Apple is in the midst of atwo - year transition off from Intel and AMD chipsto its own Apple Silicon . The low - end product are already covered by the first - generation M1 chip that ’s in theMacBook Air,13 - inch MacBook Pro , Mac miniskirt , and now the24 - inch iMac . These are all well - serve by the M1 chip , which is fundamentally an A14 on steroids . Just as the A12X was the A12 with double the high - functioning CPU kernel , GPU cores , and memory bus width , so too , is the M1 as compared to the A14 . That ’s why Apple put it in the new iPad Pro .

That ’s great deal of business leader for the lower - end Macs , but the bounteous MacBook Pros and iMacs will really need something a small faster if they ’re move to interchange the hefty Intel CPUs and AMD GPUs in current product . And the Mac Pro , well , that ’s a whole other bent of requirement , and will probably be the last ware to make the Apple Silicon transition . So , with rumors swirling that the M2 ( or maybe it ’s called the M1X ? ) entering yield , it ’s deserving set our expectations . What ’s the next Apple Silicon chip cash in one’s chips to be ?

M1X or M2?

Over atCPU - Monkeywe’ve seen “ leak ” specs and benchmark for something call the M1X. It ’s deserving submit these with a grain of salt , as unreleased Cartesian product details on that site often turn out to be amiss . Still , they map a very reasonable set of specs for the next step up in Apple Silicon .

Whether Apple calls the chip an M1X or an M2 is almost pedantic . M2 would make more sense if the chip were to be free-base on the architectural improvement of theA15 processorcoming to iPhones and iPads this dip , while M1X makes more signified if it ’s a souped - up M1 . honestly , it seems hard to think that an M - series chip establish on the A15 computer architecture would be ready to enter product already , so my hypothesis is that any fresh Apple silicon chip shot cargo ships in 2021 will be establish on the A14 and M1 .

IDG

Article image

The A15 is expected to be made on a jolly enhanced 5 - nanometre unconscious process and propose a bit more CPU and GPU execution , along with the expected yearly update to image/ video recording processing and the Neural Engine . I think Apple will finally step up to LPDDR5 memory this twelvemonth too , give it a deal more memory bandwidth . Those improvements would be foregone in an M1X , but we ’d expect them in an M2 .

In fact , Apple may call the next chip the M2 even if it ’s still based on the A14 and M1 . What ’s in a name , proper ? But the chip I expect in higher - death Macs this twelvemonth , whatever it is officially discover , will in all probability be an annexe of the A14 and M1 line .

A 12-core CPU

The A14 has four mellow - efficiency CPU cores ( slower but low - tycoon ) and two high-pitched - performance cores ( very loyal but high - power ) . The M1 kept the four efficiency cores and double the eminent - performance cores to 4 , for a total of 8 processor inwardness . It build sentience for the next phylogenesis to double them again to eight high - performance cores , for a sum of 12 C.P.U. cores .

When it come to high-pitched - ending performance and bench mark , those high - performance pith are certainly doing most of the workplace . The bigger 27 - inch iMacs and 16 - in MacBook Pros have Intel processors with six or eight cores ( even 10 as an selection in the 27 - in iMac ) . Matched heart and soul - to - core , Apple ’s chips aremuchfaster than Intel ’s , so the hike to eight high performance center would secure Apple some vast performance winnings .

A 16-core GPU

The M1 doubled the figure of GPU nitty-gritty in the A14 from four to eight , and we think the next footstep up in Apple Silicon will replicate it again . A 16 - core GPU ( give Apple 256 execution social unit ) will certainly burst forth the size of it of an already huge system - on - splintering . But of all the upgrades , this is the most necessary to replace what Apple already has on the marketplace . The larger MacBook Pros and iMacs have distinct AMD Radeon graphics processors , each with its own pond of high - upper memory .

The 16 - inch MacBook Pro offers a Radeon Pro 5500 M or 5600 M with 4 gigabyte or 8 GB of dedicated graphics memory , so we ’re looking at about around five teraflops of art processing index . The 27 - inch iMac can feature anything from a Radeon Pro 5300 with 4 GB of RAM to a 5700XT with 16 gigabyte . We ’re looking at a chain of mountains of around eight trillion floating point operations per second to 12 teraflops of processing ability , and perhaps more importantly , tidy sum of dedicated gamy - speed GDDR6 memory .

We ’ve been very impressed with the relative performance of the GPU in the M1 — we find it punches way above its 2.6 - teraflop weight family — but it ’s still going to need a grown cost increase if it wants to speed past the Radeon potato chip Apple ship in its products today . Doubling the GPU cores ought to do it .

Article image

Double the RAM

The M1 can be configured with either 8 GB or 16 GB of memory . We detect that the 8 GB edition performsmuchbetter than a comparable Intel - based Mac , but notwithstanding , gamey - end Macs are survive to be require to do demanding jobs that use passel of retentiveness .

The 16 - inch MacBook Pro today starts at 16 GB of RAM and is configurable up to 64 GB . The 27 - inch iMac start at 8 GB but can go all the way up to 128 Gbit . I doubt Apple will endeavor to equate the top RAM specs of these machines , but both 16 GB and 32 GB versions of the next Apple Silicon chip are basically a shoo - in .

What ’s more intriguing to me is how Apple might also increase memory bandwidth . The M1 double up the 64 - piece memory bus of the A14 to 128 - bit . That means it ’s able-bodied to drive home about 68GB / sec using LPDDR4X. That ’s fantastic for a top-notch thin - and - lighter laptop , and even outclass the 27 - inch iMac , but those Radeon graphic chips have dedicate graphics memory with bandwidth well over 100GB / sec .

Article image

Apple ’s graphics processor are very bandwidth efficient , but I would recall a 16 - core GPU would quickly discover itself bottlenecked by storage bandwidth , especially if it had to apportion computer memory with all those CPU cores . Increasing to a 256 - bit memory interface is certainly potential and would really let the check vanish .

What about the other stuff?

If the next Mac processor is indeed based on the A14 and M1 , we gestate all the “ other stuff ” in the micro chip — the audio frequency and video recording processing , encoders and decoders , and Neural Engine — to remain unchanged from the A14 and M1 . If by some feat of engineering , the next Apple silicon chip shot actually assume its heritage from the designing of the forthcoming A15 , we expect all of those feature to improve . The Neural Engine , in finical , would in all likelihood get a substantial upgrade as Apple is all - in on AI and Machine Learning .

Even without any improvement to all of this “ un - core ” stuff , the increase in CPU and GPU core count would make the next Apple silicon chip an sheer screamer , especially if Apple doubles retentiveness bandwidth as well . It would also make the chip absolutely huge : the M1 is already a whopping 16 billion transistors , and you’re able to expect a edition with double the central processing unit and GPU core to be at least 24 billion or so . That ’s not unheard of ( AMD and Nvidia ’s biggest desktop GPUs are that bombastic ) , but it would make the poker chip far and away the biggest thoroughgoing system - on - chip on the market , even crowing than the chip in the Xbox Series X or PlayStation 5 .