Apple's A8 may integrate DRAM into the chipset

Apple's A8 may integrate DRAM into the chipset
The next round of Apple devices isn't expected to hit the market until September, but that means Apple has to finish up designing the chip and sorting out the orders for the companies that will produce and package the chips. A new report makes it sound like Apple may be attempting to bring the upcoming A8 closer to being a true "system-on-a-chip" (SoC).

The A7 SoC included the CPU, GPU, cache, image processor, and "secure enclave" to store TouchID data, and moved the accelerometer, gyroscope, and compass processing to the M7 motion coprocessor. Presumably, there will be an M8 motion coprocessor, but word has it that in addition to everything you'd find on the A7, the A8 will also include the DRAM for the device. 

The same report from DigiTimes says that Apple has chosen Amkor Technology and STATS ChipPAC to each package 40% of the A8 orders with Advanced Semiconductor Engineering taking the last 20% of the packaging orders. Interestingly, the report closes by saying that "TSMC reportedly will start ramping up production using 20nm process technology for Apple's A8 chip in the second quarter of 2014." The interesting thing there is that the wording doesn't say that the A8 will begin production in Q2, but that TSMC will ramp up production using the same 20nm process. That could just be odd wording from DigiTimes, because it would make sense that the A8 ramps up production in Q2 so the devices can be built in time for a late Q3 release. 

source: DigiTimes



1. teja171 unregistered

Its hard to believe that Apple's dual-core chips are better than Qualcomm's quad-core ones.

2. Berzerk000

Posts: 4275; Member since: Jun 26, 2011

They're using different generations of technology, of course it would be better. Any dual core ARMv8 processor would be better than a quad core ARMv7.

12. jove39

Posts: 2148; Member since: Oct 18, 2011

"Any dual core ARMv8 processor would be better than a quad core ARMv7". Not's swift cpu core is significantly larger than qualcomm's krait400 cpu large that it's performance compares with intel's x86 cores. Apple's swift core performance comfortably sits between baytrail and haswell (more on haswell side) has done a wonderful job with swift cpu core...even arm's A57 cpu core trails behind swift...if compared for per clock performace...only challenger for swift cpu core is nvidia's denver cpu core. Edit - I have a Nexus5 and love performance of S800...but with heated rumors of new iPhone with 4.8" screen...I am very inclined to own iPhone6 this year.

17. teja171 unregistered

Yeah, i think a denver cpu is way better than both apple's and qualcomm chips.

24. eggimage

Posts: 83; Member since: Jan 19, 2012

@jove39: i like rational and informative comments like such

29. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

jove39, I normally appreacite your comments and I agree in the general statement you are doing. But this time you did two mistakes and one quite big. You are referring to the "Apple Cyclone cores" not the "Apple Switft cores" as it's all over your comment. The Switft are just around as powerful as Krait 400, far from Haswell. The second one is not as bad: BayTrail are not "cores", they are SoCs. A7 is equivalent to BayTrail (SoCs) Cyclone is equivalent to Silvermont (cores)

33. brrunopt

Posts: 742; Member since: Aug 15, 2013

Execp it doesnt , the A7 only gets a higher score in single core benchs (since is has only 2 cores it has more performance per core than quadcores). Also most of the bench are browser tests , witch are COMPLETE SH*T ... Just using a different browser will give you completely different results...

37. Finalflash

Posts: 4063; Member since: Jul 23, 2013

Any sources on that stuff because Haswell is leagues beyond A7 no matter what you test on. The A7 is as large as the quad-core qualcomm chip even being a dual core. This means Apple just doubled the size of the chip internally and that brought it level with the 800 on performance (they win single core but lose multicore).

47. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

Is this one prove enough for you? It´s the less powerful of the Haswell chips, but it´s clocked at the same frequency as the A7 and guess what? The performance is quite similar too. An A7 wins big time in single core and loses by a tiny bit in multicore agains the SND800. You forgot to mention that too.

4. ScottSchneider

Posts: 336; Member since: Dec 06, 2011

It is actually not IMO... Its because of the Software optimization in iOS... This is what u get when u lock the OS ecosystem so it is not slow to the Hardware that is available... U really dont require Quad cores to run iOS. But when it is open and that is what What the customer really needs,. U get what is known as Android...

6. darkkjedii

Posts: 31529; Member since: Feb 05, 2011

Excuses excuses.

7. GeekyGene

Posts: 88; Member since: Aug 25, 2013

Your comment gave me cancer.

14. jove39

Posts: 2148; Member since: Oct 18, 2011

I hope you are

8. Bioload25

Posts: 213; Member since: Nov 12, 2012

Thats not the point,bro see multi-platform benchmarks,Apple did a very good job with A7 and its a very capable processor by itself,it doesnt have anything to do with the OS Its true that in system operations iOS is more fluid and less laggy than Android,its basic requirements are far lower than android but talking about the hardware itself,A7 is indeed on par with Qualcomm's chips

10. Taters

Posts: 6474; Member since: Jan 28, 2013

Actually does, the z compact pretty much beats it in everything, just look at the performance section of the 5s vs the z compact. The only one that the 5s wins, sunspider, is a total software win. Apple is the master of hiding behind safari and low screen resolution to achieve nice looking benchmarks. In fact, they are masters of deception in all aspects.

18. teja171 unregistered

Well, safari has no more likes to the taste of chrome.

22. Taters

Posts: 6474; Member since: Jan 28, 2013

Run sunspider on chrome on both, which is probably the fairer test, and I am positive the z1 compact would win the sunspider benchmark too. Chrome benches like s**t.

30. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

Sory, but in that comparison there are only graphic benchmarks. And Adreno 330 has proved a worthy contender against PowerVR G6430. But in the only CPU one the A7 sweeps the floor with the Snapdragon 800 (being Sunspiede lightly threaded and with all software optimizations and all you want). Here you have more CPU benchmark examples: In any pure CPU bench the A7 beats Snapdragon, in particular in "single thread performance" You can also look to basemark OS in the "system group 1.0" part and check that the iphone 5S is best of all.

38. Finalflash

Posts: 4063; Member since: Jul 23, 2013

Yea only in single treaded benches, it got demolished in the multithreaded ones (see 3d mark physics) . And even on the single threaded one where all they do is load a thousand Web pages, those pages being rendered on the iPhone mean lower resolution images and text being rendered, if you extrapolate for the resolution difference they lose on those too.

40. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

Sorry, but geekbench does a lot more than loading web pages and it is not influenced by the resolution of the screen as it's a pure CPU test. And the result is the same: better single core performance than any other SoC. And the 3d mark physics test was obviously not working OK with the A7: it gives lower results thatn with the A6 when it has been proven once and again it nearly doubles the performance. Quote from AnandTech: "Note that I saw no improvement in this largely CPU bound test, leading me to believe that we've hit some sort of a bug with 3DMark and the new Cyclone core" I encorage you to read and not only look at the graphics:

42. Taters

Posts: 6474; Member since: Jan 28, 2013

Browser tests are hardly real CPU tests. Junk like sunspider and browser mark is the cheating Apple way if judging CPU speed.

44. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

OK, if you think geekbench is based in a browser test we can stop the conversation here. You just demostrated your lack of knowledge about benchmarks.

32. 1701nino unregistered

If that is true that you say apple is hidding behind "low resolution" screen on the iphone what about the ipad? The ipad has the the almost the same A7 proccesor but with a resolution 2048x1536 and it rules benchmarks,so i think there is no argument about A7 it is clearly a beast.

41. Taters

Posts: 6474; Member since: Jan 28, 2013

You scale up accordingly. The ipad is clocked higher but most importantly it doesn't need to throttle at all to accommodate a 1500 mah battery. If you scale the S800 accordingly, bump it past 2.3 ghz and tailor it for a fat 8000 mah battery you will get the same performance increases regardless if it's an A7 or a s800 or some crap mediated chip. Bottom line is that they are pretty on par overall but 4 cores should in theory make the S800 more powerful overall. You can argue that the A7 makes better use of its power because mobile OS would make more use of 2 cores but if you somehow put windows on it where the 4 cores might actually do something then the S800 should be more powerful because if the quad cores.

25. xfire99

Posts: 1207; Member since: Mar 14, 2012

A7 isnt better the Qualcomms Quad, in singel core perfomance A7 is faster and in multicore qualcomms is faster. ex:

31. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

Sure, but single core perfomance is the most important one of the two to know how fast a CPU is. If you like it is the one tha has more to do to the "user experience".

34. brrunopt

Posts: 742; Member since: Aug 15, 2013

what ?? lmfao ....

35. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

May I ask what do you find funny in the comment?

43. Taters

Posts: 6474; Member since: Jan 28, 2013

Um you can't just ignore cores. Single core is most important with single core vs single core comparisons but when you are comparing a dual core with a quad core, it is trickier than that. Lol That is why he was laughing at your pathetic comment. There is no absolute fair way to compare them but let's just say for a tinkerer the S800 should be better if they can find a way to maximize the performance of the 4 cores. User experience is a whole other debate but as for as pure muscle goes, the quad usually wins unless the cores are generations apart like the pentium quads vs the core 2 duo where the core 2 duo wins hands down. This one probably leans towards the s800 despite what the ifans want you to believe. If you ignore operating systems and actually max out the power of each core the S800 quad should win overall. That is how you judge CPU power, with no artificial limits, not just looking at single core performance and ignoring multidirectional performance. I mean if you can just ignore it and look at single core performance, why even make quads at all? Lol

46. TylerGrunter

Posts: 1544; Member since: Feb 16, 2012

I don´t even know where to start: 1) I´m not igoring cores, just saying that I take a couple of better performance cores over quad SoCs with lowe performing cores. Not just me, anyone who trully understands how coputers/smartphones works. If the cores are of similar performance of course I take the quad cores. 2) Most applications in mobile are NOT optimized for high number of cores, making my point 1 even more valid. 3) Because they are not optimzed the "user experience" is better with a couple of higher performance cores. 4) Precisely the case comparing Cyclone with Krait 400 is what you mention: comparing Pentium cores with Core2. Cyclone are much wider (suspected 7-wide front end compare to the Krait 400, 3-wide) and use ARMv8, pretty much doubling the IPC of Krait 400 (not the performance, as they are clocked at lower frequency) 5) Quads (and octas) are being made for 2 reasons. The most important is MARKETING. To sell to the ignorants. The second one is technical, you don´t always have a more powerful core or you can´t implement it in the current manufacturing node. You can´t scale frequency forever either due to power consumtion constrains. The solution: add more cores. But that doesn´t mean it´s the best solution. If you have more questions feel free to ask, as it seems you are quite green on the matter. To end it, a comparison Snapdragon 800 and A7 (again, geekbench is native code, no browser or software other than the native code involved): As you can see: A7 has a much higer single core performance, and nearly identical multicore. So I would take that any time. Yours faithfully, a proud owner of a Nexus 5 with Snapdragon 800 inside.

Latest Stories

This copy is for your personal, non-commercial use only. You can order presentation-ready copies for distribution to your colleagues, clients or customers at or use the Reprints & Permissions tool that appears at the bottom of each web page. Visit for samples and additional information.