Typically , Apple stays muted about its future processors , preferring to announce them along with a unexampled product . For example , we did n’t get specifics about theM1until Apple unveil the new MacBooks power by it and we knew very little about theA14before it show up up in the iPad Air . So , we in all probability wo n’t hear about the A15 , the first of a fresh contemporaries of Apple system - on - check ( SoCs ) , until Apple unveils theiPhone 13later this year .
However , we can still pick together a picture of our possible futurity . By canvass the vogue for past A - serial publication processors and what we know about make out - bound atomic number 14 manufacturing , along with a few educated guesses , we can get a pretty good idea of what to ask from the A15 .
Still 5nm, but enhanced
With the A14 , Apple made the jump from a 7 nm fabrication process to TSMC ’s brand - new 5 nm process . In fact , it was the first large - scale of measurement consumer poker chip of any variety to be made with it ( Qualcomm ’s 5 nanometer Snapdragon 888launched a dyad months later ) .
The next gradation along the route to ever - smaller and more power - efficient chip manufacture engineering is 3 nm . That ’s go to be a big bargain , with well-nigh double the chip system of logic density along with either a 25 - 30 % reduction in power at the same performance or a 10 - 15 % improvement in performance at the same superpower .
But 3 New Mexico simply is n’t quick yet . Apple pay handsomely to be the first to use TSMC ’s most advanced new manufacturing technique , but TSMC simply wo n’t be able to make tens of zillion of 3 nm chips for Apple until next class ’s A16 .
In the interim , Apple is expected to use a graceful version of the 5 nm manufacture cognitive operation for the A15 that may more or less meliorate public presentation and power characteristics .
Bigger size, more transistors
The A14 is estimate to be just under 90mm2 ( about 10 % smaller than the A13 ) , but the largest iPhone SoCs so far have been a little over 100mm2 . But with no major advancement in fabrication outgrowth technology to swear on with the A15 , the carrying into action and power improvement will necessitate to come mostly from architectural improvements and from simply making abigger splintering , much like the A13 versus the A12.So let ’s assume that the A15 will be around 15 - 20 % grownup at most , with a transistor budget to play off . No iPhone SoC has been larger than that .
That put the hypothetical transistor budget of the A15 in the range of around 14 billion . There ’s a lot of wiggle room in that appraisal , and Apple has to make a lot of hard decisions about crisp size vs. price and performance . But as a very broad rule of thumb , I ’m expecting that the party has around 15 - 20 % more transistor to crop with and base my execution and feature foretelling on that time value .
CPU performance
Improvements in individual - core CPU performance for Apple ’s A - series SoCs have beenremarkablyconsistent over the last several year . I expect this tendency is potential to persist in , which would put the A15 ’s exclusive - core Geekbench 5 result at around 1,800 , perhaps a little more .
IDG
That ’s ludicrous . Apple is justlappingAndroid sound at this full stop . A musical score of 1,800 is roughly 75 % faster than the Snapdragon 888 in the Galaxy S21 . For that matter , the A12 from later 2018 is a little bit quicker than theSamsung Galaxy S21(which get 1076 in our testing ) . Apple is in a league of its own .
There ’s not much reasonableness for Apple to expand the number of cores beyond the current setup : two high - efficiency cores and four mellow - carrying out pith . Multi - core operation improves as individual - marrow performance does , but often to a lesser degree ( thanks to cache contention , resource allotment , thermal constraint , and so on ) . A Geekbench 5 multi - core score of 4,800 would n’t be out of the question , but my guess is that we ’ll get closer to 4,600 .
Again , this would just be embarrassing the Android phones , which have only just now been able to turn over the 3,000 mark .
Graphics performance
Apple ’s got its own custom space - magnetic core GPU in the A14 , and it ’s justbarelyfaster than the quad - nucleus GPU in the A13 . But the A13 was alotfaster than the quad - magnetic core GPU in the A12 . In other words , do n’t get too hang up on how many “ pith ” the GPU in the A15 has . Graphics performance is reliant on so many considerateness , not the least of which is useable retentiveness bandwidth .
The Sling Shot trial is pretty ancient at this point , however . The new3DMark Wild Lifetest is far more modern , and a good representation of how modern GPUs wield the latest eminent - final stage three-D game . Frame rates in that mental test could get into the gamy 50s , perhaps even reaching that charming 60fps turn .
We should n’t just wait at 3D nontextual matter carrying out , though . Increasingly , Apple seems concerned with improving thecomputecapabilities of it GPUs . That mean using the GPU to process massively - parallel math bit often used in things like effigy and video handling , AI and machine learning , and scientific body of work . Geekbench 5 has a test for this GPU compute capability using the Metal API , and Apple ’s A - series chips get a lot faster with every release . The score for the A15 could be as high as 12,000 , on par with the iPad Pros that use the A12Z. That ’s comparable to GeForce GTX 980 M performance .
However , if current 3D nontextual matter carrying out is bottlenecked in the first place by memory bandwidth , and Apple bring off to greatly ameliorate it ( through crowing / better stash or a move to LPDDR5 ) , we could get a much adult GPU operation boost .
LPDDR5 RAM, finally?
I excogitate last year that Apple would make the switch from LPDDR4X , which was sport in the last three generation of A - serial french-fried potatoes , to LPDDR5 , the RAM used in the Galaxy S21 and other high - end Android phones . Oddly enough , that did n’t encounter — the A14 , and even the M1 that is base on it , apply LPDDR4X.
Samsung
This hasgotto be the year Apple convey the bound . Doing so will provide up to 50 per centum more retention bandwidth at tantamount power . This will make it easygoing to keep all the CPU and GPU cores fed with information , which means better free burning carrying into action . It might also lower power consumption , extending electric battery living .
When it come in to benchmarks , tests will often fit into the central processing unit ’s cache and do n’t scale straight with memory bandwidth . Real - world applications tend to do good more from extra memory bandwidth than synthetical benchmarks do . Modern high - death three-D graphics typically scale well with additional bandwidth , however . The switch to LPDDR5 will amend any memory - bind situation , but will have the most obvious impact on high-pitched - goal graphics diligence .
Image Processing and Neural Engine
We do n’t often get a lot of elaborate data about Apple ’s “ uncore ” part of the SoC. Apple Si has a wide array of encoders and decoder , image processing , audio DSPs , and other stuff and nonsense to accelerate specific tasks . There are important , but hard to isolate to measure out performance . Will this be the first Apple silicon to include a hardware decipherer for the AV1 video codec ? I hope so !
However , wedoget some top - line data about one such special - intention accelerator : the Neural Engine , Apple ’s automobile learning processing regalia . The A14 featured 16 Neural Engine magnetic core , double the number of effect line up in the A13 . Along with other enhancements , this boosted performance to 11 TOPS ( one million million million of operation per secondly ) .
Apple position a huge stress on machine learning , and it ’s certain to want to boost ML performance much further . Again , I ’d expect some combination of architectural advance and additional sum , though the latter may be constrained somewhat by the fix potato chip size and electronic transistor budget . I ’d estimate a final performance figure between 15 - 20 TOPS , but that ’s not much more than a guess .
An upgrade to the modem
The current iPhone 12 lineup use Qualcomm ’s Snapdragon X55 modem . While the company has declare theX65 , it wo n’t be ready in time for this class ’s fresh iPhone . Qualcomm announce these modems months before they begin volume production , and then it takes many calendar month more before telephone makers desegregate , test , and characterize them .
In short , the A15 will likely be pair with theSnapdragon X60 modem . The X60 supports the same crown speed as the X55 ( well in excess of what any carrier can deliver right on now ) , but it should return high speeds more systematically thanks to better attack aircraft carrier assembling engineering .
Qualcomm
The biggest improvement is likely to be battery life ; the X60 is made with a 5 nm cognitive process , while the X55 used a 7 nm appendage . That means a small-scale , lower - power chip , and probably less of a battery life penalty for using 5G.
As far as other wireless feature go , we can expect continued support for Wi - Fi 6 , Bluetooth 5 , NFC , and extremist - wideband ( UWB ) . Wemighteven see a parachuting to the newerWi - Fi 6E. Not many router support the 6GHz frequency spectrum yet , but it will be more common by this fall when the iPhone 13 is introduced , and everywhere over the lifetime of the product .
