I’m not sure who mentioned it, but a comment was recently made that 75% of App Store revenues were games, therefore they should be used to test performance. The implication is, obviously, that most people don’t do work on their iOS devices and people mostly played games, thus trying to undermine the requirements for fast processors to run powerful Apps.
Last year Adobe had revenues of $7 billion. Autodesk had revenues of $2 billion. Two companies that make expensive, complex software used by professionals.
Activision/Blizzard had revenues of $51 billion. Should I then claim PCs are only used to play games because game revenues vastly outpaced revenues of professional applications?
Gaming is popular and generates a LOT of revenue. You can’t draw conclusions about what people do with their devices just because gaming companies make lots of money.
GG mentioned it. But never mind. Your responses to GG and Morgle in this thread (despite your own incitement) are less than adequate. The examples provided by GG and Morgle are VALID, because those are the ones which are COMMONLY used. If you want to choose a NICHE application for comparison, then you have to "admit" that ALL the ADDITIONAL power that Apple's A series processors is useful for a small set of NICHE users ONLY.
The specific App examples that you gave in one of your comments (video encoding, complex spreadsheet, audio mixing etc) would have come up in the TOP 10 most APPS used list if they are used by "majority" of the users. They are NOT used by majority of the people (I am talking about iPhone users only), hence they are not coming up in the TOP 10 or 100 most frequently used Apps.
Either you have to admit the A series processors performance has reached a "niche" category - much like 18 core Xeon processors in the iMac Pro that is needed only by a small sub segment of Mac owners Vs iMacs with i5/i7 (in mobile terms, the snapdragons and exynos would fit that category) satisfying the needs of common people. Is an 18 core Xeon processor much more powerful than quad core i7? Of course, Yes. Does everyone NEED it? Absolutely NO. Only a small set of niche users need that 18 core Xeon processor.
OR
prove that it is useful for "generic" users with "generic" App/Game examples.
I wonder if there is any consumer facing app to break down the use of the cores over time.
My guess is that for huge usage periods, the performance cores remain idle.
I agree with Gatorguy that I would never spend long periods of time doing photo editing on a phone.
Speed stopped being a deal-breaker in the mid tiers a few years ago. I haven't seen anyone complain about it. That said, things get faster. However, should anyone with an A11 rush out and get a new phone just to have more speed? No. And they don't. Upgrade cycles for iPhones are slowing. People are sticking with their iPhones and (and old processors) for years and not worrying about it.
'Slower' can still mean 'fast' and clearly does for the vast majority of users. In fact, millions of Android users upgraded to ultra premium phones this year and got far more out of the deal than with any iPhone currently available.
That's because processor cores are less important now than before. Far less important.
It impresses me no end that the new meme is that Apple's silicon advantage doesn't matter. Yet I recall you personally touting how great the Kirin 980 was going to be; well of course before the inevitable comparisons. You even attempted to convince yourself that Huawei was first with 7nm silicon, which is actually laughable, given Apple's current iPhone X deliveries. Announced first isn't the same as shipping.
I would note that Apple's advantage in silicon is so vast, that Huawei, Samsung, and Google, are all in the race to build their own silicon. All of those companies are attempting to migrate their customers and their product lines to match Apple's marketing model, and all will fall short.
A processor goes live at its presentation. The Kirin 980 pipped Apple to that post. The A12 pipped Huawei to the post for the first phone to become available on 7nm.
In both cases by very short periods.
None of this admits discussion. The facts are too clear.
In fact both chips were rolling out of TSMC at the same time.
The Kirin 980 is great. Latest variants of virtually everything on board. Including the world's first mobile Cat 21 modem and in house designed ultra fast wi-fi chip.
So, here you are, talking about speed again, after "Speed stopped being a deal breaker in the mid tiers a few years ago"
I’m not sure who mentioned it, but a comment was recently made that 75% of App Store revenues were games, therefore they should be used to test performance. The implication is, obviously, that most people don’t do work on their iOS devices and people mostly played games, thus trying to undermine the requirements for fast processors to run powerful Apps.
Last year Adobe had revenues of $7 billion. Autodesk had revenues of $2 billion. Two companies that make expensive, complex software used by professionals.
Activision/Blizzard had revenues of $51 billion. Should I then claim PCs are only used to play games because game revenues vastly outpaced revenues of professional applications?
Gaming is popular and generates a LOT of revenue. You can’t draw conclusions about what people do with their devices just because gaming companies make lots of money.
GG mentioned it. But never mind. Your responses to GG and Morgle in this thread (despite your own incitement) are less than adequate. The examples provided by GG and Morgle are VALID, because those are the ones which are COMMONLY used. If you want to choose a NICHE application for comparison, then you have to "admit" that ALL the ADDITIONAL power that Apple's A series processors is useful for a small set of NICHE users ONLY.
The specific App examples that you gave in one of your comments (video encoding, complex spreadsheet, audio mixing etc) would have come up in the TOP 10 most APPS used list if they are used by "majority" of the users. They are NOT used by majority of the people (I am talking about iPhone users only), hence they are not coming up in the TOP 10 or 100 most frequently used Apps.
Either you have to admit the A series processors performance has reached a "niche" category - much like 18 core Xeon processors in the iMac Pro that is needed only by a small sub segment of Mac owners Vs iMacs with i5/i7 (in mobile terms, the snapdragons and exynos would fit that category) satisfying the needs of common people. Is an 18 core Xeon processor much more powerful than quad core i7? Of course, Yes. Does everyone NEED it? Absolutely NO. Only a small set of niche users need that 18 core Xeon processor.
OR
prove that it is useful for "generic" users with "generic" App/Game examples.
I wonder if there is any consumer facing app to break down the use of the cores over time.
My guess is that for huge usage periods, the performance cores remain idle.
I agree with Gatorguy that I would never spend long periods of time doing photo editing on a phone.
Speed stopped being a deal-breaker in the mid tiers a few years ago. I haven't seen anyone complain about it. That said, things get faster. However, should anyone with an A11 rush out and get a new phone just to have more speed? No. And they don't. Upgrade cycles for iPhones are slowing. People are sticking with their iPhones and (and old processors) for years and not worrying about it.
'Slower' can still mean 'fast' and clearly does for the vast majority of users. In fact, millions of Android users upgraded to ultra premium phones this year and got far more out of the deal than with any iPhone currently available.
That's because processor cores are less important now than before. Far less important.
It impresses me no end that the new meme is that Apple's silicon advantage doesn't matter. Yet I recall you personally touting how great the Kirin 980 was going to be; well of course before the inevitable comparisons. You even attempted to convince yourself that Huawei was first with 7nm silicon, which is actually laughable, given Apple's current iPhone X deliveries. Announced first isn't the same as shipping.
I would note that Apple's advantage in silicon is so vast, that Huawei, Samsung, and Google, are all in the race to build their own silicon. All of those companies are attempting to migrate their customers and their product lines to match Apple's marketing model, and all will fall short.
A processor goes live at its presentation. The Kirin 980 pipped Apple to that post. The A12 pipped Huawei to the post for the first phone to become available on 7nm.
In both cases by very short periods.
None of this admits discussion. The facts are too clear.
In fact both chips were rolling out of TSMC at the same time.
The Kirin 980 is great. Latest variants of virtually everything on board. Including the world's first mobile Cat 21 modem and in house designed ultra fast wi-fi chip.
So, here you are, talking about speed again, after "Speed stopped being a deal breaker in the mid tiers a few years ago"
To date, I haven't seen any developer state that they had enough performance, memory, storage, battery life, or thermal cooling, available in a smartphone to accomplish what they need to in their market niche.
Yes, NONE of the developers have YET come up with an App that is USEFUL for the masses (hundreds of millions of users) which can be run ONLY with the additional power that A series processers provide, but NOT by mid-range Snapdragon SoCs (Snapdragon 600/700 series, leaving aside the 835/845 for a moment). It is a fact, that smartphone SoCs have reached a level where MAJORITY of the "common" users do NOT NEED the absolute best when it comes to performance. People are buying the latest and greatest phones for MANY other reasons, performance may NOT be the first and foremost reason unlike 3-4 years ago.
I agree that most users don't require that performance, but even you would have to agree that there are benefits to performance in common applicants that are in daily use.
Computational imaging is better and faster, so the user sees little latency in the capture of an HDR* portrait image. Tasks are initiated and completed faster, something that users notice; again the reduction in latency that give a better experience throughout the day.
The fact that iOS on the iPhone X models operates so smoothly is commonly mentioned in reviews, certainly a result of all those additional computations that are available to Apple and its developers.
I'm literally laughing at all of you that are downplaying performance as if it isn't relevant to the user experience, all because the of lesser benchmarks of the Pixel.
If you are satisfied with less than optimal performance, then buy an older iPhone, or buy a flagship Android OS device like the Pixel, because it just doesn't matter to you.
For the rest of us that aspire to the smoothest user experience, those few hundred extra dollars are well spent.
Your milage evidently varies.
Partially agree!!! Agree with first 3 paragraphs of your reply. Diagree with the below:
I'm literally laughing at all of you that are downplaying performance as if it isn't relevant to the user experience, all because the of lesser benchmarks of the Pixel.
If you are satisfied with less than optimal performance, then buy an older iPhone, or buy a flagship Android OS device like the Pixel, because it just doesn't matter to you.
For the rest of us that aspire to the smoothest user experience, those few hundred extra dollars are well spent.
Your milage evidently varies.
My own take on the above points - Smoothest user experience is mostly related to software nowadays than hardware. Hardware (particularly SoC, RAM, Internal Memory and Battery - Yes, battery too in case of iPhones, not in the case of Android phones) has become more than adequate for most of the users, which you also agree. Luckily for Android users, stock Android hasn't been "bloated" (i.e. made worse in performance compared to the previous version of the OS) by Google in the last 3 generations (Marshmallow, Nougat or Oreo). I am making this statement based on my own real world experience with Android mid-range phones (not flagship phones like Pixel or Samsung S/Note phones).
Based on the comments that I have read in this forum, iOS 11 was less than optimal for performance (compared to either iOS 10 or iOS 12) and Apple has worked on improving it in iOS 12. iOS 12 has made more difference to smooth user experience THAN A12 or 4GB of RAM. I don't own any device with either iOS 11 or 12, (my iPad Air is stuck at iOS 10 due to my own decisionnot to upgrade it) so I cannot say it with 100% confidence. Please correct me if I am wrong about this.
To date, I haven't seen any developer state that they had enough performance, memory, storage, battery life, or thermal cooling, available in a smartphone to accomplish what they need to in their market niche.
Yes, NONE of the developers have YET come up with an App that is USEFUL for the masses (hundreds of millions of users) which can be run ONLY with the additional power that A series processers provide, but NOT by mid-range Snapdragon SoCs (Snapdragon 600/700 series, leaving aside the 835/845 for a moment). It is a fact, that smartphone SoCs have reached a level where MAJORITY of the "common" users do NOT NEED the absolute best when it comes to performance. People are buying the latest and greatest phones for MANY other reasons, performance may NOT be the first and foremost reason unlike 3-4 years ago.
I agree that most users don't require that performance, but even you would have to agree that there are benefits to performance in common applicants that are in daily use.
Computational imaging is better and faster, so the user sees little latency in the capture of an HDR* portrait image. Tasks are initiated and completed faster, something that users notice; again the reduction in latency that give a better experience throughout the day.
The fact that iOS on the iPhone X models operates so smoothly is commonly mentioned in reviews, certainly a result of all those additional computations that are available to Apple and its developers.
I'm literally laughing at all of you that are downplaying performance as if it isn't relevant to the user experience, all because the of lesser benchmarks of the Pixel.
If you are satisfied with less than optimal performance, then buy an older iPhone, or buy a flagship Android OS device like the Pixel, because it just doesn't matter to you.
For the rest of us that aspire to the smoothest user experience, those few hundred extra dollars are well spent.
Your milage evidently varies.
Partially agree!!! Agree with first 3 paragraphs of your reply. Diagree with the below:
I'm literally laughing at all of you that are downplaying performance as if it isn't relevant to the user experience, all because the of lesser benchmarks of the Pixel.
If you are satisfied with less than optimal performance, then buy an older iPhone, or buy a flagship Android OS device like the Pixel, because it just doesn't matter to you.
For the rest of us that aspire to the smoothest user experience, those few hundred extra dollars are well spent.
Your milage evidently varies.
My own take on the above points - Smoothest user experience is mostly related to software nowadays than hardware. Hardware (particularly SoC, RAM, Internal Memory and Battery - Yes, battery too in case of iPhones, not in the case of Android phones) has become more than adequate for most of the users, which you also agree. Luckily for Android users, stock Android hasn't been "bloated" (i.e. made worse in performance compared to the previous version of the OS) by Google in the last 3 generations (Marshmallow, Nougat or Oreo). I am making this statement based on my own real world experience with Android mid-range phones (not flagship phones like Pixel or Samsung S/Note phones).
Based on the comments that I have read in this forum, iOS 11 was less than optimal for performance (compared to either iOS 10 or iOS 12) and Apple has worked on improving it in iOS 12. iOS 12 has made more difference to smooth user experience THAN A12 or 4GB of RAM. I don't own any device with either iOS 11 or 12, (my iPad Air is stuck at iOS 10 due to my own decisionnot to upgrade it) so I cannot say it with 100% confidence. Please correct me if I am wrong about this.
Easy way to tell if iOS 12 was the primary generator of optimal performance, is to compare the iPhone X with iOS11 against iOS 12, and this has been done, many times, by many reviewers.
Still, the A12 as reviewed by Anandtech, leaves no doubt that there is substantial improvement in performance, but that should be compared iPhone X with iOS 12 vs iPhone XS with iOS 12, which has also been a commonly reviewed comparison, including just today here at AI.
I would upgrade any iPhone or iPad eligible to iOS 12 if there are no other factors beneficial to not doing it.
I’m not sure who mentioned it, but a comment was recently made that 75% of App Store revenues were games, therefore they should be used to test performance. The implication is, obviously, that most people don’t do work on their iOS devices and people mostly played games, thus trying to undermine the requirements for fast processors to run powerful Apps.
Last year Adobe had revenues of $7 billion. Autodesk had revenues of $2 billion. Two companies that make expensive, complex software used by professionals.
Activision/Blizzard had revenues of $51 billion. Should I then claim PCs are only used to play games because game revenues vastly outpaced revenues of professional applications?
Gaming is popular and generates a LOT of revenue. You can’t draw conclusions about what people do with their devices just because gaming companies make lots of money.
GG mentioned it. But never mind. Your responses to GG and Morgle in this thread (despite your own incitement) are less than adequate. The examples provided by GG and Morgle are VALID, because those are the ones which are COMMONLY used. If you want to choose a NICHE application for comparison, then you have to "admit" that ALL the ADDITIONAL power that Apple's A series processors is useful for a small set of NICHE users ONLY.
The specific App examples that you gave in one of your comments (video encoding, complex spreadsheet, audio mixing etc) would have come up in the TOP 10 most APPS used list if they are used by "majority" of the users. They are NOT used by majority of the people (I am talking about iPhone users only), hence they are not coming up in the TOP 10 or 100 most frequently used Apps.
Either you have to admit the A series processors performance has reached a "niche" category - much like 18 core Xeon processors in the iMac Pro that is needed only by a small sub segment of Mac owners Vs iMacs with i5/i7 (in mobile terms, the snapdragons and exynos would fit that category) satisfying the needs of common people. Is an 18 core Xeon processor much more powerful than quad core i7? Of course, Yes. Does everyone NEED it? Absolutely NO. Only a small set of niche users need that 18 core Xeon processor.
OR
prove that it is useful for "generic" users with "generic" App/Game examples.
I wonder if there is any consumer facing app to break down the use of the cores over time.
My guess is that for huge usage periods, the performance cores remain idle.
I agree with Gatorguy that I would never spend long periods of time doing photo editing on a phone.
Speed stopped being a deal-breaker in the mid tiers a few years ago. I haven't seen anyone complain about it. That said, things get faster. However, should anyone with an A11 rush out and get a new phone just to have more speed? No. And they don't. Upgrade cycles for iPhones are slowing. People are sticking with their iPhones and (and old processors) for years and not worrying about it.
'Slower' can still mean 'fast' and clearly does for the vast majority of users. In fact, millions of Android users upgraded to ultra premium phones this year and got far more out of the deal than with any iPhone currently available.
That's because processor cores are less important now than before. Far less important.
It impresses me no end that the new meme is that Apple's silicon advantage doesn't matter. Yet I recall you personally touting how great the Kirin 980 was going to be; well of course before the inevitable comparisons. You even attempted to convince yourself that Huawei was first with 7nm silicon, which is actually laughable, given Apple's current iPhone X deliveries. Announced first isn't the same as shipping.
I would note that Apple's advantage in silicon is so vast, that Huawei, Samsung, and Google, are all in the race to build their own silicon. All of those companies are attempting to migrate their customers and their product lines to match Apple's marketing model, and all will fall short.
A processor goes live at its presentation. The Kirin 980 pipped Apple to that post. The A12 pipped Huawei to the post for the first phone to become available on 7nm.
In both cases by very short periods.
None of this admits discussion. The facts are too clear.
In fact both chips were rolling out of TSMC at the same time.
The Kirin 980 is great. Latest variants of virtually everything on board. Including the world's first mobile Cat 21 modem and in house designed ultra fast wi-fi chip.
So, here you are, talking about speed again, after "Speed stopped being a deal breaker in the mid tiers a few years ago"
WTF.
Were you not paying attention?
The processor speed stopped being an key issue some years ago.
I even mentioned that many cores would largely go idle for long periods for lack of need.
The same cannot be said for WiFi or modem speed. You just need supporting infrastructure and you will see a noticeable difference.
AI even has a piece demonstrating the modem speed improvements on the XS over the X.
While cores may sit idle, the same cannot be said of either the modem or the wi-fi chip when activated. In both cases the user cannot control (for example by opening a processor intensive app) how many resources will be used. It is the surrounding infrastructure that will decide and the faster the better in all scenarios. The same can be said of antenna design there is still room for improvement that will translate into noticeable performance gains.
We largely haven't 'seen' those gains in the cores as most current apps simply don't max the cores out anyway.
We largely haven't 'seen' those gains in the cores as most current apps simply don't max the cores out anyway.
Posted this earlier, but worth reiterating that speed and processing power directly translates into efficiency and power management,
If processor A is twice as fast as processor S, then when loads occur processor A only spends half the time running at full speed before its able to clock back down and return to efficiency mode. Or, depending on the task, processor S may need to run flat out to handle a sustained load that processor A can handle with half the effort. Which again translates into better battery life and improved performance.
Either way, those short, quick, high intensity load cycles are much more common to most users than needing to race around running at full speed for extended periods of time.
So going back to the car analogy, it may not seem to matter when you're just "driving around town", but when you arrive at your destination driving processor A you're going to end up having more gas left in the tank, all while having had a better, smoother user experience to boot.
I’m not sure who mentioned it, but a comment was recently made that 75% of App Store revenues were games, therefore they should be used to test performance. The implication is, obviously, that most people don’t do work on their iOS devices and people mostly played games, thus trying to undermine the requirements for fast processors to run powerful Apps.
Last year Adobe had revenues of $7 billion. Autodesk had revenues of $2 billion. Two companies that make expensive, complex software used by professionals.
Activision/Blizzard had revenues of $51 billion. Should I then claim PCs are only used to play games because game revenues vastly outpaced revenues of professional applications?
Gaming is popular and generates a LOT of revenue. You can’t draw conclusions about what people do with their devices just because gaming companies make lots of money.
GG mentioned it. But never mind. Your responses to GG and Morgle in this thread (despite your own incitement) are less than adequate. The examples provided by GG and Morgle are VALID, because those are the ones which are COMMONLY used. If you want to choose a NICHE application for comparison, then you have to "admit" that ALL the ADDITIONAL power that Apple's A series processors is useful for a small set of NICHE users ONLY.
The specific App examples that you gave in one of your comments (video encoding, complex spreadsheet, audio mixing etc) would have come up in the TOP 10 most APPS used list if they are used by "majority" of the users. They are NOT used by majority of the people (I am talking about iPhone users only), hence they are not coming up in the TOP 10 or 100 most frequently used Apps.
Either you have to admit the A series processors performance has reached a "niche" category - much like 18 core Xeon processors in the iMac Pro that is needed only by a small sub segment of Mac owners Vs iMacs with i5/i7 (in mobile terms, the snapdragons and exynos would fit that category) satisfying the needs of common people. Is an 18 core Xeon processor much more powerful than quad core i7? Of course, Yes. Does everyone NEED it? Absolutely NO. Only a small set of niche users need that 18 core Xeon processor.
OR
prove that it is useful for "generic" users with "generic" App/Game examples.
I wonder if there is any consumer facing app to break down the use of the cores over time.
My guess is that for huge usage periods, the performance cores remain idle.
I agree with Gatorguy that I would never spend long periods of time doing photo editing on a phone.
Speed stopped being a deal-breaker in the mid tiers a few years ago. I haven't seen anyone complain about it. That said, things get faster. However, should anyone with an A11 rush out and get a new phone just to have more speed? No. And they don't. Upgrade cycles for iPhones are slowing. People are sticking with their iPhones and (and old processors) for years and not worrying about it.
'Slower' can still mean 'fast' and clearly does for the vast majority of users. In fact, millions of Android users upgraded to ultra premium phones this year and got far more out of the deal than with any iPhone currently available.
That's because processor cores are less important now than before. Far less important.
It impresses me no end that the new meme is that Apple's silicon advantage doesn't matter. Yet I recall you personally touting how great the Kirin 980 was going to be; well of course before the inevitable comparisons. You even attempted to convince yourself that Huawei was first with 7nm silicon, which is actually laughable, given Apple's current iPhone X deliveries. Announced first isn't the same as shipping.
I would note that Apple's advantage in silicon is so vast, that Huawei, Samsung, and Google, are all in the race to build their own silicon. All of those companies are attempting to migrate their customers and their product lines to match Apple's marketing model, and all will fall short.
A processor goes live at its presentation. The Kirin 980 pipped Apple to that post. The A12 pipped Huawei to the post for the first phone to become available on 7nm.
In both cases by very short periods.
None of this admits discussion. The facts are too clear.
In fact both chips were rolling out of TSMC at the same time.
The Kirin 980 is great. Latest variants of virtually everything on board. Including the world's first mobile Cat 21 modem and in house designed ultra fast wi-fi chip.
So, here you are, talking about speed again, after "Speed stopped being a deal breaker in the mid tiers a few years ago"
WTF.
Were you not paying attention?
The processor speed stopped being an key issue some years ago.
I even mentioned that many cores would largely go idle for long periods for lack of need.
The same cannot be said for WiFi or modem speed. You just need supporting infrastructure and you will see a noticeable difference.
AI even has a piece demonstrating the modem speed improvements on the XS over the X.
While cores may sit idle, the same cannot be said of either the modem or the wi-fi chip when activated. In both cases the user cannot control (for example by opening a processor intensive app) how many resources will be used. It is the surrounding infrastructure that will decide and the faster the better in all scenarios. The same can be said of antenna design there is still room for improvement that will translate into noticeable performance gains.
We largely haven't 'seen' those gains in the cores as most current apps simply don't max the cores out anyway.
"You need supporting infrastructure and you will see a noticeable difference"
Almost none of that infrastructure exists for any but a very few urban areas, and for most people, they aren't going to get better performance that LTE anyway.
Either you have to admit the A series processors performance has reached a "niche" category - much like 18 core Xeon processors in the iMac Pro that is needed only by a small sub segment of Mac owners Vs iMacs with i5/i7 (in mobile terms, the snapdragons and exynos would fit that category) satisfying the needs of common people. Is an 18 core Xeon processor much more powerful than quad core i7? Of course, Yes. Does everyone NEED it? Absolutely NO. Only a small set of niche users need that 18 core Xeon processor.
To be blunt, why are you guys discussing things that you really don’t know anything about? Think carefully, do you really understand your rhetorical questions vis a vis 18 core iMac Pros? Then, think again, do you really understand this analogy you are making?
The performance in the A12 benefits every single customer, every single moment they use the device. Every single app, every single UI interaction benefits from the faster A12 SoC and memory hierachy. The architecture of the SoC is basically designed entirely for consumer use cases: fastest single threaded performance possible in the given power envelope, lowest power usage possible for idling and background tasks, and fastest memory hierarchy possible from storage to memory to CPU for the given power environment. Anytime a user does some action, latencies will be lower, animations will have less frame drops, renders will be faster, anything compute bound will be faster. Image manipulations, video manipulations, PDF renders, anything that needs the CPU.
Users are able to make scripts that can take an image, resize, apply a filter, send it to another app to do whatever and than publish it. That needs CPU. I use OmniGraffle on my iPhone, a SVG style diagramming app, and that apps needs both more CPU and GPU. It’s really an idiotic argument to say that apps don’t use the power or people don’t use the power. Every single web app can stand to be faster, and virtually all web app properties these days are JavaScript monstrosities that everyone should be ashamed of.
Moreover, your analogy using an 18 core iMac Pro is flat out wrong as you don’t seem understand how computers work. Such large CPU core count devices are really only useful for embarrassingly parallel applications. Most of Apple’s content creation niche can barely use it (can Premiere or FCPX even scale to 18 cores?), and it’s really only the numerical modelers (CFD et al) that would need such a machine, or someone made a very poor choice to use an iMac Pro as a server. The A12 will actually outperform the 18-core Xeon in a variety of single thread dominant tasks (JavaScript, UI latency, most office automation tasks), and that’s the right way to do it.
Consumer needs are basically inversely proportional to the level of performance they need, specifically single threaded performance. The less proficient a user, the more important it is for them to have lower latency devices, ie, the highest per core performance possible. Lower latencies and more responsive devices make the devices easier to user, more friendly to use, more fun to use, and in the end making computing a less daunting thing. Most of the computing industry does consumers a disservice by shoveling crap bin parts to the poorest customers, when they really need the fastest, low core count (probably 2 cores) CPU possible. Intel, HiSilicon, Qualcomm, AMD, really should be offering 2-core (with frequency-voltage scaling or +2 low power cores) that offer the fastest single threaded performance possible. It would be cheap, low power, and offer the 99% of the experience for them as compared to the 6 and 8 core monstrosities they are selling to OEMs now.
Ie, consumers are better off if Intel shipped a large cache 6 GHz turbo dual-core over a 4 GHz turbo 18-core machine, or the 6/8 Core i7/i9 chips. Apple is definitely following the former model with the best single threaded performance possible.
Now, when comparing performance, everything is always gated by the slowest thing in the computing train. Gigabit LTE is great, but the vast majority of data coming off the network starts out coming off a hard drive somewhere. Slow and latent 200 to 500 MBit/s hard drives. Not to mention the multi-year buildout of gigabit LTE towers that is only just getting started. If you have 1 TB microSD card, it will limit the performance of an app start or data load because the data is coming off the card at a low 500 Mbit/s or whatever, the same with any OS operation that has to pull data from storage instead of RAM. A SSD has been and is the number performance upgrade for consumers computers for nearly a decade now. Maybe in 10 years, when everything is on SSDs and the network works at gigabit speeds everywhere, gigibit networks will be useful for most customers.
You can can make the argument that human perception means that if most latencies or 500 ms or less, your at the point of diminishing returns, which is true a large number of things. I wouldn’t agree with the argument in general as I’d prefer instantaneous for all ops. As the saying goes, we always end up using all the compute available, generalizing a bit here as multi-core CPU architectures essentially only benefit embarrassingly parallel problems, and what I really mean is single thread performance. You need to be running pretty specific for multicore architectures (which are indeed quite compute hungry).
Either you have to admit the A series processors performance has reached a "niche" category - much like 18 core Xeon processors in the iMac Pro that is needed only by a small sub segment of Mac owners Vs iMacs with i5/i7 (in mobile terms, the snapdragons and exynos would fit that category) satisfying the needs of common people. Is an 18 core Xeon processor much more powerful than quad core i7? Of course, Yes. Does everyone NEED it? Absolutely NO. Only a small set of niche users need that 18 core Xeon processor.
To be blunt, why are you guys discussing things that you really don’t know anything about? Think carefully, do you really understand your rhetorical questions vis a vis 18 core iMac Pros? Then, think again, do you really understand this analogy you are making?
The performance in the A12 benefits every single customer, every single moment they use the device. Every single app, every single UI interaction benefits from the faster A12 SoC and memory hierachy. The architecture of the SoC is basically designed entirely for consumer use cases: fastest single threaded performance possible in the given power envelope, lowest power usage possible for idling and background tasks, and fastest memory hierarchy possible from storage to memory to CPU for the given power environment. Anytime a user does some action, latencies will be lower, animations will have less frame drops, renders will be faster, anything compute bound will be faster. Image manipulations, video manipulations, PDF renders, anything that needs the CPU.
Users are able to make scripts that can take an image, resize, apply a filter, send it to another app to do whatever and than publish it. That needs CPU. I use OmniGraffle on my iPhone, a SVG style diagramming app, and that apps needs both more CPU and GPU. It’s really an idiotic argument to say that apps don’t use the power or people don’t use the power. Every single web app can stand to be faster, and virtually all web app properties these days are JavaScript monstrosities that everyone should be ashamed of.
Moreover, your analogy using an 18 core iMac Pro is flat out wrong as you don’t seem understand how computers work. Such large CPU core count devices are really only useful for embarrassingly parallel applications. Most of Apple’s content creation niche can barely use it (can Premiere or FCPX even scale to 18 cores?), and it’s really only the numerical modelers (CFD et al) that would need such a machine, or someone made a very poor choice to use an iMac Pro as a server. The A12 will actually outperform the 18-core Xeon in a variety of single thread dominant tasks (JavaScript, UI latency, most office automation tasks), and that’s the right way to do it.
Consumer needs are basically inversely proportional to the level of performance they need, specifically single threaded performance. The less proficient a user, the more important it is for them to have lower latency devices, ie, the highest per core performance possible. Lower latencies and more responsive devices make the devices easier to user, more friendly to use, more fun to use, and in the end making computing a less daunting thing. Most of the computing industry does consumers a disservice by shoveling crap bin parts to the poorest customers, when they really need the fastest, low core count (probably 2 cores) CPU possible. Intel, HiSilicon, Qualcomm, AMD, really should be offering 2-core (with frequency-voltage scaling or +2 low power cores) that offer the fastest single threaded performance possible. It would be cheap, low power, and offer the 99% of the experience for them as compared to the 6 and 8 core monstrosities they are selling to OEMs now.
Ie, consumers are better off if Intel shipped a large cache 6 GHz turbo dual-core over a 4 GHz turbo 18-core machine, or the 6/8 Core i7/i9 chips. Apple is definitely following the former model with the best single threaded performance possible.
Now, when comparing performance, everything is always gated by the slowest thing in the computing train. Gigabit LTE is great, but the vast majority of data coming off the network starts out coming off a hard drive somewhere. Slow and latent 200 to 500 MBit/s hard drives. Not to mention the multi-year buildout of gigabit LTE towers that is only just getting started. If you have 1 TB microSD card, it will limit the performance of an app start or data load because the data is coming off the card at a low 500 Mbit/s or whatever, the same with any OS operation that has to pull data from storage instead of RAM. A SSD has been and is the number performance upgrade for consumers computers for nearly a decade now. Maybe in 10 years, when everything is on SSDs and the network works at gigabit speeds everywhere, gigibit networks will be useful for most customers.
You can can make the argument that human perception means that if most latencies or 500 ms or less, your at the point of diminishing returns, which is true a large number of things. I wouldn’t agree with the argument in general as I’d prefer instantaneous for all ops. As the saying goes, we always end up using all the compute available, generalizing a bit here as multi-core CPU architectures essentially only benefit embarrassingly parallel problems, and what I really mean is single thread performance. You need to be running pretty specific for multicore architectures (which are indeed quite compute hungry).
You made many good points, which I agree with completely. while I may have used the technical terminology wrong, but the key point in your response is bolded by me. With a qualifier - It is applicable to the APPs that most of the people are using. And most of the Apps that "common/mainstream" people are using have fallen into that category off late, to the level where "raw performance" is NOT the PRIMARY necessity while looking out for a new phone for the majority of the "common/mainstream" people.
Well, I have to say, this has been the most bizarre pitch invasion I've seen it quite some time.
In order to prop up their favourite platform, the Android brigade are now insisting that increasing phone performance doesn't matter (ignoring the fact that Apple does a lot more AI processing on the phone than back end servers). First time I've ever heard a whiff of this nonsense since the off-cited, often-disproven claim that Bill Gates said 640K memory was all we'd ever need.
Well, I have to say, this has been the most bizarre pitch invasion I've seen it quite some time.
In order to prop up their favourite platform, the Android brigade are now insisting that increasing phone performance doesn't matter (ignoring the fact that Apple does a lot more AI processing on the phone than back end servers). First time I've ever heard a whiff of this nonsense since the off-cited, often-disproven claim that Bill Gates said 640K memory was all we'd ever need.
Don't give up on the removals business, chaps.
Few simple questions to you:
1. Whom does Apple target selling iPhones towards? Is it a "niche" group of users who have "specific" needs OR "generic/mainstream" users (about 1 billion of the installed base) whose needs are NOT so demanding (restricted to browsing/multimedia consumption/few social media Apps and few other non-demanding Apps)?
2. If it is the latter, what are the "performance expectations" from them of their phones? How do the older iPhones (6s/6s Plus/7/7 Plus) fulfill their performance needs with iOS 12 Vs latest iPhones (8/8 Plus/X/Xr/Xs/Xs Plus)?
3. Has the performance of 6s/6s Plus/7/7 Plus "degraded" to the level of "nearly unusable/damn slow/frustrating" for basic operations for most of the "mainstream" user's use cases that performance available in new SoCs are MUST for even basic operations that "mainstream users" end up using?
The key point that you/tmay/tht are "refusing" to see is this - Target audience for iPhones, i.e. "mainstream users" (hundreds of millions of them whose performance NEEDs are NOT as demanding as you "imagine" them to be), NOT the techies (people like you/tmay/tht etc) in this forum who know how to use their smartphones for more demanding tasks and do get maximum out of it.
We largely haven't 'seen' those gains in the cores as most current apps simply don't max the cores out anyway.
Posted this earlier, but worth reiterating that speed and processing power directly translates into efficiency and power management,
If processor A is twice as fast as processor S, then when loads occur processor A only spends half the time running at full speed before its able to clock back down and return to efficiency mode. Or, depending on the task, processor S may need to run flat out to handle a sustained load that processor A can handle with half the effort. Which again translates into better battery life and improved performance.
Either way, those short, quick, high intensity load cycles are much more common to most users than needing to race around running at full speed for extended periods of time.
So going back to the car analogy, it may not seem to matter when you're just "driving around town", but when you arrive at your destination driving processor A you're going to end up having more gas left in the tank, all while having had a better, smoother user experience to boot.
Agreed, but that is power efficiency and its management, not performance per se.
In power efficiency and management, other factors come into play such as battery capacity and charging speeds. If your phone can go for over a day before needing a charge, efficiency isn't as important as on a phone that struggles to get a regular user through the day.
From a purely performance angle, the user is not seeing a difference in most situations. Apps open quickly and work quickly
You can can make the argument that human perception means that if most latencies or 500 ms or less, your at the point of diminishing returns, which is true a large number of things. I wouldn’t agree with the argument in general as I’d prefer instantaneous for all ops.
You made many good points, which I agree with completely. while I may have used the technical terminology wrong, but the key point in your response is bolded by me. With a qualifier - It is applicable to the APPs that most of the people are using. And most of the Apps that "common/mainstream" people are using have fallen into that category off late, to the level where "raw performance" is NOT the PRIMARY necessity while looking out for a new phone for the majority of the "common/mainstream" people.
You are threading a needle too much. Apple, and many other companies, are offering a product that is comparatively best in class or class competitive among several axes of the smartphone experience, such as performance, industrial design (includes software design), service, build quality, etc. People will pay more if they get more, especially for things that they value or proves to be a value to them. I don’t think it is any more complex than that.
A company that is able to determine the features that people will value will get a leg up. A company that are unable to do that will have an unsuccessful product, or will have to sell at a lower price. Apple has been on both ends of the spectrum in determining what people want, and no company gets it always right.
Perhaps there will be a time when smartphones become so commoditized that new features don’t provide any value to customers, and therefore prices will come down and performance will stagnate. If smartphones continue to be items of personal value, continue to be a general purpose computer, there will continue to be a significant fraction of the market who will pay more to get more.
And I disagree with the good enough notion. More performance continues to improve the user experience. We’re not done yet, and the stuff that appears on the horizon like assistants (they were called smart agents 20 years ago), neural network computing, web apps (99% of the CPU is wasted on ads here and when has ads ever gotten less?), security, so on and so forth, mean there is a long way to go still.
Comments
WTF.
Partially agree!!! Agree with first 3 paragraphs of your reply. Diagree with the below:
If you are satisfied with less than optimal performance, then buy an older iPhone, or buy a flagship Android OS device like the Pixel, because it just doesn't matter to you.
For the rest of us that aspire to the smoothest user experience, those few hundred extra dollars are well spent.
Your milage evidently varies.
Still, the A12 as reviewed by Anandtech, leaves no doubt that there is substantial improvement in performance, but that should be compared iPhone X with iOS 12 vs iPhone XS with iOS 12, which has also been a commonly reviewed comparison, including just today here at AI.
I would upgrade any iPhone or iPad eligible to iOS 12 if there are no other factors beneficial to not doing it.
The processor speed stopped being an key issue some years ago.
I even mentioned that many cores would largely go idle for long periods for lack of need.
The same cannot be said for WiFi or modem speed. You just need supporting infrastructure and you will see a noticeable difference.
AI even has a piece demonstrating the modem speed improvements on the XS over the X.
While cores may sit idle, the same cannot be said of either the modem or the wi-fi chip when activated. In both cases the user cannot control (for example by opening a processor intensive app) how many resources will be used. It is the surrounding infrastructure that will decide and the faster the better in all scenarios. The same can be said of antenna design there is still room for improvement that will translate into noticeable performance gains.
We largely haven't 'seen' those gains in the cores as most current apps simply don't max the cores out anyway.
Posted this earlier, but worth reiterating that speed and processing power directly translates into efficiency and power management,
If processor A is twice as fast as processor S, then when loads occur processor A only spends half the time running at full speed before its able to clock back down and return to efficiency mode. Or, depending on the task, processor S may need to run flat out to handle a sustained load that processor A can handle with half the effort. Which again translates into better battery life and improved performance.
Either way, those short, quick, high intensity load cycles are much more common to most users than needing to race around running at full speed for extended periods of time.
So going back to the car analogy, it may not seem to matter when you're just "driving around town", but when you arrive at your destination driving processor A you're going to end up having more gas left in the tank, all while having had a better, smoother user experience to boot.
Almost none of that infrastructure exists for any but a very few urban areas, and for most people, they aren't going to get better performance that LTE anyway.
The performance in the A12 benefits every single customer, every single moment they use the device. Every single app, every single UI interaction benefits from the faster A12 SoC and memory hierachy. The architecture of the SoC is basically designed entirely for consumer use cases: fastest single threaded performance possible in the given power envelope, lowest power usage possible for idling and background tasks, and fastest memory hierarchy possible from storage to memory to CPU for the given power environment. Anytime a user does some action, latencies will be lower, animations will have less frame drops, renders will be faster, anything compute bound will be faster. Image manipulations, video manipulations, PDF renders, anything that needs the CPU.
Users are able to make scripts that can take an image, resize, apply a filter, send it to another app to do whatever and than publish it. That needs CPU. I use OmniGraffle on my iPhone, a SVG style diagramming app, and that apps needs both more CPU and GPU. It’s really an idiotic argument to say that apps don’t use the power or people don’t use the power. Every single web app can stand to be faster, and virtually all web app properties these days are JavaScript monstrosities that everyone should be ashamed of.
Moreover, your analogy using an 18 core iMac Pro is flat out wrong as you don’t seem understand how computers work. Such large CPU core count devices are really only useful for embarrassingly parallel applications. Most of Apple’s content creation niche can barely use it (can Premiere or FCPX even scale to 18 cores?), and it’s really only the numerical modelers (CFD et al) that would need such a machine, or someone made a very poor choice to use an iMac Pro as a server. The A12 will actually outperform the 18-core Xeon in a variety of single thread dominant tasks (JavaScript, UI latency, most office automation tasks), and that’s the right way to do it.
Consumer needs are basically inversely proportional to the level of performance they need, specifically single threaded performance. The less proficient a user, the more important it is for them to have lower latency devices, ie, the highest per core performance possible. Lower latencies and more responsive devices make the devices easier to user, more friendly to use, more fun to use, and in the end making computing a less daunting thing. Most of the computing industry does consumers a disservice by shoveling crap bin parts to the poorest customers, when they really need the fastest, low core count (probably 2 cores) CPU possible. Intel, HiSilicon, Qualcomm, AMD, really should be offering 2-core (with frequency-voltage scaling or +2 low power cores) that offer the fastest single threaded performance possible. It would be cheap, low power, and offer the 99% of the experience for them as compared to the 6 and 8 core monstrosities they are selling to OEMs now.
Ie, consumers are better off if Intel shipped a large cache 6 GHz turbo dual-core over a 4 GHz turbo 18-core machine, or the 6/8 Core i7/i9 chips. Apple is definitely following the former model with the best single threaded performance possible.
Now, when comparing performance, everything is always gated by the slowest thing in the computing train. Gigabit LTE is great, but the vast majority of data coming off the network starts out coming off a hard drive somewhere. Slow and latent 200 to 500 MBit/s hard drives. Not to mention the multi-year buildout of gigabit LTE towers that is only just getting started. If you have 1 TB microSD card, it will limit the performance of an app start or data load because the data is coming off the card at a low 500 Mbit/s or whatever, the same with any OS operation that has to pull data from storage instead of RAM. A SSD has been and is the number performance upgrade for consumers computers for nearly a decade now. Maybe in 10 years, when everything is on SSDs and the network works at gigabit speeds everywhere, gigibit networks will be useful for most customers.
You can can make the argument that human perception means that if most latencies or 500 ms or less, your at the point of diminishing returns, which is true a large number of things. I wouldn’t agree with the argument in general as I’d prefer instantaneous for all ops. As the saying goes, we always end up using all the compute available, generalizing a bit here as multi-core CPU architectures essentially only benefit embarrassingly parallel problems, and what I really mean is single thread performance. You need to be running pretty specific for multicore architectures (which are indeed quite compute hungry).
In order to prop up their favourite platform, the Android brigade are now insisting that increasing phone performance doesn't matter (ignoring the fact that Apple does a lot more AI processing on the phone than back end servers). First time I've ever heard a whiff of this nonsense since the off-cited, often-disproven claim that Bill Gates said 640K memory was all we'd ever need.
Don't give up on the removals business, chaps.
Few simple questions to you:
1. Whom does Apple target selling iPhones towards? Is it a "niche" group of users who have "specific" needs OR "generic/mainstream" users (about 1 billion of the installed base) whose needs are NOT so demanding (restricted to browsing/multimedia consumption/few social media Apps and few other non-demanding Apps)?
2. If it is the latter, what are the "performance expectations" from them of their phones? How do the older iPhones (6s/6s Plus/7/7 Plus) fulfill their performance needs with iOS 12 Vs latest iPhones (8/8 Plus/X/Xr/Xs/Xs Plus)?
3. Has the performance of 6s/6s Plus/7/7 Plus "degraded" to the level of "nearly unusable/damn slow/frustrating" for basic operations for most of the "mainstream" user's use cases that performance available in new SoCs are MUST for even basic operations that "mainstream users" end up using?
The key point that you/tmay/tht are "refusing" to see is this - Target audience for iPhones, i.e. "mainstream users" (hundreds of millions of them whose performance NEEDs are NOT as demanding as you "imagine" them to be), NOT the techies (people like you/tmay/tht etc) in this forum who know how to use their smartphones for more demanding tasks and do get maximum out of it.
In power efficiency and management, other factors come into play such as battery capacity and charging speeds. If your phone can go for over a day before needing a charge, efficiency isn't as important as on a phone that struggles to get a regular user through the day.
From a purely performance angle, the user is not seeing a difference in most situations. Apps open quickly and work quickly
A company that is able to determine the features that people will value will get a leg up. A company that are unable to do that will have an unsuccessful product, or will have to sell at a lower price. Apple has been on both ends of the spectrum in determining what people want, and no company gets it always right.
Perhaps there will be a time when smartphones become so commoditized that new features don’t provide any value to customers, and therefore prices will come down and performance will stagnate. If smartphones continue to be items of personal value, continue to be a general purpose computer, there will continue to be a significant fraction of the market who will pay more to get more.
And I disagree with the good enough notion. More performance continues to improve the user experience. We’re not done yet, and the stuff that appears on the horizon like assistants (they were called smart agents 20 years ago), neural network computing, web apps (99% of the CPU is wasted on ads here and when has ads ever gotten less?), security, so on and so forth, mean there is a long way to go still.