Every year , Apple introduces a unexampled A - series processor with its iPhone upgrades in the fall . We do n’t expect anything unlike this twelvemonth . In fact , the parachuting to a new manufacturing process technology — TSMC ’s 3 nm process — gives the A17 an opportunity to be the most significant leaping in performance and feature of speech in several years .
By looking at the past decennium of in - house Apple - designed A - series bit , together with what we recognize about the manufacturing engineering available and the caller ’s direction and destination , we can nibble together a pretty good educated guess about what to expect from the A17 .
Just for iPhone 15 Pro (or Ultra)
Last year , the A16 was undivided to the iPhone 14 Pro , while the standard iPhone 14 used the A15 . We expect a repeat of that normal this year , where the raw A17 is sole to theiPhone 15 Pro and iPhone 15 Pro Max(oriPhone 15 Ultra , if the hearsay are true ) and the stock iPhone 15 go the A16 used in the iPhone 14 Pro models .
Is this the way forwards for Apple for the foreseeable hereafter ? With smartphone features and tone reach something of a tableland and Apple ’s phone microprocessor chip running circles around most Android phones , there ’s little reason to change . For Apple , it reduces monetary value and help produce a bounteous differentiator between regular and Pro models , which be given to crowd customers toward the more expensive iPhone .
Apple’s first 3nm chip
The A14 , A15 , and A16 were all made using a 5 nm manufacturing process from TSMC . allot , that process has germinate over clip , create micro chip that are denser and have meliorate index efficiency , but there ’s nothing quite like the leap to the next major process node . And that ’s what we ’re almost certainly going to get with the A17 — the first big - scale consumer bit made with TSMC ’s 3 micromillimeter cognitive operation .
I write at distance about the reward Apple will savor with a 3 nm physical process , and the big one is more concentration – while the A16 was about 16 billion transistors , we can expect well over 20 billion for the A17 , perhaps as mellow as about 24 billion .
The 3 nm process offer more exponent efficiency , with a like chip at a comparable speed , but Apple is n’t going to make a corresponding chip at a comparable focal ratio . maximal power draw will be limit by battery size , thermal dissipation , and other factor , and I do n’t think we can gestate a monumental shift in electric battery life from the move to 3 Land of Enchantment alone . At least , not for combat-ready use under full mightiness – not only will the chip in all probability ware nearly as much power in that scenario , but the display and radios also contribute so much to the power drainage .
A projected Geekbench score for the A17, following the trend of the last several years.
Where we may see some melioration is in standby style , which may get perceptibly better with the move to the 3 micromillimeter process .
Foundry
CPU performance and features
ARMlaunched its v9 architecture in 2021 , and we think the A16 might be Apple ’s first cow dung to endorse the new v9 instruction set . alternatively , it fend for ARM v8.6 with peck of Apple ’s own extensions . This class , with a higher junction transistor budget , it seems ARM v9 support is likely .
What advantage do the ARM v9 instruction circle and architecture provide ? orchard apple tree designs its own CPU core , and many of the performance benefit promised by the v9 computer architecture are already realized in Apple ’s designs and ARM extensions . Indeed , the Snapdragon 8 Gen 1 was one of the first high - end smartphone CPUs with ARM ’s Cortex - X2 core with support for ARM v9 , and Apple ’s A15 outperformed it by a wide margin .
You ’ll see a lot of call that ARM v9 offers a 30 percent performance improvement over ARM v8 , but that ’s for ARM ’s own essence blueprint , and does n’t take into story the use of custom filename extension . Apple ’s in a whole other league here – we probably wo n’t see 30 percentage faster CPU performance from the A17 .
A projected Geekbench score for the A17, following the trend of the last several years.
Apple ’s new CPU center for the A17 will almost certainly be quicker , but notnecessarilybecause of the chemise to ARM v9 . central processing unit core operation is influenced by the teaching set , branch prediction , education decode , execution unit , cache structure and size , clock speeds , and many other ingredient .
As far as general core count work , there does n’t seem to be much reason for Apple to move beyond the 4 efficiency heart and soul and 2 performance cores that have been with us since the A11 Bionic . We just expect a good 15 percent more public presentation out of them .
just project the last several years of mainframe performance improvement , we can probably require a Geekbench 5 exclusive - core score between 2,100 and 2,200 , and a multi - core score of just over 6,000 . Geekbench 6 was just released and we do n’t have years of benchmark datum to make an exact projection , but a single - core grade of over 2,800 and a multi - core grade of 7,300 or more seems justifiable . Arecent leakthat claims a individual - center sexual conquest of 3019 and a multi - core musical score of 7,860 is n’t out of the kingdom of possibleness — especially since we saw a larger - than - expected leap when Apple moved from a 7 nm process to a 5 nm one with the A14 — but numbers this betimes are probably wholly made up .
A projected Geekbench score for the A17, following the trend of the last several years.
In guinea pig you ’re wondering how that stack up to other processors , that would give the A17 a single - core score similar to the previous Ryzen in high spirits - ending desktop CPUs and 13th - gen Core i7 Intel mainframe , but a far abject multi - core scotch ( which makes sentiency , as we ’re talking about only two gamey - performance pith versus 12 or more in those desktop processors ) . The A16 already soundly beat Android earpiece with the top - tier Snapdragon 8 Gen 2 , and the A17 should only widen the crack .
If we ’ve learned anything over the years , it ’s just howsteadythe performance melioration is for Apple ’s central processing unit . Single- and multi - core performance goes up at a nearly straight line , no matter which year have heavy architectural change or manufacturing process jump . It ’s very reasonable to expect a interchangeable improvement this class .
GPU performance and features
The GPU is one area where the A17 could potentially be very interesting . Apple has been increase GPU execution by an average of around 20 per centum or so with each unexampled A - series chip , though it can be anywhere from 15 per centum to 30 per centum . What has n’t vary a whole lot is the overall feature set of the GPU . It ’s getting faster , and there have been some minor new features like varying rasterization pace and SIMD improvements for GPU compute , but Apple isyearsbehind desktop GPUs in important feature like ray trace quickening .
A unelaborated rumor said that the A16 was signify to have a major GPU architecture but it was n’t quick in clip , so it got the same GPU as the A15 ( but more memory bandwidth improve performance ) . I do n’t know if that ’s dependable , but Apple has n’t update itsMetal lineament rig table for developersto admit the A16 , which is tell .
I call up it ’s potential that Apple has an update GPU architecture ready to go for A17 . lineament like ray tracing quickening may not be critical for iPhone , but this GPU pattern will find its direction into next M - series Mac mainframe , where the lack of innovative GPU feature like ray trace acceleration put them far behind state of the artwork .
A projected Geekbench score for the A17, following the trend of the last several years.
We can also require improved performance in current 3D games and applications that apply the GPU for reckoning . When the computer architecture changes , the quickening tends to be variable – some things get a lot faster , and other things not so much .
stay put the course with a some 20 percent speedup bring on a Geekbench 5 GPU compute scotch of over 18,000 . Remember , that benchmark measures thecomputationalperformance of the GPU , not its ability to interpret 3D graphics .
For 3D graphics , a 20 per centum speedup would lead in the mod 3DMark Wild Life Unlimited tryout running at around 88fps versus 74fps with the A16 . presently , the Snapdragon 8 Gen 2 is faster than the A16 in this test ( and other 3D graphic tests ) , but this would put Apple slightly ahead .
No Android device comes close to Apple in Geekbench compute scores.
proportional to CPU improvements , GPU performance is more wide-ranging from year to year . The betterment from A15 to A16 was muted , owe mostly to slight clock speed improvements and improved retentiveness bandwidth . This year , if we get both a novel GPU computer architecture and a major manufacturing procedure improvement , the spring could be much bigger .
The media railway locomotive is often loosely associated with the GPU , and so this is as dependable a time as any to say we will once again hope for hardware to speed up encode and decode of the AV1 data formatting . It ’s in most of the new generation of PC GPUs , after all . We expect Apple to continue to invest in the carrying into action and powerfulness efficiency of its encoders for H.264 , HEVC , and ProRes formats .
A continued focus on ML and AI
Apple is big into political machine learning and AI . While it does n’t appear to be pushing as hard as its competitors in thegenerative AI raceled by projects like ChatGPT , Midjourney , and Stable Diffusion , the company has used AI and machine study throughout its operating system and apps for years . New features like the ability to select textual matter in photograph are implemented all the meter , and Apple dedicates a hatful of orbit to its Neural Engine ( the accelerator for machine scholarship tasks ) .
In the A16 , Apple did n’t seem to change the Neural Engine much at all . It ’s still 16 cores , and at 17 trillion operations per second is only 8 % faster than the A15 ’s Neural Engine . That ’s a small enough leap to adopt it simply runs at a higher clock velocity . honestly , we expected much more .
With the 3 nm mental process cause a much bigger transistor budget possible , the Neural Engine will probably take a significant saltation . It could be just more core , big innovation changes to the room the cores engage , or both . I ’d be surprised if it does n’t deliver over 20 trillion operations per sec , though there is some argument over whether “ trillions of operations per second ” is really the best way to measure performance .
We expect better performance out of the A17 GPU, but it’s new features we hope for the most.
Faster LPDDR5x RAM
With the A16 , Apple boosted the RAM to LPDDR5 ( from LPDDR4x in the A15 ) . vie top - tier chips like the Snapdragon 8 Gen 2 use LPDDR5x , which delivers about 33 percent more bandwidth and lower memory latency while using the same power .
More memory bandwidth is good for everything , particularly when it does n’t utilize more power . The most obvious benefactive role will be mellow - destruction 3-D games , which stress both processor and GPU in ways that drive the limits of storage bandwidth .
Apple ’s not always the flying to jump on new memory banner , but it does prioritise memory bandwidth and self-aggrandising caches and seems to recognize the welfare ofmoving things aroundfaster instead of justprocessingthings faster . I give it 50/50 odds that we ’ll see LPDDR5x RAM in the A17 .
Still a Snapdragon modem (for now)
Apple is expect to bulge usingits own 5 G modemsstarting next yr , credibly with the iPhone SE in the spring and , if things go well , later in the iPhone 16 line in the crepuscule .
TheSnapdragon X70is probably the modem Apple will apply for the iPhone 15 line ( at least the Pro exemplar ) . Most of the X70 ’s top - line of work characteristic sound fundamentally the same as the X65 that you incur in the iPhone 14 Pro , but it has a small integrated AI processor that constantly monitors and optimizes connection states , which is presuppose to result in more static and optimal link . That it is supposed to improve substantial - globe stop number and promote electric battery spirit as well .
