Other A15 Bionic - is it a disappointing year for Apple Silicon?

[kiranmk2]

After watching the Apple Keynote I've been looking at what Apple didn't say (or what I tihnk they accidently said). One thing I noticed was what they didn't say about the A15 chip. When they detailed the A15 chip in the iPhone update they only compared the performance improvement to competitors rather than earlier A chips. They did possibly let the cat out of the bag when introducing the iPad Mini: in that reveal they said that the new iPad mini had 40% faster CPU and an 80% faster GPU than the previous version. The previous iPad Mini used an A12 Bionic chip. From What Apple have said previously I believe the A14 had a 40% CPU improvement and 30% GPU improvement over the A12. To me this is saying that there is basically no CPU improvement from A14 to A15 and about 40% improvement in GPU performance. The main improvement was to the neural engine.

I'm sure that there will be some other improvements due to the extra cache etc but it seems like this could be another A8 (technically faster than the previous chip but much less of an improvement than the A7 or A9 were).

 

[MrKennedy]

Don’t worry about it.

 

[slippery-pete]

Laps the competitors.

 

[now i see it]

The previous A14 has more power than is necessary as it is. At this point "faster" isn't "better" anymore when the previous generation wasn't slow by any means.

 

[man4857]

"Disappointing" chip designs means that the expectations was set too high or not realistic.
The A-series chips already has the best performance and power efficiency of any mobile SoC design out there. (Find any comparable competitive phone - similar screen size and feature set and compare the size of the battery vs the rated battery life and the performance you'd get via single-thread benchmarks). iPhones does what it does best - do something fundamentally a smartphone would do - call, text, web access, photos/videos, and music very well and efficient.

 

[DSTOFEL]

…….. seems like this could be another A8 (technically faster than the previous chip but much less of an improvement than the A7 or A9 were).

Maybe similar also to the small gains in the A12x in the 2018 iPad Pro to A12Z in the 2020 iPad Pro.

 

[NoBoMac]

At this point "faster" isn't "better" anymore when the previous generation wasn't slow by any means.

This. We might be seeing something similar to desktop processors: GHz "wars", then cores. Camera sensors and mega-pixels. And like those, Apple might have decided that 2x, 3x speed "improvements" have little value-add at this time, so, concentrate on improvements elsewhere, like power consumption (speculation here re: power).

 

[rui no onna]

After watching the Apple Keynote I've been looking at what Apple didn't say (or what I tihnk they accidently said). One thing I noticed was what they didn't say about the A15 chip. When they detailed the A15 chip in the iPhone update they only compared the performance improvement to competitors rather than earlier A chips. They did possibly let the cat out of the bag when introducing the iPad Mini: in that reveal they said that the new iPad mini had 40% faster CPU and an 80% faster GPU than the previous version. The previous iPad Mini used an A12 Bionic chip. From What Apple have said previously I believe the A14 had a 40% CPU improvement and 30% GPU improvement over the A12. To me this is saying that there is basically no CPU improvement from A14 to A15 and about 40% improvement in GPU performance. The main improvement was to the neural engine.

I'm sure that there will be some other improvements due to the extra cache etc but it seems like this could be another A8 (technically faster than the previous chip but much less of an improvement than the A7 or A9 were).

Caveat here, what are the implications for upcoming Apple Silicon Macs? Does this mean there won't be significant improvements over the M1?

 

[Cayden]

It seems like they used the new fab size to optimize battery performance rather than optimizing for speed. I personally prefer that, since speed improvements are much less noticed than battery improvements these days

 

[mprime_17]

I would be quite happy if they went with the efficiency route this time.

 

[kiranmk2]

Don’t worry about it.

Lol - absolutely right, but I am interested about this. It's commonly said there is never a good to to upgrade as there is always something better around the corner, but I disagree with this - I think it's better to upgrade in a year with a big improvement from the previous model than a year with small improvements.

While I'll reserve judgement until the phones are out in the hands of users, at the moment if someone with an iPhone 7/8/X asked whether they should upgrade I'd suggest either picking up a discounted iPhone 12 (new form factor, 5G, OLED displays on non-pro models) or waiting another year rather than getting something from the iPhone 13 range.

 

[Bodhitree]

It seems like they used the new fab size to optimize battery performance rather than optimizing for speed. I personally prefer that, since speed improvements are much less noticed than battery improvements these days

That sounds very possible, especially as they announced longer battery use times as well.

I also noticed that they didn’t announce any speed increases compared to the previous generation, which they usually do. It really does pose a question about what the silicon design teams are working on, because 2x larger cache is very much a ‘meh’ upgrade. Previously the iPhone would often debut new generations of architecture, but I’m wondering whether Apple is acknowledging that the focus has shifted towards the M2, and that for the iPhone it is hard to leverage additional speed into features that people genuinely care about.

I thought it was interesting that they spent so much time looking at Center Stage, which is a cool new feature, rather than chip and performance improvements.

 

[Taz Mangus]

This is a pretty good analysis on the A15. Apple is very deliberate in what they do. What is the most interesting part is Apple is designing the efficiency cores to be so good in performance that in most cases the performance cores don't need to be started. This saves a lot on battery life.

 

[KittyKatta]

I’d love to see what people obsessed with specs actually do on their phones.

 

[GoodGuy313]

They usually do talk a lot about the A chips. To me it might be more of a marketing strategy so that next year it will feel like “oh, you didn’t notice any performance improvement last year, here’s a 30% improvement in CPU!” and be more eye candy.

It’s also worth noting that now, as attention regarding processors/gpu has been centered on the M1, the A chips this year have been getting a ‘back-seat’ treatment.

 

[now i see it]

I’d love to see what people obsessed with specs actually do on their phones.

They run benchmarks

 

[rui no onna]

They usually do talk a lot about the A chips. To me it might be more of a marketing strategy so that next year it will feel like “oh, you didn’t notice any performance improvement last year, here’s a 30% improvement in CPU!” and be more eye candy.

It’s also worth noting that now, as attention regarding processors/gpu has been centered on the M1, the A chips this year have been getting a ‘back-seat’ treatment.

I don't think it's a marketing strategy. I think they just simply can't boast about major performance gains when there is none.

The A14 3.0 GHz is 1600 single/4200 multi. If you increase clock rate to 3.2 GHz which is what the A15 has, then that's 1700 single/4500 multi assuming linear gains with clock rate. That's pretty much on par with A15. It seems to me they mostly focused on getting power consumption down.

Mind, getting 2018/2020 iPad Pro performance or better with a lower power iPhone chipset is an impressive feat.

That said, I know Apple's way ahead of the competition but so was Intel at one point in time and look where Intel is now.

 

[Bstephens]

I don't think it's a marketing strategy. I think they just simply can't boast about major performance gains when there is none.

The A14 3.0 GHz is 1600 single/4200 multi. If you increase clock rate to 3.2 GHz which is what the A15 has, then that's 1700 single/4500 multi assuming linear gains with clock rate. That's pretty much on par with A15. It seems to me they mostly focused on getting power consumption down.

Mind, getting 2018/2020 iPad Pro performance or better with a lower power iPhone chipset is an impressive feat.

That said, I know Apple's way ahead of the competition but so was Intel at one point in time and look where Intel is now.

Yeah AMD is pwning Intel.

 

[GoodGuy313]

I don't think it's a marketing strategy. I think they just simply can't boast about major performance gains when there is none.

The A14 3.0 GHz is 1600 single/4200 multi. If you increase clock rate to 3.2 GHz which is what the A15 has, then that's 1700 single/4500 multi assuming linear gains with clock rate. That's pretty much on par with A15. It seems to me they mostly focused on getting power consumption down.

Mind, getting 2018/2020 iPad Pro performance or better with a lower power iPhone chipset is an impressive feat.

That said, I know Apple's way ahead of the competition but so was Intel at one point in time and look where Intel is now.

Agree. What I meant by marketing strategy is that Apple likes to leave some time between updating a different part of the product. For example, the big feature of the iPhone 11 Pro was the cameras (with some other features), and while the 12 Pro had some camera improvements, that upgrade was more focused into a redesign and 5G. This year the focus comes back on the camera (mostly). Another example of this is the 2020 iPad Pro that had the A12Z which was, arguably, a minor upgrade but of course, the iPad is already ridiculously powerful so there is no need for more performance. Then Apple updates it in 2021 with the M1, boasting on performance.

This way, the yearly upgrade seems more appealing.

 

[MacWorld78]

Just remember that lots of people work from home especially who work for apple campus because of coronavirus situation between 2020 to 2021, Whatever apple did this time it must be fantastic performance - never have higher expectation from Apple as you know the design patents and lots confidential information must be deal at the Campus not at home.

Just relax it's just a phone If you think it is minor/major upgrade it's all depend what iPhone you come from, Learn to enjoy the life more than material.

 

[kiranmk2]

I do wonder whether Apple is turning their attention more to the M chip performance. It seems like the next gen chips (whether M1X or M2) are slated to use more cores, so the focus for this year has to be getting the power consumption per core down so as to not impact the battery life seen with M1.

 

[DanTSX]

I’m not expecting anything but incremental improvements at this stage of the game. It’s the iPhone 13 not the iPhone 3. These things are already faster than almost anything else put there, and will again get updated in 12 months anyways.

 

[dallegre]

The brain drain that reportedly lead to the marginal gains with the A15 aren't really a concern this year since Apple is in such a strong position with their chips at present. It could become a bigger concern in future years if it leads to a long term slowdown in Apple's pace of chip innovation, esp with the move to the M series.

 

[JulianL]

This is a pretty good analysis on the A15. Apple is very deliberate in what they do. What is the most interesting part is Apple is designing the efficiency cores to be so good in performance that in most cases the performance cores don't need to be started. This saves a lot on battery life.

I wonder where that could lead to next. It seems to me that the efficiency cores could get sufficiently good that even they become overkill when composing a text message, reading an ebook, reading a news article or many other tasks where a glacially slow human is staring at an unmoving page for what, to a computer, feels like a few aeons of time before doing something like scrolling down or turning a page or typing the next character. Might we one day see a three-level cascade of core power - performance, efficiency and hardly-trying-at-all cores?

 

[3Rock]

Isn’t the iOS iPhone and iPad fast enough for everybody already? So a small incremental increase in speed should be more than enough for everybody. I have no complaints and not disappointed.