“In a note out late yesterday, Robert Maire of boutique research firm Semiconductor Advisors follows up on some positive observations about Apple’s ‘A8’ chip, introduced last week along with its iPhone 6 models,” Tiernan Ray reports for Barron’s. “In an initial note published on Friday, Maire observed that Apple’s stated specification of 2 billion transistors in the A8 is more than the 1.4 billion transistors in Intel’s ‘Haswell’ desktop chip.”
“He also notes Apple’s claim of a 25% performance improvement over its A7 chip in the iPhone 5S, a claim of 50% GPU performance increase, and a claim of the A8 being ‘50% more power efficient,'” Ray reports. “Observed Maire, ‘This is more than double the normal gain of 15-20% expected from a shrink to 20nm technology that A8 is based on. This suggests that Apple made some significant design improvements on top of just the hardware shrink.'”
Ray reports, “Maire, noting that such processors could take over some ‘light’ application tasks in a notebook like the MacBook Air, wondered ‘When will Apple start building its own laptop and desktop processors?’ While he doesn’t think Intel will lose Apple’s desktop and laptop business ‘anytime soon,’ he opined, ‘It should also be viewed as a threat to Intels entry into the tablet and smartphone market. Apple has set a very high bar in performance that it seems would be very difficult for Intel to match on a price/performance basis without substantial subsidies (contra revenue…).'”
Read more in the full article here.
Related articles:
Apple’s A8 processor launch and its impact on Intel’s mobile CPUs – August 27, 2014
Intel-powered Macs: The end is nigh – August 4, 2014
Intel’s Broadwell chips further delayed; not shipping for most Macs until early-mid 2015 – July 9, 2014
Apple will inevitably drop Intel for their own A-series processors in the Mac – June 26, 2014
How long before Apple dumps Intel from MacBook Air? – June 26, 2013
I’d also like to know how they compare with the snapdragon chips
We’ll never know unless that Snapdragon can run iOS. I’m sure Apple must be fine-tuning the processor for running iOS and vice-versa. Benchmarks are a waste of time because every processor I’ve seen seems to be better in certain benchmarks but weaker in others. Apple builds the iPhone to be an overall good performer and not to necessarily beat specific benchmarks. Some of the fastest Android smartphones stutter at times with lots of system RAM, quad cores and high clock speeds despite having great benchmarks.
Apple is not following the Android bunch because we see how low a clock speed Apple’s A-Series run with dual cores. Yet all the tech heads reason Apple should mimic NVidia, Qualcomm and Intel. Does Apple’s way have theoretical limitations? Those limitations are probably not going to be reached in a smartphone although they might be met in a tablet.
Don’t forget that most android manufacturers got caught cheating on benchmarks. LOL!
As with the RISC/CISC arguments of 15-20 years ago, MHz isn’t what matters most. At last, people are finally recognizing that fact! Good times for Apple.
Intel might be worried if Apple was selling their A8 chip to others. But since they are not, it leaves Intel plenty of customers to acquire.
Intel’s got nothing to worry about. They could lose all of Apple’s desktop business and they’d still be rolling in money. I really can’t see a downside in this for Intel.
Now, if this were Microsoft, that would be a concern for them.
What’s AAPL’s market share again? Taking a hit like that overnight gets noticed, because replacing 8% or 13% business isn’t necessarily all that easy.
But I think they’d certainly take it hard (if just pride) to be excluded from the market leader’s products and working on the market leader’s cutting edge projects. Can’t put a price on that, but it would hurt.
The last news I heard about PC sales was Mac share was growing while Windows machines were shrinking. Losing the 13% that is growing would be a huge loss. This is what is keeping Intel even or growing slowly.
But, you have to agree that losing 87% of your business is far worse than losing 13% of your business.
So this is why they bought PA Semi back in the day.
One of their most brilliant decisions!
Yes, it does.
Most of that A8 transistor count is prolly for the GPU also included.
How many transistors in Intel’s integrated graphics?
… and RAM.
RAM consumes the lion’s share of space on SoC.
Exactly. And that is why transistor count means nothing. Laying out memory is just a step-and-repeat process. There is nothing high tech about that.
Did you know that ‘prolly’ is not a word? Probably, you are from a lower socio-economic class and are not educated and using an Android device that doesn’t have autocorrect.
I had to stop my MacBook Pro from correcting ‘prolly’ to probably 4 times in typing this post.
From Apple’s own dictionary:
prolly |ˈprälē|
adverb informal
probably: you prolly know this already. I am prolly the last person he should be asking for advice.
ORIGIN 1940s: representing an informal pronunciation of probably.
Educated?
School-marmish.
So I wonder what would happen if you put 4 A8 Chips in a Macbook Air? We’ve always bragged about OSXs multiprocessor capabilities would that offer any performance gains over a single Intel i7 Quad core? Seems like you would have a lot more graphics capabilities but I am sure the answer is probably very little gain as most programmers don’t write code that takes advantage of multiple cores or multiple processors.
I think that would be interesting, however the A8 would probably not be the best chip. What you would want is 4 CPU only chips with a separate GPU – OR a massive SoC with 4 – 8 CPUs and suitably scaled GPU.
As for the battery – just get one out of a Mack….
truck 😛
The A8 to laptop/desktop has been well covered here:
http://seekingalpha.com/article/2478635-apples-low-cost-business-mac-pc-implications
Apple is probably rather angry with Intel over Thunderbolt. Even if they don’t ACTUALLY make an A8 mac, the fact that they _could_ do so might give them leverage over Intel.
Why would Apple be angry?
Apple was tired of competing with USB and wanted to join with Intel on the next big thing. After going it alone with Firewire for so long, Apple wanted to be in on a fresh new universal standard so that Macs and PCs could use the same high performance external devices at lower cost.
Firewire devices were needlessly expensive ($30-$50 more than the same device running USB 2.) because of low volume. You had a niche market having to bear all the development cost. (Bizarrely, PCs often DID have firewire, but would mislabel the port so users didn’t know they could use firewire devices.)
Apple and Intel were supposed to introduce Thunderbolt together as the next big thing for PCs. But the instant Apple started shipping Thunderbolt hardware, thus committing itself publicly to the standard, Intel REMOVED Thunderbolt from their reference designs, and started pushing USB 3 instead. As a result, Apple was left in exactly the same situation they were trying to escape with firewire. And thunderbolt devices are even more expensive than firewire!
Apple didn’t need Intel’s help to design a faster standard – Firewire 3200 would have cost less to implement, and been more compatible.
What (aside from cost) is wrong with thunderbolt?
Last year Anandtech’s 5s review included a comparison to Intel’s Bay Trail, against which it made a good showing. I’m hoping for a iPhone 6 review as comprehensive this year.
Still, I doubt Intel has much to worry about. X86 compatibility will remain important, and Apple is unlikely to be selling any chips to any of Intel’s competitors. I think the Apple Intel relationship will be happy and healthy for a long time.
My chip’s bigger than yours. Er… smaller than yours. Er… more transistors than yours. Er… whatever makes me the winner.
Apple is using every processor it makes for itself. Intel should be so lucky. Some damn idiot is always looking for or trying to start a rivalry between companies where none exists. Apple is building a processor specifically for its own product needs. Intel’s Haswell is up for grabs by company willing to buy it. Apple has never claimed its A Series processor was better than a Haswell processor, so I don’t know what all this arguing is about.
But, Apple does continually use the phrase “desktop class”. So, if Haswell can be considered a desktop processor, Apple’s doing some heavy inferring there.
Macs would be much faster if Intel would get off their lazy ass.
apple has tried to ameliorate Intel’s lackadaisical attitude by doing acrobatics with the GPUs in the Mac Pro etc but I wish Intel would get on the ball…
Remember Motorola and IBM, Intel will be a distant memory in 3 years, Apple will move on.
PPC was with us for over a decade and got the boot because no one could create a G5 efficient enough for a laptop. A 13″ Macbook Air can now get 12 hrs of battery life AND run Windows to attract switchers. Intel will be with us for years to come.
“Apple is probably rather angry with Intel over Thunderbolt. Even if they don’t ACTUALLY make an A8 mac, the fact that they _could_ do so might give them leverage over Intel.”
They are even more pissed about having to delay all their new Mac product lines because of Intel’s problems with their next processor introduction (was supposed to be earlier this year and now looks like next year).
I love the discussion here today. It will be some time before even Apple makes a “final” decision, I suspect. But the facts are: They have started a move back in the old PPC direction with ARM. They are making iOS and OSX work together better. Intel hasn’t been as innovative since Apple moved to their processors. The last hurdle is getting over people feeling they need to run Windoze – and they started pulling people away from that with free Notes, Pages, etc.
There are plenty of us who will *have* to keep running Windoze at work who will never even worry about the differences on our home Macs. I’ll be getting an early maxiPad. I would love to then be able to add an A10 (or whatever) based iBook running iOS in tandem to OSX someday. (Oh, and here’s an early rumor – that I just made up, but it has as much validity as most of the analysts’ garbage – it will be called OSXi. 🙂 )