we) should not bring ourselves down to the level of the egregious trolls we reply to.
I'm very curious to see what chip Apple ends up using. I'm sure it will be 32 nm (they are already using some 32nm A5s now).
The big question for me is whether it will just be a die-shrunk, and higher clocked A5 or if it will be an A6. And if A6, does that necessarily mean quad cores? Or could it mean two beefier cores? Or could it be something radically new?
My guess is that it will be a 32nm quad core chip that they call the A6, and that the next iPad will be a 32 nm quad core chip that they call the A6X (beefier graphics to support the larger number of pixels in an iPad). If not that, then I'd guess the die shrunk A5.
I'm also guessing that the iPad Mini will get a 32 nm A5.
If you watch Apple's hardware moves, you'll notice they don't make any giant leaps just to help out consumers. It's pretty obvious that if they update the screens to be a different widescreen resolution, and change the look, all while adding 4G, the likelihood that they'd also be adding a quadcore chip is pretty low. You can almost consider that to be a rumor that pans out for next years iPhone 5S
Inaccuracies. Couple of things, Apple designs their own custom CPUs based on the ARM architecture, not based on what Samsung is doing (even though Samsung makes ARM chips too). Apple partners with Samsung to fabricate the Apple designed chips, currently using Samsung's 45nm fabrication process. It's important to note these are not based on a chip design by Samsung. Yes, there is a huge difference in meaning... Also, Apple is working with other fabs to build these CPUs on an even smaller 32nm process, and hopefully 22nm at some point. Whether Samsung will remain Apple largest silicon provider is uncertain.
It's also premature to say that Apple will be making a quad-core CPU for the next iPhone, or that it will "heat up competition" or whatever. The number of cores or the level of Gigahertz does not correlate to actual performance. Having four cores is basically useless when 3 of them sit idle 90% of the time, as is the case with Android products. Apple w/iOS gets significantly more mileage out of their dual-core ARM CPUs than Android currently can with 2, 4, or even 8 to 12 cores (if such a beast existed). I would not be at all surprised to see the next iPhone ship with a dual-core CPU based on a more advanced ARM architecture with a hex-core GPU built in. Nobody will know for sure until it ships.
It's also premature to say that Apple will be making a quad-core CPU for the next iPhone, or that it will "heat up competition" or whatever. The number of cores or the level of Gigahertz does not correlate to actual performance. Having four cores is basically useless when 3 of them sit idle 90% of the time, as is the case with Android products. Apple w/iOS gets significantly more mileage out of their dual-core ARM CPUs than Android currently can with 2, 4, or even 8 to 12 cores (if such a beast existed). I would not be at all surprised to see the next iPhone ship with a dual-core CPU based on a more advanced ARM architecture with a hex-core GPU built in. Nobody will know for sure until it ships.
Google Nexus 7 has a 1.2 GHz Nvidia Tegra 3 quad-core ARM Cortex-A9 CPU with additional low-speed companion core and a 416 MHz Nvidia GeForce ULP dodeca-core (12 core) GPU. (1)
The fact that Google Nexus 7 has all that computing power in addition to "Project Butter" which introduced Android 4.1 "Jelly Bean" to extend vsync timing across all drawing and animation done by the Android framework, triple buffering in the graphics pipeline, synchronizing touch to vsync timing and anticipating where your finger will be at the time of the screen refresh as well as applying a CPU input boost at the next touch event after periods of inactivity. (2)
Google Nexus 7 exemplifies your statement, "The number of cores or the level of Gigahertz does not correlate to actual performance."
1. Unattributed. Published Date Unknown. Nexus 7 Features. Google. Retrieved 6 July 2012.
2. Unattributed. Published Date Unknown. Android 4.1 for Developers . developers.android.com. Retrieved 6 July 2012.
As others have posted, I think a big determining factor to what ARM architecture goes into the A6 chip, is mainly an issue of logistics -- can they get the yields they need at the time they need them.
Other issues such as need for additional battery, 4G chips, etc., may be obviated by a redesign of the device, e.g. the next iPhone may have a thinner rear case and greater overall height!
What complicates Apple's logistics problem with newly available technology is that they have 2 high-demand product lines -- the iPhone and the iPad, which are typically refreshed 4-6 months apart. So, Apple would like to have:
1) the latest technology (chip architecture and manufacturing)
2) yields to meet the anticipated demands
3) low cost -- efficiency and economies of scale
Here's a possibility -- Say Apple:
1) Specs a maxed out [latest available] ARM and manufacturing process for the next iPad -- for availability in anticipated iPad quantities for March 2013.
2) Specs a lower-capability version of the same chip for the iPhone for October 2012.
3) Begins manufacturing of iPad chip ASAP at maximum capacity -- even at low yield, and bank the qualified chips for the next iPad.
4) Tests the rejects from 3) at the lower-speced iPhone chip.
5) Testing for the lower spec should give a greater yield of iPhone chips.
If the process still works the way [I understood] that it used to, you can cripple failing components (cores, etc.), lower processing speed... and use the chip at a lower spec.
Say, that Apple determines it needs 100 million iPhone A6 chips from October thru March and 30 million iPad chips beginning in March -- then some number of both flavors going forward.
So, If Apple (based on history and sampling) can reliably estimate yields, they could start say, 200 million (300 million?) iPad chips today, realizing that a high number of the iPad rejects will qualify as reduced-spec iPhone chips.
If the iPad yields are higher/sooner than expected they can just cripple some chips and use them in the iPhone.
If the iPhone (iPad rejects) yields are lower than expected, they can use the iPad chips crippled as above -- and delay announcement of the new iPad (That's their ace-in-the-hole).
If they do a reasonable job of estimating and manufacturing -- the yields should ramp up to meet demand. They will be manufacturing the latest technology, in the needed quantities, at the desired time.
And, they will have increased the efficiency and reduced costs (primary and secondary yields) of a single architecture and manufacturing process.
Then the next members of the iDevice family can, in turn, utilize this technology (crippled as needed).
Is this still possible/practical in today's world of semiconductor manufacturing -- with all the SOC & POP?
Edit: It occurs to me that one way to make delaying the March iPad less painful (if necessary) -- would be to release the iPad 8" (based on currently available technology) around Sept--Nov.
Edit 2: Likely, I have embarrassed myself and exhausted my dated knowledge of semiconductor manufacturing....
My last real involvement in the field was in the early 1980's when we setup LANs at IBM, National Semi, Fairchild/Schlumberger and the like. One LAN in particular had Apple ][s in the Board Room at Applied Materials headquarters (they did their forecasting and yield computations using VisiCalc).
Applied Materials had coasters similar to this -- laying around everywhere:
Second, while Samsung could conceivably provide the fab services to make the chip, I doubt Apple would even let Samsung in the door unless absolutely necessary considering the hostility towards each other right now.
Why does this sound more like wishful thinking?
Perhaps this is how YOU would respond if you were Apple?
At the end of the day, its the bottom line that matters.
Nothing personal in business (except when you are Steve Jobs or his minions)
Let it be known that even Qualcomm is making contracts with Samsung to produce Snapdragon chips for them.
One thing is clear: its very hard to untangle yourself from Samsung's wide tenticles.
Well, I really doubt that it will turn out to be an Exynos-like processor, but I can really see the possibility of a quad core processor. It's just the next step in the evolution of smartphones. Apple can't always be far behind. They proved that by topping the market tech with the new Macbook Pros that sport an ivy bridge processor
It doesn't help to reply in kind tho, particularly as a moderator. You (we) should not bring ourselves down to the level of the egregious trolls we reply to.
In theory, you're right. But theory and practice don't always coincide, as you know. The very fact that someone is stupid enough to insult a mod is in itself worth of a deletion, at least, and at many sites, an immediate ban. I'm much more forgiving.
Comments
In post 56, where I also get ticked off at a really dumb, insulting post about it, aimed at me.
Quote:
Originally Posted by melgross
In post 56, where I also get ticked off at a really dumb, insulting post about it, aimed at me.
It doesn't help to reply in kind tho, particularly as a moderator. You (we) should not bring ourselves down to the level of the egregious trolls we reply to.
Quote:
Originally Posted by Blastdoor
I'm very curious to see what chip Apple ends up using. I'm sure it will be 32 nm (they are already using some 32nm A5s now).
The big question for me is whether it will just be a die-shrunk, and higher clocked A5 or if it will be an A6. And if A6, does that necessarily mean quad cores? Or could it mean two beefier cores? Or could it be something radically new?
My guess is that it will be a 32nm quad core chip that they call the A6, and that the next iPad will be a 32 nm quad core chip that they call the A6X (beefier graphics to support the larger number of pixels in an iPad). If not that, then I'd guess the die shrunk A5.
I'm also guessing that the iPad Mini will get a 32 nm A5.
If you watch Apple's hardware moves, you'll notice they don't make any giant leaps just to help out consumers. It's pretty obvious that if they update the screens to be a different widescreen resolution, and change the look, all while adding 4G, the likelihood that they'd also be adding a quadcore chip is pretty low. You can almost consider that to be a rumor that pans out for next years iPhone 5S
Inaccuracies. Couple of things, Apple designs their own custom CPUs based on the ARM architecture, not based on what Samsung is doing (even though Samsung makes ARM chips too). Apple partners with Samsung to fabricate the Apple designed chips, currently using Samsung's 45nm fabrication process. It's important to note these are not based on a chip design by Samsung. Yes, there is a huge difference in meaning... Also, Apple is working with other fabs to build these CPUs on an even smaller 32nm process, and hopefully 22nm at some point. Whether Samsung will remain Apple largest silicon provider is uncertain.
It's also premature to say that Apple will be making a quad-core CPU for the next iPhone, or that it will "heat up competition" or whatever. The number of cores or the level of Gigahertz does not correlate to actual performance. Having four cores is basically useless when 3 of them sit idle 90% of the time, as is the case with Android products. Apple w/iOS gets significantly more mileage out of their dual-core ARM CPUs than Android currently can with 2, 4, or even 8 to 12 cores (if such a beast existed). I would not be at all surprised to see the next iPhone ship with a dual-core CPU based on a more advanced ARM architecture with a hex-core GPU built in. Nobody will know for sure until it ships.
Google Nexus 7 has a 1.2 GHz Nvidia Tegra 3 quad-core ARM Cortex-A9 CPU with additional low-speed companion core and a 416 MHz Nvidia GeForce ULP dodeca-core (12 core) GPU. (1)
The fact that Google Nexus 7 has all that computing power in addition to "Project Butter" which introduced Android 4.1 "Jelly Bean" to extend vsync timing across all drawing and animation done by the Android framework, triple buffering in the graphics pipeline, synchronizing touch to vsync timing and anticipating where your finger will be at the time of the screen refresh as well as applying a CPU input boost at the next touch event after periods of inactivity. (2)
Google Nexus 7 exemplifies your statement, "The number of cores or the level of Gigahertz does not correlate to actual performance."
1. Unattributed. Published Date Unknown. Nexus 7 Features. Google. Retrieved 6 July 2012.
2. Unattributed. Published Date Unknown. Android 4.1 for Developers . developers.android.com. Retrieved 6 July 2012.
As others have posted, I think a big determining factor to what ARM architecture goes into the A6 chip, is mainly an issue of logistics -- can they get the yields they need at the time they need them.
Other issues such as need for additional battery, 4G chips, etc., may be obviated by a redesign of the device, e.g. the next iPhone may have a thinner rear case and greater overall height!
What complicates Apple's logistics problem with newly available technology is that they have 2 high-demand product lines -- the iPhone and the iPad, which are typically refreshed 4-6 months apart. So, Apple would like to have:
1) the latest technology (chip architecture and manufacturing)
2) yields to meet the anticipated demands
3) low cost -- efficiency and economies of scale
Here's a possibility -- Say Apple:
1) Specs a maxed out [latest available] ARM and manufacturing process for the next iPad -- for availability in anticipated iPad quantities for March 2013.
2) Specs a lower-capability version of the same chip for the iPhone for October 2012.
3) Begins manufacturing of iPad chip ASAP at maximum capacity -- even at low yield, and bank the qualified chips for the next iPad.
4) Tests the rejects from 3) at the lower-speced iPhone chip.
5) Testing for the lower spec should give a greater yield of iPhone chips.
If the process still works the way [I understood] that it used to, you can cripple failing components (cores, etc.), lower processing speed... and use the chip at a lower spec.
Say, that Apple determines it needs 100 million iPhone A6 chips from October thru March and 30 million iPad chips beginning in March -- then some number of both flavors going forward.
So, If Apple (based on history and sampling) can reliably estimate yields, they could start say, 200 million (300 million?) iPad chips today, realizing that a high number of the iPad rejects will qualify as reduced-spec iPhone chips.
If the iPad yields are higher/sooner than expected they can just cripple some chips and use them in the iPhone.
If the iPhone (iPad rejects) yields are lower than expected, they can use the iPad chips crippled as above -- and delay announcement of the new iPad (That's their ace-in-the-hole).
If they do a reasonable job of estimating and manufacturing -- the yields should ramp up to meet demand. They will be manufacturing the latest technology, in the needed quantities, at the desired time.
And, they will have increased the efficiency and reduced costs (primary and secondary yields) of a single architecture and manufacturing process.
Then the next members of the iDevice family can, in turn, utilize this technology (crippled as needed).
Is this still possible/practical in today's world of semiconductor manufacturing -- with all the SOC & POP?
Edit: It occurs to me that one way to make delaying the March iPad less painful (if necessary) -- would be to release the iPad 8" (based on currently available technology) around Sept--Nov.
Edit 2: Likely, I have embarrassed myself and exhausted my dated knowledge of semiconductor manufacturing....
My last real involvement in the field was in the early 1980's when we setup LANs at IBM, National Semi, Fairchild/Schlumberger and the like. One LAN in particular had Apple ][s in the Board Room at Applied Materials headquarters (they did their forecasting and yield computations using VisiCalc).
Applied Materials had coasters similar to this -- laying around everywhere:
Quote:
Originally Posted by sflocal
Second, while Samsung could conceivably provide the fab services to make the chip, I doubt Apple would even let Samsung in the door unless absolutely necessary considering the hostility towards each other right now.
Why does this sound more like wishful thinking?
Perhaps this is how YOU would respond if you were Apple?
At the end of the day, its the bottom line that matters.
Nothing personal in business (except when you are Steve Jobs or his minions)
Let it be known that even Qualcomm is making contracts with Samsung to produce Snapdragon chips for them.
One thing is clear: its very hard to untangle yourself from Samsung's wide tenticles.
Well, I really doubt that it will turn out to be an Exynos-like processor, but I can really see the possibility of a quad core processor. It's just the next step in the evolution of smartphones. Apple can't always be far behind. They proved that by topping the market tech with the new Macbook Pros that sport an ivy bridge processor
In theory, you're right. But theory and practice don't always coincide, as you know. The very fact that someone is stupid enough to insult a mod is in itself worth of a deletion, at least, and at many sites, an immediate ban. I'm much more forgiving.