A10 does appear to have higher IPC than Skylake, but remember that Skylake is a design that scales to well past 4GHz. Can the A10 CPU do the same?
According to geekbench the A10 have higher ipc compared to skylake, i wonder why they do not make a desktop version since it would be faster and use it for they own mac family instead of using intel cpu's..
The A10 fusion is a performance beast in both CPU and GPU metrics. Apple is embarassing Qualcomm in CPU performance.
According to geekbench the A10 have higher ipc compared to skylake, i wonder why they do not make a desktop version since it would be faster and use it for they own mac family instead of using intel cpu's..
A10 does appear to have higher IPC than Skylake, but remember that Skylake is a design that scales to well past 4GHz. Can the A10 CPU do the same?
I'm not knowledgeable enough to know how much of an arquitectural change adding SMT would mean. But it would make sense to add it to an A?X variant in a future iPad pro, specially if it requires extra die space.
Although who knows, if the A10X ends up using TSMC 10nm next spring maybe they'll cram 4 full big cores in there instead of a big.little config.
I'm not so doubtful anymore given the 30% increase in clocks they got on the same node, what next on 10nm: 3GHz? That and higher than Skylake IPC (if true) would make an hell of a laptop cpu, then double/triple the cores and it reaches easily the same overall performance even without SMT.
What I heard about SMT die increase is around 5% when Intel added it back on Nehalem, there's many implementations though and likely IBM's power uses a lot more than 5% (but it also gets a 2x speedup rather than 1.25x :O).
A10X is very interesting at this point... if Apple goes with the route more cores again I think they'll use 3 like with A8X, but considered 10nm should grant some clockspeed boost and more cores isn't their style a 2.6-3.0 GHz dual core is more plausible. (without the small cores because the benefits might be smaller in a tablet, same way A9 has an extra 4MB victim cache to save power but A9X doesn't)
Heck that thing might reach 4200 points on GB4 single! Isn't a 6700K stock barely over 5000?
The power efficiency gains at 10nm are not really great (18% over 16FF+) so the clocks will increase only marginally like we saw from A7 at 28nm to A8 at 20nm. But at TSMC 7nm we will see a good increase in clocks. Apple will break the 3 Ghz barrier at TSMC 7nm. Given the clock speed increase from A7 at 1.4 Ghz to A10 at 2.34 Ghz (mature 28nm to mature 16FF+) its going to be exciting to see how much Apple can push clocks from TSMC 16FF+ to TSMC 7nm. The 2018 A12 and 2019 A13 chips are going to be pretty much notebook/desktop class CPUs (even compared to Kabylake / Cannonlake).
At 10nm I expect marginal clock increase just as we saw from A7 to A8. At 7nm the jump would be substantial and I expect Apple to break the 3 Ghz barrier in a smartphone.
10nm is mainly a cost reduction node. There is some performance enhancement, but it's not the kind of thing that's going to make something like Kryo or A73 come within A10 Fuson's orbit, IMO.
Architecture > Process. Always.
The way Apple is going right now there is only one meaningful comparison . Apple Ax vs Core M. At TSMC 7nm Apple has an opportunity to design a core which can scale from 3 to 4+ Ghz and serve their entire product stack from phones to desktops and basically replace Intel as their CPU supplier for iMacs and Macbooks. TSMC has a high performance variant at 7nm so if Apple chooses they can have two designs - one optimized for perf/watt and low power using a low power process and the other optimized for high performance and frequency using the high performance process.
http://www.eetimes.com/document.asp?doc_id=1329217&page_number=2
"The 7nm platforms will have their own development kits, EDA tools and IP blocks. In addition, companies such as ARM and Imagination Technologies will provide IP reference subsystems and optimized processors to speed SoC integration and verification, said Hou.
The HPC variant will make chips that run 10-15% faster than those made in the mobile process, enabling clock trees optimized to hit 4 GHz data rates. The variant will include optimized SRAM compilers supporting L3 caches and taller standard cell libraries. Early versions of the tools and IP blocks will be available later this year."
There's nothing in theory that would stop Apple from going to 3GHz today. If they need 7nm for that so their chips don't become furnaces, that just shows how far ahead Intel is, given they're doing 3.6GHz boost at 4.5W TDP today.At 10nm I expect marginal clock increase just as we saw from A7 to A8. At 7nm the jump would be substantial and I expect Apple to break the 3 Ghz barrier in a smartphone.
There's nothing in theory that would stop Apple from going to 3GHz today. If they need 7nm for that so their chips don't become furnaces, that just shows how far ahead Intel is, given they're doing 3.6GHz boost at 4.5W TDP today.
Well, there is a reason those Intel 4.5 W TDP chips are not labelled 3.6 GHz chips. IOW, you're basically contradicting yourself there.There's nothing in theory that would stop Apple from going to 3GHz today. If they need 7nm for that so their chips don't become furnaces, that just shows how far ahead Intel is, given they're doing 3.6GHz boost at 4.5W TDP today.
Not good cross platform, but one might say they could be basic intra-platform comparators.A kind reminder to stay away from Ars' Javascript benchmark methodology and numbers.
A kind reminder to stay away from Ars' Javascript benchmark methodology and numbers. Only they could end up with a 300% discrepancy from reality such as on the S7 in this chart.
Even intra-platfom, if you don't retest the older iPhone's with the newer OS (and browser) the comparison makes little sense.Not good cross platform, but one might say they could be basic intra-platform comparators.
True. However, I don't know what OSes he had installed. However, iOS 10 is now out officially, and the GM has been out for a week already.Even intra-platfom, if you don't retest the older iPhone's with the newer OS (and browser) the comparison makes little sense.
I'd say it does matter, although it may only be relevant in the long term. Let's say the gulf between Apple's chips and Qualcomm's is large for a long time. Customers notice that hey, the iPhone users are doing things that our phones can't do.
Qualcomm is a small fry compared to Apple, not sure why this is a surprise
There isn't a whole lot that you can do with an iPhone that you can't on an Android phone. You could argue that iOS runs a little smoother, but I think that has more to do with Android than the SoC as they were designed differently such that even if you put Android on an Apple SoC, it wouldn't fix the fundamental underlying design choices that lead to that difference. Otherwise at best you get something that's slightly faster on the iPhone, and unless you're really pegging the SoC, it won't be human noticeable. Look at some video speed test comparisons where they open apps and do simple tasks. In most cases the results are so close that it's imperceptible or may be more of a result of slightly different start times. Beyond that you have gaming, which doesn't matter too much when the most popular games are things like Angry Birds or Candy Crush that don't require much computational muscle.
Where this difference might matter is with tablets where people are starting to use the kind of apps that can tax the CPU/GPU to a larger degree. I think that this explains why Android can be the leader in mobile phones, but more of an also ran in the tablet space.
Apple's aggressive approach to SoC performance suggests that they want to try to make their own processors for all of their products lines at some point. They don't care about competing with Qualcomm or Samsung as much as they want to be able to beat Intel, at which point they're that much closer to vertical integration for their traditional computing platforms.
Maybe that speeds are being targeted to the iPad. Also... why not start to put Macs with ARM processors?Sure there is, the design probably can't handle running at 3GHz without consuming way too much power, if it can even hit those frequencies at all.
Anyway, A10 is a wider/higher performance/MHz design than any Intel core today, at least in non-SIMD applications (aka most client applications). So straight up comparing frequency and claiming one company is way further ahead than the other is ridiculous.
Look at delivered performance. If you look at it objectively, you will see that Apple is delivering Skylake Core m class performance in a handheld smartphone, an impressive achievement.
Intel is really good at what it does, and Apple is not a competitor to Intel across the board, but when it comes to CPU performance in a ~4W power envelope, Intel doesn't have that big of an advantage.
There isn't a whole lot that you can do with an iPhone that you can't on an Android phone. You could argue that iOS runs a little smoother, but I think that has more to do with Android than the SoC as they were designed differently such that even if you put Android on an Apple SoC, it wouldn't fix the fundamental underlying design choices that lead to that difference. Otherwise at best you get something that's slightly faster on the iPhone, and unless you're really pegging the SoC, it won't be human noticeable. Look at some video speed test comparisons where they open apps and do simple tasks. In most cases the results are so close that it's imperceptible or may be more of a result of slightly different start times. Beyond that you have gaming, which doesn't matter too much when the most popular games are things like Angry Birds or Candy Crush that don't require much computational muscle.
Where this difference might matter is with tablets where people are starting to use the kind of apps that can tax the CPU/GPU to a larger degree. I think that this explains why Android can be the leader in mobile phones, but more of an also ran in the tablet space.
Performance is not just the only goal for Apple having a brilliant in-house SoC design team. It's also a carefully planned out move to retain as much control over their end product. It's amazing how hardware nerds focus so much on benchmarks that they don't get the latter.
If I'm Apple I wouldn't want to source SoCs from Intel or QC either when there are so much conflicts of interests working against me.