Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
A5 Chip in Tweaked Apple TV Still Manufactured by Samsung at 32nm
- Thread starter MacRumors
- Start date
- Sort by reaction score
To an end user, a chip maker is just a provider of plumbing and there are many others who can do it . After all, Apple users already use graphics chips made by TSMC and CPUs made by Intel on their Macs, and iPhones and iPads with wireless chips made also by TSMC
Ideally, we would all be using the chips made by the most advanced process available now, which would be Intel, not Samsung. Not that Samsung is a bad chip maker by any means but I suspect the biggest advantage of Samsung is that they have worked together with Apple's engineering team for a long time. Also it's always a big risk when you move your big manufacturing away from your long time supplier.
It's not just a "vain" attempt though. Apple tells Samsung ahead of time what kind of processor they need and what quantity, that could be a valuable information for the competition. Just look at all the noise created by the analysts checking with the suppliers.
A good reason to distance themselves from Samsung don't you think? They are not the only fish in the sea and Apple does not need to feed the Samsung engine any more. They have proven to not be friendly.
business partners have always sued each other and continued to do business
no one paid any attention to it until bloggers came on the scene to link to any halfway interesting story to keep the clicks going
----------
the process to manufacture a CPU is different for each design
and when you get to Apple's size going from 32nm to 28nm is a huge money saver so it makes sense to choose your manufacturing partners wisely.
If Apple could possibly do without Samsung, that is exactly what would be happening.
But that is not reality.
The reality is that Apple is completely and totally dependent on Samsung as of now. And if Samsung were to pull the cork, Apple would founder and sink to the icy depths.
Does anybody really cares about it?!
Some of us like to try and predict Apple's future product plans based on past behaviors.
The Apple TV has a small enough volume that it can test out new processor designs without having to worry about iPhone-level volume and ramp-up time.
Looking over recent events we see the iPod Nano was turned into a watch, almost entirely a third-party invention that Apple tried to embrace with additional watch faces and minor multi-touch on a small screen (rotation mostly, and swiping). The problem was the device looked like an iOS device, but wasn't. It was too limiting (from an OS perspective for what Users wanted to really do).
Now the iPod nano is back to looking like an iPod, not a watch. But people really seemed to like the iWatch concept. Why would Apple abandon a really popular market segment (fashion-computing/wearable computing)?
Typically Apple may appear to ignore/abandon a popular market until it gets the product "Right." And I think we are seeing that happen now. Apple may very well be making a true iOS-based A5-running iWatch to fill the gap between iPhone/iPad and iPods like the former nano.
Receive a phone call, your iWatch shows the callers image/name and vibrates. You answer the call on your iPhone-connected earbuds. Swipe your iWatch at NFC stations to pay, or to unlock your car or login to your Mac. Its all an iOS-ecosystem Apple is creating.
Meanwhile Google makes Stalker Glasses. Derp.
Yea, that makes sense really. I hope apple use this chip + full ios on the iwatch.
There really isn't anything Apple could use this on besides tweaked versions of what's out now or a new device as everything already has an A5 besides the iPhone 4 that they are still making. It could just be a test run for 28nm and possibly so they can test TMSC's 28nm.
You say this as if it's a bad thing. Samsung is one of the only Android manufacturers scaring Apple into not doing the same things over and over again. Plus, this is a marketplace. If you're operating a business and your goals aren't expanding and growing profits as well as trying to dive into new markets, you're poorly operating your business.
It isn't quite like running a Xerox machine. A foundries experience with what they are making and how that interacts with the processes used in the foundry makes a difference in yields. Yields differences make a difference in cost.
One contributing factor to why Intel is out in front on process tech is that they don't try to make their process work better for anything but exactly what they make. Generic "all comers" foundries have their own quirks and are no where near as optimized.
ARM's basic logic layout designs has assumptions built into it. Those probably don't match Intel's because for the most part Intel doesn't share what its assumptions are.
In short, there is a real information feedback cycle that flows between the designs and the foundries for the more advanced offerings.
P.S. There are zero indications that Intel is out shopping as a "bring me anything" foundry. Not at all. All of that is just hand waving analyst wet dream material.
MBAN.... More boring Apple news....
Honestly, who cares what size the chip is? It's just an old chip in a machine that streams Netflix and iTunes.
Honestly, who cares what size the chip is? It's just an old chip in a machine that streams Netflix and iTunes.
Apple should buy as many components as possible from Samsung at commodity prices. The USA should buy as much oil as possible from other countries at commodity prices. The profit is in the value add.
And then we have our own oil when the others run out, and they are.
Rocketman
And then we have our own oil when the others run out, and they are.
Rocketman
The Sammy would loose a LOT of business, after all if they will treat their biggest customer like that their smaller ones will be treated worse. Sammy will suddenly become untrustworthy as a business partner (more so than they are now).
Disappointing to see Eric Slivka using slang words like fabbed.
Some of the writing on MacRumors is fairly poor, but I've always found Eric's articles to be very well-written.
I guess this is a true single core A5. Even, then there are likely even more cutbacks to achieve the 50% size reduction since CPU cores and even GPU cores aren't big enough to make up the difference.
Kind of quirky to label it A5.
Perhaps same general functionality but a hand tuned transistor layout for most of the sub-components for the CPUs but still same tweaked ARM A9 basic functionality. Maybe a more efficiently implemented GPU that takes less space. in that context it would make sense to keep the A5 label but different.
They could dump a core if though that AppleTV didn't need it. ( and anything else that isn't used on an AppleTV. ) Effectively an optimized for only AppleTV SoC. As the volumes get bigger some of these products can get their own SoC if Apple has enough designers on staff and the optimizations pay off in lower foundry costs.
Same process means this will be cheaper to make on very mature processes (i.e.,g very high yields. ).
----------
Frankly that cuts both ways. Apple going patent troll mode and suing every tech company on the planet isn't really helping them long term either. Nor is firing/terminating Samsung subsidiaries because they do an above average job of fulfilling orders.
If Apple wants to sit at the "big boy" table then big companies have overlap. That is just a reality.
Neither one of these two companies would be helped economically to cut the other off "cold turkey".
Could you enlighten me? If I had to guess, it means "selling old crap to get your pipeline ready for new stuff".
That seems the logical inference.
It is probably also the case that Apple are experimenting with alternative ways to achieve certain functionality (IO, cache, memory controller, etc) at lower power, lower area, or higher performance. Obviously the sensible thing to do with the first version of the A5 is to utilize the safe version of each piece of functionality, but later, when you are not under time pressure, there is the chance to try out whatever better ideas you have had.
My guess is that all these A5 variants are not the long-term game; they are all ultimately learning experiences. I'd imagine the longer term game involves the same sort of thing (a constant stream of ever smaller, lower power, but superficially similar chips), but based not on A5 but on A7 (or whatever the 64-bit version of A6 will be). Going to 64-bit ARM allows Apple (just like it has allowed ARM) to strip out bits of the ARM spec that are no longer useful, and obviously Apple wants that transition over-and-done as soon as possible, so it only has to maintain one OS and one toolchain (just like they hurried along the OSX 64-bit transition as rapidly as possible).
So I expect we'll see more of this sort of thing over the next few years (perhaps in A6 variants once the iPhone 7 launches), all as "secondary" chips, for low-volume items, and without time pressure, all exploring aspects of the design space which may not seem sexy ("ho hum, the new L2 cache design is based on a flubar circuit with non-MRU replacement algorithm; who cares") but acting as a way for Apple (which let's remember, is very new to the custom CPU game) to refine its parts as it were, in parallel, even more so than a Tick-Tock like strategy.
A comparison would be to the venerable PPC 750 CPU (which, back in the day when Apple used it, was called the G3 by Apple). IBM STILL ships that part, which has gone through a dozen die shrinks and tweaks of one sort or another. My guess is that that's the sort of place Apple will strive to be in, once they have their 64-bit ARM part nailed down --- something that works well, has been optimized to the max, and that can just be cranked out at minimal cost by the billions.