Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
The list of included components would not be limited to memory, however, so future device teardowns will be interesting as mobile devices begin to include these technologies. More information on TSMC's packaging technologies can be accessed via this PDF.
This might be the major takeaway if the story had been posted at IFIXIT. "We give the iPhone 10 a repairability score of -20, as we can no longer replace the modem chip as a separate component." Shouldn't the takeaway at MacRumors be, "thinner?" A technique that replaces discrete wires with other interconnect methods would seem to offer greater reliability and space savings, though the latter is not immediately apparent from the illustrations.

There's no denying the advantages of three-dimensional fabrication. They were clear to Jack Kilby at the dawn of the IC (per Wikipedia, emphasis added)
A precursor idea to the IC was to create small ceramic squares (wafers), each containing a single miniaturized component. Components could then be integrated and wired into a bidimensional or tridimensional compact grid. This idea, which seemed very promising in 1957...
What's remarkable to me, as someone to whom "IC" once meant 4 logic gates on a 14-pin DIP, is the utter sophistication and density.

As we know, "two-dimensional thinking" was the downfall of Khan (as in, Star Trek II: the Wrath of...), though I have to wonder why a late-20th Century genius would be unaware of three-dimensional aerial combat techniques pioneered during World War I...
 
  • Like
Reactions: Avieshek and Zirel
The Commercial Times report mentions TSMC's integrated fan-out wafer-level packaging (InFO WLP) technology as one of the key inclusions in the production contract.

Pack this! Samsung!

Also, when someone asks why Apple doesn't manufacture their own chips? This is the answer, either TSMC is better or Samsung is better, Apple can always chose the best one for the iPhone.
 
It really is hard to fathom what the iPhone 7/A10 combo will look like. I expected a tri-core processor w/ 2GB of RAM for the A9, but it turned out to be a dual core processor w/ 2GB of RAM instead.

I mean we're a year away, so it's just pure speculation at this point. I wouldn't be surprised it's a dual core w/ 2GB of RAM again. I would always say more RAM would benefit the consumer within an iPhone, but it's too much of stretch given Apple's past tendencies. I think the iPhone ought to shift towards more practical and ergonomic ideas. A water-resistant iPhone would be great.
 
CPU possibilities all fairly interesting but this is the bit that really excites me with its possibilities...

According to a paper abstract from TSMC engineers, InFO WLP also allows for better thermal performance as well as superior performance for radio frequency (RF) components such as cellular modems. We reported last year about Apple hiring more engineers to potentially bring RF component development in house, ...

Apple caused quite a stir when its CPU engineers came up with the A7 ground-up redesign of an ARM-based SoC. If these Apple RF engineers can do a similarly great re-implementation of some of the core RF components, maybe with a single super-optimised R10 ("R" for "Radio") chip that combines all the functionality of the existing LTE and WiFi/Bluetooth chips with significantly reduced power consumption and reduced use of PCB real-estate the main PCB could be made smaller and the extra space given over to a bigger battery. Those two factors could make a valuable contribution to battery life and significantly better battery life is about all I want from the next iPhone.
 
  • Like
Reactions: Avieshek
Easy solution: Samsung will steal TSMC 's technology.
Not for the very first time .....

This is great news.

Just as Apple fixed Bendgate with the iPhone 6s, whilst publicly denying it, so they will fix Chipgate with the iPhone 7, whilst similarly denying it.

Finally, we can look forward to iPhones that all have the same battery life, as opposed to the current iPhone 6s, which has battery life that can be as much as two hours shorter if you are afflicted with the Samsung chip.
Interesting way to twist the reality.
There was no bendgate, nor a chipgate....

....but people will still complain next year o_O
That's for sure.
Buttongate... Colorgate... Whatever reason Apple bashers are going to make up....
 
Apple caused quite a stir when its CPU engineers came up with the A7 ground-up redesign of an ARM-based SoC. If these Apple RF engineers can do a similarly great re-implementation of some of the core RF components, maybe with a single super-optimised R10 ("R" for "Radio") chip that combines all the functionality of the existing LTE and WiFi/Bluetooth chips with significantly reduced power consumption

Apple weaves RAM into the SoC packages now and does no real, substantive RAM chip/die development work. The Fan Out package means can put other peoples stuff in the package at least as much as can put your own stuff in. You just have to be find partners willing to sell you raw, finished dies of what is inside their more mainstream packages. That is the key factor.

Apple is unlikely to do both RF subsystems in one step. With the highly fragmented LTE space it also seems unlikely that Apple is going to take that on internally.

If Intel or Qualcomm was willing to sell a modem chip die instead of a whole package then the Fan Out packaging could work without Apple doing the work in house. May need some in house folks to work with the models of the packaging impacts and layout/interconnection/interference impacts but not particuarly doing to the whole RF subsystem implementation.

Of the two RF subsystems I could see Apple cutting off Broadcomm/Qualcomm on the WifiBluetooth before LTE radios. Wifi/Bluetooth goes into all Macs and iOS devices. Apple could make modules that work with both product lines and get incrementally higher volumes of usage. The LTE radios not so much.

RF wouldn't be my first bet though. The Motion processor could be the side car die inside the package (well Motion plus some other run always supervisoring functions ) . Similar if keeping 2GB RAM constant, but shrinking it to smaller process size might now allow for side-by-side placement. There should be cooling upsides to not having the RAM sitting on top/under the CPU. Cooler longer then don't have to throttle sooner.

and reduced use of PCB real-estate the main PCB could be made smaller and the extra space given over to a bigger battery.

I don't think that Fan Out is a major reduction in PCB space. "Fan Out" means taking up more 2D space and perhaps dropping from taking up 3D space. You'd get some saving because the packages the dies go in are typically bigger than the dies ( due to pins and ball pads connections needing more space. ). However, there is an offset. The package doing a "fan out" into gets slightly bigger. If that increase is 10% bigger and get rid of a 20% bigger 2nd package then get a reduction, but only a subset of the original 2nd package. Only get a big win when large majority of the pins/pads from the 2nd package were 100% going to the one collapsing into. Stuff like RF subsystems hooked to far more than just the CPU main package.


PCB isn't a major battery volume blocker. More so what this could do is lower costs a bit and may open up space on the PCB for more/pricier gadgets to be added for the PCB same volume. May get some limited battery bumps with shorter, less lossy linkages, but battery volume likely isn't a huge win here.[/quote][/QUOTE]
 
Last edited:
http://semiaccurate.com/2013/03/25/what-is-apple-doing-at-14nm/
http://semiaccurate.com/2014/03/25/details-apples-gpu-emerge/

In the second link look at the tags. It will be very interesting chip. We can see first HBM memory in mobile SoC instead of DDR memory of any type.

Nice but I'm not about to sign up for access to semiaccurates services. Nothing negative or positive here I simply have better uses for my money.


All this speculation does make me wonder what GPU is going into the coming iPad Pro. Will it be a beefing up of the current A9 GPU or something that is a bigger step forward? Make you wonder.
 
There was no "chipgate" and no issue. To really understand that though, you would need to be an engineer. The fact the many aren't, and really don't understand the underlying technology, manufacturing, second-sourcing of components, semiconductor processes, tolerances, etc, is evident, and is what ultimately froths up the tech media and its readers.

One shouldn't need to be an engineer to understand some of the issues with mass production parts. I'm actually surprised that many people posting here even made it to high school much less further in life. It isn't like the yields of the various electronics manufacturing technologies has changed drastically over the years, even the lowly restore varies widely coming off the production line.
 
I'll be like 98% of the macrumors forum posters on here who don't know anything about cutting edge engineering or the intracacies of huge multi-national corporate economics, politics, legalities, and supply chain goings-on, etc.

And say Apple is doomed because I read this article today.
 


FC technology for Packaging current Apple processors ( nothing new, developed it back in 1992 in AZ, been used by Intel etc. since 1998, Silicon Valley no longer develops hardware technology, sponges off those developed by others ) has 3 expensive components ( wafer bumping, organic substrate, FC assembly ), all done outside the Foundry ( like TSMC ). Now TSMC wants to keep all that within their Foundry with somewhat cheaper processing for their InFOWLP. But TSMCs InFOWLP technology does not have unique advantages, others too can do very fine interconnect lines and vias key to packing em in. But InFOWLP does not address the Bandwidth issue very well within the footprint limits of Smart Phones etc., hence the line width advantage is moot. So why would Apple give up their leverage to play one Foundry against another by signing up with just TSMC ?

tsmc_logo_new-250x205.jpg
According to a recent report from Taiwan's Commercial Times, via EE Times and a separate research report from KGI Securities' Ming-Chi Kuo, Taiwan-based TSMC may have won sole production rights on the A10 chip slated for the next-generation iPhone 7.

This is in contrast to the split production of the A9 processor between Samsung and TSMC featured in the iPhone 6s and iPhone 6s Plus. Apple's decision to revert back to TSMC as a single supplier, as was seen in A8 chip production, could be motivated by advanced device packaging techniques offered by TSMC that may not have equivalents in Samsung's packaging offerings.

The Commercial Times report mentions TSMC's integrated fan-out wafer-level packaging (InFO WLP) technology as one of the key inclusions in the production contract. InFO WLP is one of many competing 3D IC technologies that promise higher levels of component integration in a single package with better electrical characteristics.

Among those improvements is the possibility for higher-width memory buses that support lower-power operation necessary for mobile devices, which for consumers means better performance and efficiency. 3D IC technologies are just beginning to emerge in the consumer space, with AMD's use of High Bandwidth Memory (HBM) in its Fiji XT line of discrete graphics cards being one of the first implementations.

According to a paper abstract from TSMC engineers, InFO WLP also allows for better thermal performance as well as superior performance for radio frequency (RF) components such as cellular modems. We reported last year about Apple hiring more engineers to potentially bring RF component development in house, so this packaging technology could serve as additional motivation to Apple for packaging in the future. Even if Samsung could offer Apple a comparable technology, the challenges of verifying a design on two new manufacturing flows may be a motivating factor for Apple to stick with one supplier for its next processor.

In the near term, the thermal advantages and potential increased memory bandwidth are the more immediate sources of improvement for Apple's potential next chip. Many 3D IC technologies have seen slow adoption due to increased costs and processing steps, but the simpler InFO WLP technology offers an easier, cheaper entry point for Apple, which also has the luxury of uncommonly high margins on its devices.

tsmc_chip_packaging.jpg

Comparison of packaging technologies offered by TSMC

TSMC's InFO WLP differs from many competing 3D IC solutions in that it does not require an additional silicon interposer along with the existing package substrate used for component integration. Though they do not feature active components, silicon interposers are made on silicon wafers just like the application processors featured in mobile devices, making them a costly addition to the device assembly.

InFO WLP allows multiple flip chip components to be placed side-by-side on a package substrate resembling a traditional assembly, but with the ability to interconnect to one another through the package substrate. This is in contrast to traditional methods which feature stacked packages (package on package, or PoP) interconnected with tiny wires. As mobile memory technologies advance, with LPDDR4 being the latest iteration, electrical signaling becomes an increasing technological challenge which begins to make 3D IC technologies more attractive for enhanced performance.

The list of included components would not be limited to memory, however, so future device teardowns will be interesting as mobile devices begin to include these technologies. More information on TSMC's packaging technologies can be accessed via this PDF.

Article Link: Rumored A10 Production Win for TSMC Could Be Tied to Device Packaging Advances
the Fabless Wonders of Si Valley do not develop hardware mfg. technology, sponges off more traditional Co.s,
 
It really is hard to fathom what the iPhone 7/A10 combo will look like. I expected a tri-core processor w/ 2GB of RAM for the A9, but it turned out to be a dual core processor w/ 2GB of RAM instead.

I mean we're a year away, so it's just pure speculation at this point. I wouldn't be surprised it's a dual core w/ 2GB of RAM again. I would always say more RAM would benefit the consumer within an iPhone, but it's too much of stretch given Apple's past tendencies. I think the iPhone ought to shift towards more practical and ergonomic ideas. A water-resistant iPhone would be great.
Just to put a fine point on what you said (which I agree with, for the most part), from a production point of view, we're about 5 months (end of June) away from final firm configuration of the phone, and then there is pre-production, test, and the "golden master" for the phone will be in about 7 months (end of August), and then the factories in China will be putting these together in a furious pace for the launch in September.

It's kind of like Christmas... People in the major retailers are sweating in June to get the orders in for December.
 
A10 can have higher single threaded performance and Hyper Threading on dual core CPU. Also if it will have HBM as a system memory, and cache for CPU and GPU, we can expect even higher performance from it.
 
Just to put a fine point on what you said (which I agree with, for the most part), from a production point of view, we're about 5 months (end of June) away from final firm configuration of the phone, and then there is pre-production, test, and the "golden master" for the phone will be in about 7 months (end of August), and then the factories in China will be putting these together in a furious pace for the launch in September.

It's kind of like Christmas... People in the major retailers are sweating in June to get the orders in for December.

I meant from the consumer perspective. It's a cyclical approach from the production side. Based off Apple's past tendencies, they keep upgrades relatively small, especially since their fans are so passionate about their products that they can get away with it.
 
  • Like
Reactions: JeffyTheQuik
Please use your power as a shareholder to insist they update their logo to something that no longer looks like it's come from a 1970's TV sci-fi show
I prefer an old logo (being invisible on the end product anyway) and better products over burning money for a new logo design that is eventually missing in R&D.

btw: The 70s were a cool decade ...

Upps - just realized that I answered to a thread from Jan ... my apology for zombie posting :oops:
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.