Hubbry Logo
Apple siliconApple siliconMain
Open search
Apple silicon
Community hub
Apple silicon
logo
8 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Apple silicon
Apple silicon
from Wikipedia

The A16 Bionic chip

Apple Silicon is a series of system on a chip (SoC) and system in a package (SiP) processors designed by Apple Inc., mainly using the ARM architecture. They are used in nearly all of the company's devices including Mac, iPhone, iPad, Apple TV, Apple Watch, AirPods, AirTag, HomePod, and Apple Vision Pro.

The first Apple-designed system-on-a-chip was the Apple A4, which was introduced in 2010 with the first-generation iPad and later used in the iPhone 4, fourth generation iPod Touch and second generation Apple TV.

Apple announced its plan to switch Mac computers from Intel processors to its own chips at WWDC 2020 on June 22, 2020, and began referring to its chips as Apple silicon.[1][2] The first Macs with Apple silicon, built with the Apple M1 chip, were unveiled on November 10, 2020. The Mac lineup completed its transition to Apple chips in June 2023.

Apple fully controls the integration of Apple silicon in the company's hardware and software products. Johny Srouji, the senior vice president for Apple's hardware technologies, is in charge of the silicon design.[3] Apple is a fabless manufacturer; production of the chips is outsourced to contract foundries including TSMC and Samsung.

A-series SoCs

[edit]

The A series is a family of SoCs used in the iPhone, certain iPad models (including iPad Mini and entry-level iPad), and the Apple TV. A-series chips were also used in the discontinued iPod Touch line and the original HomePod. They integrate one or more ARM-based processing cores (CPU), a graphics processing unit (GPU), cache memory and other electronics necessary to provide mobile computing functions within a single physical package.[4]

Apple A4

[edit]

The Apple A4 is a PoP SoC manufactured by Samsung, the first SoC Apple designed in-house.[5] It combines an ARM Cortex-A8 CPU – also used in Samsung's S5PC110A01 SoC[6][7] – and a PowerVR SGX 535 graphics processor (GPU),[8][9][10] all built on Samsung's 45-nanometer silicon chip fabrication process.[11][12] The design emphasizes power efficiency.[13] The A4 commercially debuted in 2010, in Apple's iPad tablet,[8] and was later used in the iPhone 4 smartphone,[14] the fourth-generation iPod Touch, and the 2nd-generation Apple TV.[15]

The Cortex-A8 core used in the A4, dubbed Hummingbird, is thought to use performance improvements developed by Samsung in collaboration with chip designer Intrinsity, which was subsequently acquired by Apple[16][17] It can run at far higher clock rates than other Cortex-A8 designs yet remains fully compatible with the design provided by ARM.[18] The A4 runs at different speeds in different products: 1 GHz in the first iPads and 2nd-generation Apple TV,[19][20] and 800 MHz in the iPhone 4 and fourth-generation iPod Touch.

The A4's SGX535 GPU could theoretically push 35 million polygons per second and 500 million pixels per second, although real-world performance may be considerably less.[21] Other performance improvements include additional L2 cache.

The A4 processor package does not contain RAM, but supports PoP installation. The 1st-generation iPad, fourth-generation iPod Touch,[22] and the 2nd-generation Apple TV[23] have an A4 mounted with two low-power 128 MB DDR SDRAM chips (totaling 256 MB), while the iPhone 4 has two 256 MB packages for a total of 512 MB.[24][25][26] The RAM is connected to the processor using ARM's 64-bit-wide AMBA 3 AXI bus. To give the iPad high graphics bandwidth, the width of the RAM data bus is double that used in previous ARM11- and ARM9-based Apple devices.[27]

Apple A5

[edit]

The Apple A5 is an SoC manufactured by Samsung[28] that replaced the A4. The chip commercially debuted with the release of Apple's iPad 2 tablet in March 2011,[29] followed by its release in the iPhone 4S smartphone later that year. Compared to the A4, the A5 CPU "can do twice the work" and the GPU has "up to nine times the graphics performance",[30] according to Apple.

The A5 contains a dual-core ARM Cortex-A9 CPU[31] with ARM's advanced SIMD extension, marketed as NEON, and a dual core PowerVR SGX543MP2 GPU. This GPU can push between 70 and 80 million polygons/second and has a pixel fill rate of 2 billion pixels/second. The iPad 2's technical specifications page says the A5 is clocked at 1 GHz,[32] though it can adjust its frequency to save battery life.[31][33] The clock speed of the unit used in the iPhone 4S is 800 MHz. Like the A4, the A5 process size is 45 nm.[34]

An updated 32 nm version of the A5 processor was used in the third-generation Apple TV, the fifth-generation iPod Touch, the iPad Mini, and the new version of iPad 2 (version iPad2,4).[35] The chip in the Apple TV has one core locked.[36][37] Markings on the square package indicate that it is named APL2498, and in software, the chip is called S5L8942. The 32 nm variant of the A5 provides around 15% better battery life during web browsing, 30% better when playing 3D games and about 20% better battery life during video playback.[38]

In March 2013, Apple released an updated version of the 3rd-generation Apple TV (Rev A, model A1469) containing a smaller, single-core version of the A5 processor. Unlike the other A5 variants, this version of the A5 is not a PoP, having no stacked RAM. The chip is very small, just 6.1×6.2 mm, but as the decrease in size is not due to a decrease in feature size (it is still on a 32 nm fabrication process), this indicates that this A5 revision is of a new design.[39] Markings tell that it is named APL7498, and in software, the chip is called S5L8947.[40][41]

Apple A5X

[edit]

The Apple A5X is an SoC announced on March 7, 2012, at the launch of the third-generation iPad. It is a high-performance variant of the Apple A5; Apple claims it has twice the graphics performance of the A5.[42] It was superseded in the fourth-generation iPad by the Apple A6X processor.

The A5X has a quad-core graphics unit (PowerVR SGX543MP4) instead of the previous dual-core as well as a quad-channel memory controller that provides a memory bandwidth of 12.8 GB/s, roughly three times more than in the A5. The added graphics cores and extra memory channels add up to a very large die size of 165 mm2,[43] for example twice the size of Nvidia Tegra 3.[44] This is mainly due to the large PowerVR SGX543MP4 GPU. The clock frequency of the dual ARM Cortex-A9 cores have been shown to operate at the same 1 GHz frequency as in A5.[45] The RAM in A5X is separate from the main CPU package.[46]

Apple A6

[edit]

The Apple A6 is a PoP SoC introduced on September 12, 2012, at the launch of the iPhone 5, then a year later was inherited by its minor successor the iPhone 5C. Apple states that it is up to twice as fast and has up to twice the graphics power compared to its predecessor the Apple A5.[47] It is 22% smaller and draws less power than the 45 nm A5.[48]

The A6 is said to use a 1.3 GHz[49] custom[50] Apple-designed ARMv7 based dual-core CPU, called Swift,[51] rather than a licensed CPU from ARM like in previous designs, and an integrated 266 MHz triple-core PowerVR SGX 543MP3[52] graphics processing unit (GPU). The Swift core in the A6 uses a new tweaked instruction set, ARMv7s, featuring some elements of the ARM Cortex-A15 such as support for the Advanced SIMD v2, and VFPv4.[50] The A6 is manufactured by Samsung on a high-κ metal gate (HKMG) 32 nm process.[53]

Apple A6X

[edit]

Apple A6X is an SoC introduced at the launch of the fourth-generation iPad on October 23, 2012. It is a high-performance variant of the Apple A6. Apple claims the A6X has twice the CPU performance and up to twice the graphics performance of its predecessor, the Apple A5X.[54]

Like the A6, this SoC continues to use the dual-core Swift CPU, but it has a new quad core GPU, quad channel memory and slightly higher 1.4 GHz CPU clock rate.[55] It uses an integrated quad-core PowerVR SGX 554MP4 graphics processing unit (GPU) running at 300 MHz and a quad-channel memory subsystem.[55][56] Compared to the A6 the A6X is 30% larger, but it continues to be manufactured by Samsung on a high-κ metal gate (HKMG) 32 nm process.[56]

Apple A7

[edit]

The Apple A7 is a 64-bit PoP SoC whose first appearance was in the iPhone 5S, which was introduced on September 10, 2013. The chip would also be used in the iPad Air, iPad Mini 2 and iPad Mini 3. Apple states that it is up to twice as fast and has up to twice the graphics power compared to its predecessor the Apple A6.[57] The Apple A7 chip is the first 64-bit chip to be used in a smartphone and later a tablet computer.[58]

The A7 features an Apple-designed 1.3[59]–1.4[60] GHz 64-bit[61] ARMv8-A[62][63] dual-core CPU,[59] called Cyclone,[62] and an integrated PowerVR G6430 GPU in a four cluster configuration.[64] The ARMv8-A architecture doubles the number of registers of the A7 compared to the A6.[65] It now has 31 general-purpose registers that are each 64-bits wide and 32 floating-point/NEON registers that are each 128-bits wide.[61] The A7 is manufactured by Samsung on a high-κ metal gate (HKMG) 28 nm process[66] and the chip includes over 1 billion transistors on a die 102 mm2 in size.[59]

Apple A8

[edit]

The Apple A8 is a 64-bit PoP SoC manufactured by TSMC. Its first appearance was in the iPhone 6 and iPhone 6 Plus, which were introduced on September 9, 2014.[67] A year later it would drive the iPad Mini 4. Apple states that it has 25% more CPU performance and 50% more graphics performance while drawing only 50% of the power compared to its predecessor, the Apple A7.[68] On February 9, 2018, Apple released the HomePod, which is powered by an Apple A8 with 1 GB of RAM.[69]

The A8 features an Apple-designed 1.4[70] GHz 64-bit[71] ARMv8-A[71] dual-core CPU, and an integrated custom PowerVR GX6450 GPU in a four cluster configuration.[70] The GPU features custom shader cores and compiler.[72] The A8 is manufactured on a 20 nm process[73] by TSMC,[74] which replaced Samsung as the manufacturer of Apple's mobile device processors. It contains 2 billion transistors. Despite that being double the number of transistors compared to the A7, its physical size has been reduced by 13% to 89 mm2 (consistent with a shrink only, not known to be a new microarchitecture).[75]

Apple A8X

[edit]

The Apple A8X is a 64-bit SoC introduced at the launch of the iPad Air 2 on October 16, 2014.[76] It is a high performance variant of the Apple A8. Apple states that it has 40% more CPU performance and 2.5 times the graphics performance of its predecessor, the Apple A7.[76][77]

Unlike the A8, this SoC uses a triple-core CPU, a new octa-core GPU, dual channel memory and slightly higher 1.5 GHz CPU clock rate.[78] It uses an integrated custom octa-core PowerVR GXA6850 graphics processing unit (GPU) running at 450 MHz and a dual-channel memory subsystem.[78] It is manufactured by TSMC on their 20 nm fabrication process, and consists of 3 billion transistors.

Apple A9

[edit]

The Apple A9 is a 64-bit ARM-based SoC that first appeared in the iPhone 6S and 6S Plus, which were introduced on September 9, 2015.[79] Apple states that it has 70% more CPU performance and 90% more graphics performance compared to its predecessor, the Apple A8.[79] It is dual sourced, a first for an Apple SoC; it is manufactured by Samsung on their 14 nm FinFET LPE process and by TSMC on their 16 nm FinFET process. It was subsequently included in the first-generation iPhone SE, and the iPad (5th generation). The Apple A9 was the last CPU that Apple manufactured through a contract with Samsung, as all A-series chips after are manufactured by TSMC. The Apple A9 runs on a higher 1.85 GHz clock speed.[80]

Apple A9X

[edit]

The Apple A9X is a 64-bit SoC that was announced on September 9, 2015, and released on November 11, 2015, and first appeared in the iPad Pro.[81] It offers 80% more CPU performance and two times the GPU performance of its predecessor, the Apple A8X. It is manufactured by TSMC using a 16 nm FinFET process.[82] Unlike its predecessor, the A8X, the Apple A9X runs on an increased 2.16 to 2.26 GHz clock rate.[83]

Apple A10 Fusion

[edit]

The Apple A10 Fusion is a 64-bit ARM-based SoC that first appeared in the iPhone 7 and 7 Plus, which were introduced on September 7, 2016.[84] The A10 is also featured in the sixth-generation iPad, seventh-generation iPad and seventh-generation iPod Touch.[85] It has a new ARM big.LITTLE quad core design with two high performance cores, and two smaller highly efficient cores. It is 40% faster than the A9, with 50% faster graphics, and runs at an improved 2.34 GHz clock rate.[86] It is manufactured by TSMC on their 16 nm FinFET process.

Apple A10X Fusion

[edit]

The Apple A10X Fusion is a 64-bit ARM-based SoC that first appeared in the 10.5″ iPad Pro and the second generation of the 12.9″ iPad Pro, which were both announced on June 5, 2017.[87] It is a variant of the A10 and Apple claims that it has 30 percent faster CPU performance and 40 percent faster GPU performance than its predecessor, the A9X.[87] On September 12, 2017, Apple announced that the Apple TV 4K would be powered by an A10X chip. It is made by TSMC on their 10 nm FinFET process, and its base clock speed is 2.36 GHz, although some sources say that it is 2.38 GHz.[88][89]

Apple A11 Bionic

[edit]

The Apple A11 Bionic is a 64-bit ARM-based SoC[90] that first appeared in the iPhone 8, iPhone 8 Plus, and iPhone X, which were introduced on September 12, 2017.[90] It has two high-performance cores, which are 25% faster than the A10 Fusion, four high-efficiency cores, which are 70% faster than the energy-efficient cores in the A10, and for the first time an Apple-designed three-core GPU with 30% faster graphics performance than the A10.[90][91] It is also the first A-series chip to feature Apple's "Neural Engine," which enhances artificial intelligence and machine learning processes.[92]

Apple A12 Bionic

[edit]

The Apple A12 Bionic is a 64-bit ARM-based SoC that first appeared in the iPhone XS, XS Max and XR, which were introduced on September 12, 2018. It is also used in the third-generation iPad Air, fifth-generation iPad Mini, the eighth-generation iPad, and the second generation Apple TV 4K. It has two high-performance cores, which are 15% faster than the A11 Bionic, and four high-efficiency cores, which have 50% lower power usage than the energy-efficient cores in the A11 Bionic.[93] The A12 is manufactured by TSMC[94] using a 7 nm[95] FinFET process, the first to ship in a smartphone.[96][94]

Apple A12X Bionic

[edit]

The Apple A12X Bionic is a 64-bit ARM-based SoC that first appeared in the 11.0″ iPad Pro and the third generation of the 12.9″ iPad Pro, which were both announced on October 30, 2018.[97] It offers 35% faster single-core and 90% faster multi-core CPU performance than its predecessor, the A10X. It has four high-performance cores and four high-efficiency cores. The A12X is manufactured by TSMC using a 7 nm FinFET process.

Apple A12Z Bionic

[edit]

The Apple A12Z Bionic is an updated version of the A12X Bionic, first appearing in the fourth generation iPad Pro, which was announced on March 18, 2020.[98] It adds an additional GPU core, compared to the A12X, for improved graphics performance.[99] The A12Z is also used in the Developer Transition Kit prototype computer that helps developers prepare their software for Macs based on Apple silicon.[100]

Apple A13 Bionic

[edit]

The Apple A13 Bionic is a 64-bit ARM-based SoC that first appeared in the iPhone 11, 11 Pro, and 11 Pro Max, which were introduced on September 10, 2019. It is also featured in the second-generation iPhone SE (released April 15, 2020), the 9th generation iPad (announced September 14, 2021) and in the Studio Display (announced March 8, 2022)

The entire A13 SoC features a six-core CPU, four-core GPU, and an eight-core Neural Engine, which is dedicated to handling on-board machine learning processes; four of the six cores on the CPU are low-powered cores that are dedicated to handling less CPU-intensive operations, such as voice calls, browsing the Web, and sending messages, while two higher-performance cores are used only for more CPU-intensive processes, such as recording 4K video or playing a video game.[101]

Apple A14 Bionic

[edit]

The Apple A14 Bionic is a 64-bit ARM-based SoC that first appeared in the fourth-generation iPad Air and iPhone 12, released on October 23, 2020. It is the first commercially available 5nm chipset and contains 11.8 billion transistors and a 16-core Neural Engine.[102] It includes the Samsung LPDDR4X DRAM, a 6-core CPU, and a 4-core GPU with real-time machine learning capabilities. It was later used in the tenth-generation iPad, released on October 26, 2022.

Apple A15 Bionic

[edit]

The Apple A15 Bionic is a 64-bit ARM-based SoC that first appeared in the iPhone 13, unveiled on September 14, 2021. The A15 is built on a 5-nanometer manufacturing process with 15 billion transistors. It has 2 high-performance processing cores, 4 high-efficiency cores, a new 5-core graphics for iPhone 13 Pro series (4-core for iPhone 13 and 13 mini) processing unit, and a new 16-core Neural Engine capable of 15.8 trillion operations per second.[103][104] It is also used in the third-generation iPhone SE, iPhone 14, iPhone 14 Plus, sixth-generation iPad Mini and third generation Apple TV 4K.

Apple A16 Bionic

[edit]

The Apple A16 Bionic is a 64-bit ARM-based SoC that first appeared in the iPhone 14 Pro, unveiled on September 7, 2022. The A16 has 16 billion transistors and is built on TSMC's N4P fabrication process, being touted by Apple as the first 4 nm processor in a smartphone.[105][106] However, N4 is an enhanced version of N5 technology, a de facto fourth-generation 5 nm manufacturing process.[107][108][109] The chip has 2 high-performance processing cores, 4 high-efficiency cores and 5-core graphics for iPhone 14 Pro series. Memory is upgraded to LPDDR5 for 50% higher bandwidth and a 7% faster 16-core Neural Engine capable of 17 trillion operations per second. It was later used in the iPhone 15 and iPhone 15 Plus,[110] as well as the iPad (A16).

Apple A17 Pro

[edit]

The Apple A17 Pro is a 64-bit ARM-based SoC that first appeared in the iPhone 15 Pro, unveiled on September 12, 2023. It is Apple's first 3 nm SoC. The chip has 2 high-performance processing cores, 4 high-efficiency cores, a 6-core GPU for iPhone 15 Pro series, and a 16-core Neural Engine capable of 35 trillion operations per second. The GPU was described as their biggest redesign in the history of Apple GPUs, adding hardware accelerated ray tracing and mesh shading support.[111] It is also used in the iPad Mini (A17 Pro).

Apple A18 and Apple A18 Pro

[edit]

The Apple A18 and Apple A18 Pro are 64-bit ARM-based SoCs designed by Apple that first appeared in the iPhone 16 and iPhone 16 Pro respectively, unveiled on September 9, 2024. Both SoCs are built on TSMC's N3E process and have 2 high-performance cores and 4 high-efficiency cores. The A18 has 5-core graphics (4-core for iPhone 16e), while the A18 Pro has 6-core graphics. The A18 and A18 Pro use LPDDR5X for 17% higher memory bandwidth, and the 16-core Neural Engine has the same quoted power as the A17 Pro.

Apple A19 and Apple A19 Pro

[edit]

The Apple A19 and Apple A19 Pro are 64-bit ARM-based SoCs designed by Apple that first appeared in the iPhone 17, iPhone Air, and iPhone 17 Pro respectively, unveiled on September 9, 2025. Both SoCs are built on TSMC's N3P process and have 2 high-performance cores and 4 high-efficiency cores. The A19 has 5-core graphics, while the A19 Pro has 6-core graphics (5-core for iPhone Air).

Comparison of A-series processors

[edit]
General Semiconductor technology Computer architecture CPU GPU AI accelerator Memory technology First release
Name Codename Part number Image Node Manufacturer Transistors count Die size CPU ISA Bit width Performance core Efficiency core Overall cores Cache Vendor Cores SIMD EU count FP32 ALU count Frequency FP32 FLOPS Cores OPS Memory bus width Total channel
Bit per channel
Memory type Theoretical
bandwidth
Available capacity
Core name Cores Core speed Core name Cores Core speed L1 L2 L3 SLC
APL0098 S5L8900 90 nm
[112]
Samsung 72 mm2
[11]
ARMv6 32-bit ARM11 1 412 MHz Single-core L1i: 16 KB
L1d: 16 KB
PowerVR
MBX Lite
1 1 8 60 MHz – 103 MHz 0.96 GFLOPS – 1.64 GFLOPS 16-bit 1 channel
16-bit/channel
LPDDR-266
(133.25 
MHz)
533 MB/s 128 MB June 29, 2007
APL0278 S5L8720 65 nm
[11]
36 mm2
[11]
533 MHz 103 MHz – 133 MHz 1.64 GFLOPs – 2.12 GFLOPS 32-bit 1 channel
32-bit/channel
1066 MB/s July 11, 2008
APL0298 S5L8920 71.8 mm2
[12]
ARMv7 Cortex-A8 600 MHz L1i: 32 KB
L1d: 32 KB
256 KB PowerVR
SGX535
[113]
2 16 200 MHz 6.4 GFLOPS LPDDR-400
(200 MHz)
1.6 GB/s 256 MB June 19, 2009
APL2298 S5L8922 45 nm
[11][12]
[34]
41.6 mm2
[11]
September 9, 2009
A4 APL0398 S5L8930 53.3 mm2
[11][12]
800 MHz 512 KB 200 MHz – 250 MHz 6.4 GFLOPS – 8.0 GFLOPS 64-bit 2 channels
32-bit/channel
3.2 GB/s April 3, 2010
1.0 GHz
800 MHz 512 MB
A5 APL0498 S5L8940 122.2 mm2
[34]
Cortex-A9 2 800 MHz Dual-core 1 MB PowerVR
SGX543
[114][52]
2 4 32 200 MHz 12.8 GFLOPS LPDDR2-800
(400 MHz)
6.4 GB/s March 11, 2011
1.0 GHz
APL2498 S5L8942 32 nm
MG
[35][41]
69.6 mm2
[35]
800 MHz March 7, 2012
1.0 GHz
2[a] Dual-core[b]
APL7498 S5L8947 37.8 mm2
[41]
1 Single-core January 28, 2013
A5X APL5498 S5L8945 45 nm
[11][12]
[34]
165 mm2
[43]
2 Dual-core 4 8 64 25.6 GFLOPS 128-bit 4 channels
32-bit/channel
12.8 GB/s 1 GB March 16, 2012
A6 APL0598 S5L8950 32 nm
MG
[53][115]
[56]
96.71 mm2
[53][115]
ARMv7s[116] Swift[50] 1.3 GHz
[117]
3 6 48 266 or 709 MHz 25.5 or 68.0 GFLOPS 64-bit 2 channels
32-bit/channel
LPDDR2-1066
(533 MHz)
8.5 GB/s September 21, 2012
A6X APL5598 S5L8955 123 mm2
[56]
1.4 GHz
[55]
PowerVR
SGX554
[55][118]
4 16 128 300 MHz 76.8 GFLOPS 128-bit 4 channels
32-bit/channel
17.0 GB/s November 2, 2012
A7 APL0698 S5L8960 28 nm
MG
[66][119]
1 billion 102 mm2
[61][119]
ARMv8.0-A
[62][70]
64-bit Cyclone 1.3 GHz L1i: 64 KB
L1d: 64 KB
4 MB (Inclusive)
[62][120][60]
PowerVR
G6430
[64][118]
450 MHz 115.2 GFLOPS 64-bit 1 channel
64-bit/channel
LPDDR3-1600
(800 MHz)
12.8 GB/s September 20, 2013
APL5698 S5L8965 1.4 GHz November 1, 2013
A8 APL1011 T7000 20 nm
MG
[71][70]
TSMC 2 billion 89 mm2
[121][78]
[122]
Typhoon 1.1 GHz PowerVR
GX6450
[72][123][124]
533 MHz 136.4 GFLOPS September 19, 2014
1.4 GHz
1.5 GHz 2 GB
A8X APL1021 T7001 3 billion 128 mm2
[78]
3 3-core 2 MB PowerVR
GX6850
[72][78][122]
8 32 256 450 MHz 230.4 GFLOPS 128-bit 2 channels
64-bit/channel
25.6 GB/s October 22, 2014
A9 APL0898 S8000 14 nm
FinFET
[125]
Samsung ≥ 2 billion 96 mm2
[126]
Twister 2 1.85 GHz
[127][128]
Dual-core 3 MB 4 MB (Victim)

[120][129]

PowerVR
GT7600
[72][130]
6 24 192 650 MHz 249.6 GFLOPS 64-bit 1 channel
64-bit/channel
LPDDR4-3200
(1600 MHz)
September 25, 2015
APL1022 S8003 16 nm
FinFET
[126][131]
[132]
TSMC 104.5 mm2
[126]
A9X APL1021 S8001 ≥ 3 billion 143.9 mm2
[131][88]
2.16 GHz
[133][134]

[120][131]
PowerVR
GT7850
[72][131]
12 48 384 499.2 GFLOPS 128-bit[c] 2 channels[d]
64-bit/channel
November 11, 2015
2.26 GHz 128-bit 2 channels
64-bit/channel
51.2 GB/s 4 GB
A10 Fusion APL1W24 T8010 3.3 billion 125 mm2
[132]
ARMv8.1-A Hurricane 2 1.64 GHz Zephyr 2 1.09 GHz Quad-core[e] P-core:
L1i: 64 KB
L1d: 64 KB

E-core:
L1i: 32 KB
L1d: 32 KB
P-core:
3 MB

E-core:
1 MB
4 MB PowerVR
GT7600
Plus
[135][72]
[136][137]
6 24 192 900 MHz 345.6 GFLOPS 64-bit 1 channel
64-bit/channel
25.6 GB/s 2 GB September 16, 2016
2.34 GHz
3 GB
A10X Fusion APL1071 T8011 10 nm
FinFET
[88]
≥ 4 billion 96.4 mm2
[88]
3 2.38 GHz 3 1.30 GHz 6-core[f] P-core:
8 MB

E-core:
1 MB

[138][139]
4 MB 12 48 384 1000 MHz 768.0 GFLOPS 128-bit 2 channels
64-bit/channel
51.2 GB/s 3 GB June 13, 2017
4 GB
A11
Bionic
APL1W72 T8015 4.3 billion 87.66 mm2
[140]
ARMv8.2-A
[141]
Monsoon 2 2.39 GHz Mistral 4 1.19 GHz 6-core 1st
generation Apple-
designed
3 12 192 1066 MHz 409.3 GFLOPS 2 600 billion OPS 64-bit 4 channels
16-bit/channel
LPDDR4X-4266
(2133 MHz)
34.1 GB/s 2 GB September 22, 2017
3 GB
A12
Bionic
APL1W81 T8020 7 nm (N7)
FinFET
6.9 billion 83.27 mm2
[142]
ARMv8.3-A
[143]
Vortex 2.49 GHz Tempest 4 1.59 GHz P-core:
L1i: 128 KB
L1d: 128 KB

E-core:
L1i: 32 KB
L1d: 32 KB
P-core:
8 MB

E-core:
2 MB
8 MB 2nd
generation Apple-
designed (Apple G11P)
4 16 256 1125 MHz 576.0 GFLOPS 8 5 TOPS September 21, 2018
4 GB
A12X Bionic APL1083 T8027 10 billion 135 mm2
[144]
4 8-core Second generation Apple-
designed (Apple G11G)
7
28 448 1.008 TFLOPS 128-bit 2 channels
64-bit/channel
68.2 GB/s November 7, 2018
6 GB
A12Z Bionic 8 32 512 1.152 TFLOPS March 25, 2020
16 GB June 22, 2020
A13
Bionic
APL1W85 T8030 7 nm (N7P)
FinFET
8.5 billion 98.48 mm2
[145]
ARMv8.4-A
[146]
Lightning 2 2.66 GHz Thunder 1.72 GHz 6-core P-core:
L1i: 128 KB
L1d: 128 KB

E-core:
L1i: 96 KB
L1d: 48 KB
P-core:
8 MB

E-core:
4 MB
16 MB 3rd
generation Apple-
designed
[147]
4 16
[148]
256 1350 MHz 691.2 GFLOPS 5.5 TOPS 64-bit 4 channels
16-bit/channel
34.1 GB/s 3 GB September 20, 2019
4 GB
A14
Bionic
APL1W01 T8101 5 nm (N5)
FinFET
11.8 billion 88 mm2
[149]
ARMv8.5-A
[150]
Firestorm 3.00 GHz Icestorm 1.82 GHz P-core:
L1i: 192 KB
L1d: 128 KB

E-core:
L1i: 128 KB
L1d: 64 KB
4th
generation Apple-
designed
[151][147][152]
[153]
1462.5 MHz 748.8 GFLOPS 16 11 TOPS October 23, 2020
4 GB
A15
Bionic
APL1W07
[154]
T8110 5 nm (N5P)
FinFET
15 billion 108.01 mm2
[154]
ARMv8.6-A
[150]
Avalanche 3.24 GHz Blizzard 2.02 GHz P-core:
12 MB

E-core:
4 MB
32 MB 5th
generation Apple-
designed
[155][156][157]
512
[148]
1338 MHz
[148][158]
1.370 TFLOPS[159] 15.8 TOPS 4 GB September 24, 2021
2.93 GHz 5 20
[158][160]
640
[158][160]
1.713 TFLOPS[161]
3.24 GHz 6 GB
A16
Bionic
APL1W10

[162]

T8120 4 nm
(N4P)
FinFET

[107][108]
[109][106]
[163]

16 billion 112.75 mm2
Everest
[164][165]
3.46 GHz Sawtooth
[164][165]
P-core:
16 MB

E-core:
4 MB

[166]

24 MB

[166]

6th
generation Apple-
designed
1398 MHz
[160]
1.789 TFLOPS
[160]
17 TOPS LPDDR5-6400 (3200 MHz) 51.2 GB/s September 16, 2022
A17
Pro
APL1V02 T8130 3 nm (N3B) FinFET 19 billion 103.80 mm2
Everest (2nd generation) 3.78 GHz
[167]
Sawtooth (2nd generation) 2.11 GHz
[167]
7th
generation Apple-
designed
35 TOPS 8 GB October 15, 2024
6 24 768 2.147 TFLOPS[168] September 22, 2023
A18 APL1V08 T1840a 3 nm (N3E) FinFET 90 mm2[169] ARMv9.2-A[170] Everest (3rd generation) 4.05 GHz Sawtooth (3rd generation) 2.42 GHz[171] P-core:
8 MB

E-core:
4 MB
12 MB

[172]

8th
generation Apple-
designed
4 16 512 1490 MHz[173] 1.526 TFLOPS LPDDR5X-7500 (3750 MHz) 60.0 GB/s[172] February 28, 2025
5 20[172] 640[172] 1.907 TFLOPS September 9, 2024
A18
Pro
APL1V07 T8140 105 mm2[169] P-core:
16 MB

E-core:
4 MB
24 MB

[172]

6 24[172] 768[172] 2.289 TFLOPS
A19 T8150a 3 nm (N3P) FinFET 4.26 GHz 2.6 GHz 9th
generation Apple-
designed
5 20 640 1620 MHz 2.074 TFLOPS[174] LPDDR5X-8533 (4266 MHz) 68.3 GB/s September 9, 2025
A19
Pro
T8150 12 GB
6 24 768 3.578 TFLOPS[175] LPDDR5X-9600 (4800 MHz) 75.8 GB/s

M-series SoCs

[edit]

The M series is a family of system on a chip (SoC) used in Mac computers from November 2020 and later, iPad Pro tablets from April 2021 and later, iPad Air tablets from March 2022 and later, and Vision Pro. The M designation was previously used for Apple motion coprocessors.

Apple M1

[edit]

The M1, Apple's first system on a chip designed for use in Macs, is manufactured using TSMC's 5 nm process. Announced on November 10, 2020, it was first used in the MacBook Air, Mac mini and 13-inch MacBook Pro, and later used in the iMac, 5th-generation iPad Pro and 5th-generation iPad Air. It comes with 4 performance cores and 4 efficiency cores, for a total of 8 CPU cores. It comes with up to 8 GPU cores, with the entry level MacBook Air having only 7 GPU cores. The M1 has 16 billion transistors[176] and operates at a clock frequency of up to 3.2 GHz.[177]

Apple M1 Pro

[edit]

The M1 Pro is a more powerful version of the M1, with six to eight performance cores, two efficiency cores, 14 to 16 GPU cores, 16 Neural Engine cores, up to 32 GB unified RAM with up to 200 GB/s memory bandwidth, and more than double the transistors. It was announced on October 18, 2021, and is used in the 14- and 16-inch MacBook Pro. Apple claimed the CPU performance is about 70% faster than the M1, and that its GPU performance is about double. Apple claims the M1 Pro can deliver up to 20 streams of 4K or 7 streams of 8K ProRes video playback (up from 6 offered by Afterburner card for 2019 Mac Pro). Like the M1, the M1 Pro operates at a clock frequency of up to 3.2 GHz.[178]

Apple M1 Max

[edit]

The M1 Max is a larger version of the M1 Pro chip, with eight performance cores, two efficiency cores, 24 to 32 GPU cores, 16 Neural Engine cores, up to 64 GB unified RAM with up to 400 GB/s memory bandwidth, and more than double the number of transistors. It was announced on October 18, 2021, and is used in the 14- and 16-inch MacBook Pro, as well as the Mac Studio. Apple claims the M1 Max can deliver up to 30 streams of 4K (up from 23 offered by Afterburner card for 2019 Mac Pro) or 7 streams of 8K ProRes video playback. Like the M1, the M1 Max operates at a clock frequency of up to 3.2 GHz.[179]

Apple M1 Ultra

[edit]

The M1 Ultra consists of two M1 Max dies connected together by a silicon interposer through Apple's UltraFusion interconnect.[180] It has 114 billion transistors, 16 performance cores, 4 efficiency cores, 48 to 64 GPU cores and 32 Neural Engine cores; it can be configured with up to 128 GB unified RAM of 800 GB/s memory bandwidth. It was announced on March 8, 2022, as an optional upgrade for the Mac Studio. Apple claims the M1 Ultra can deliver up to 18 streams of 8K ProRes video playback.[181] Like the M1 Max, the M1 Ultra operates at a clock frequency of up to 3.2 GHz.[182]

Apple M2

[edit]

Apple announced the M2 SoC on June 6, 2022, at WWDC, along with a redesigned MacBook Air and a revised 13-inch MacBook Pro and later the sixth-generation iPad Pro and the sixth-generation iPad Air. The M2 is made with TSMC's "enhanced 5-nanometer technology" N5P process and contains 20 billion transistors, a 25% increase from the previous generation M1, also operating at a higher clock frequency of up to about 3.5 GHz.[183] The M2 can be configured with up to 24 gigabytes of RAM and 2 terabytes of storage. It has 8 CPU cores (4 performance and 4 efficiency) and up to 10 GPU cores. The M2 also increases the memory bandwidth to 100 GB/s. Apple claims CPU improvements up to 18% and GPU improvements up to 35% compared to the previous M1.[184]

Apple M2 Pro

[edit]

The M2 Pro is a more powerful version of the M2, with six to eight performance cores, four efficiency cores, 16 to 19 GPU cores, 16 Neural Engine cores, up to 32 GB unified RAM with up to 200 GB/s memory bandwidth, and double the transistors. It was announced on January 17, 2023, in a press release and it is used in the 14- and 16-inch 2023 MacBook Pro as well as the Mac Mini. Apple claims the CPU performance is 20 percent faster than the M1 Pro and the GPU is 30 percent faster than the M1 Pro.[185] Like the M2, the M2 Pro operates at a clock frequency of up to about 3.5 GHz.[186]

Apple M2 Max

[edit]

The M2 Max is a larger version of the M2 Pro, with eight performance cores, four efficiency cores, 30 to 38 GPU cores, 16 Neural Engine cores, up to 96 GB unified RAM with up to 400 GB/s memory bandwidth, and more than double the transistors. It was announced on January 17, 2023, in a press release and it is used in the 14- and 16-inch 2023 MacBook Pro, as well as the Mac Studio.[187] Apple claims the CPU performance is 20 percent faster than M1 Max and the GPU is 30 percent faster than the M1 Max.[185] Unlike the M2 and the M2 Pro, the M2 Max operates at a clock frequency of up to about 3.6 GHz.[188]

Apple M2 Ultra

[edit]

The M2 Ultra consists of two M2 Max dies connected together by a silicon interposer through Apple's UltraFusion interconnect. It has 134 billion transistors, 16 performance cores, 8 efficiency cores, 60 to 76 GPU cores and 32 Neural Engine cores; it can be configured with up to 192 GB unified RAM of 800 GB/s memory bandwidth. It was announced on June 5, 2023, as an optional upgrade for the Mac Studio and the sole processor for the Mac Pro. Apple claims the M2 Ultra can deliver up to 22 streams of 8K ProRes video playback.[189] Like the M2 Max, the M2 Ultra operates at a clock frequency of up to about 3.6 GHz.[190]

Apple M3

[edit]

Apple announced the M3 series on October 30, 2023, along with the new MacBook Pro and iMac, later used in the MacBook Air and the seventh-generation iPad Air. The M3 is based on the 3nm process and contains 25 billion transistors, a 25% increase from the previous M2 generation, and operates at a clock frequency of up to 4.1 GHz.[191] It has 8 CPU cores (4 performance cores and 4 efficiency cores) and up to 10 GPU cores. Apple claims CPU improvements up to 35% and GPU improvements up to 65% compared to the M1 series.[192]

Apple M3 Pro

[edit]

The M3 Pro is a more powerful version of the M3, with five or six performance cores, six efficiency cores, 14 to 18 GPU cores, 16 Neural Engine cores, up to 36 GB unified RAM with 150 GB/s memory bandwidth, and 48% more transistors. It is used in the 14- and 16-inch MacBook Pro and operates at a clock frequency of up to about 4.1 GHz.[193] Apple claims the CPU performance is 30 percent faster than the M1 Pro and the GPU is 40 percent faster than the M1 Pro.[192]

Apple M3 Max

[edit]

The M3 Max is a larger version of the M3 Pro, with ten or twelve performance cores, four efficiency cores, 30 to 40 GPU cores, 16 Neural Engine cores, up to 128 GB unified RAM with up to 400 GB/s memory bandwidth, and more than double the transistors. It is used in the 14- and 16-inch MacBook Pro. Apple claims the CPU performance is 80 percent faster than the M1 Max and the GPU is 50 percent faster than the M1 Max.[192] Like the M3 and the M3 Pro, the M3 Max operates at a clock frequency of up to about 4.1 GHz.[194]

Apple M3 Ultra

[edit]

The M3 Ultra consists of two M3 Max dies connected together by a silicon interposer through Apple's UltraFusion interconnect. It has 184 billion transistors, 20 or 24 performance cores, 8 efficiency cores, 60 to 80 GPU cores and 32 Neural Engine cores; it can be configured with up to 512 GB unified RAM of 800 GB/s memory bandwidth. It was announced on March 5, 2025, as an optional upgrade for the Mac Studio, and operates at a clock frequency of up to about 4.0 GHz unlike the M3 Max.[195] Apple claims the M3 Ultra can deliver up to 24 streams of 8K ProRes video playback.[196]

Apple M4

[edit]

Apple announced the M4 chip on May 7, 2024, along with the seventh-generation iPad Pro; it would later be used for the iMac, Mac mini, MacBook Pro and MacBook Air. The M4 is based on a "second-generation 3-nanometer" process and contains 28 billion transistors. It has up to 10 CPU cores (3 or 4 performance and 4 or 6 efficiency) and up to 10 GPU cores. Apple claims the M4 has up to 1.5x faster CPU performance compared to the M2.[197] The M4 operates at a clock frequency of 4.4 GHz.[198]

The M4 is the first Apple Silicon using ARMv9.2-A instruction set without SVE2.[199]

Apple M4 Pro

[edit]

The M4 Pro is a more powerful version of the M4, with eight or ten performance cores, four efficiency cores, 16 to 20 GPU cores, 16 Neural Engine cores, and up to 64 GB unified RAM with 273 GB/s memory bandwidth. It is used in the 14- and 16-inch MacBook Pro as well as the Mac Mini. Apple claims the CPU performance is 1.9x faster than the M1 Pro and the GPU is 2x faster than the M1 Pro.[200] The M4 Pro operates at a clock frequency of 4.5 GHz.[201]

Apple M4 Max

[edit]

The M4 Max is a larger version of the M4 Pro, with ten or twelve performance cores, four efficiency cores, 32 to 40 GPU cores, 16 Neural Engine cores, and up to 128 GB unified RAM with up to 546 GB/s memory bandwidth. It is used in the 14- and 16-inch MacBook Pro as well as the Mac Studio. Apple claims the CPU performance is 2.2x faster than the M1 Max and the GPU is 1.9x faster than the M1 Max.[200] The M4 Max operates at a clock frequency of 4.5 GHz.[202]

Apple M5

[edit]

Apple announced the M5 chip on October 15, 2025, along with the new iPad Pro, 14-inch MacBook Pro and Vision Pro. The M5 is based on a "third-generation 3-nanometer" process. It has up to 10 CPU cores (3 or 4 performance and 4 or 6 efficiency) and up to 10 GPU cores.[203]

Comparison of M-series processors

[edit]
General Semiconductor technology CPU GPU AI accelerator Media Engine Memory technology First release
Name Codename
and part no.
Image Process Transistor count Die size Transistor density CPU ISA Performance core Efficiency core Overall cores Cache Vendor Cores SIMD EU count FP32 ALU count Frequency FP32 FLOPS
(TFLOPS)
Hardware-accelerated ray tracing Cores OPS Hardware Acceleration Media Decode/Encode Engine Memory bus width Total channel
Bit per channel
Memory type Theoretical
bandwidth
Available capacity
Core name Cores Core speed Core name Cores Core speed L1 L2 SLC Video decode Video encode ProRes decode & encode AV1 decode
M1 APL1102
T8103
Apple M1 processor TSMC
N5
16 billion 118.91 mm2[204] ~134 MTr/mm2 ARMv8.5-A
[150]
Firestorm 4 3.20 GHz Icestorm 4 2.06 GHz 8-core P-core:
L1i: 192 KB
L1d: 128 KB

E-core:
L1i: 128 KB
L1d: 64 KB
P-core:
12 MB

E-core:
4 MB
8 MB 4th generation Apple-designed 7 28 896 1278 MHz 2.290 No 16 11 TOPS H264, HEVC 1 1 128-bit 2 channels
64-bit/channel
LPDDR4X-4266
(2133 MHz)
68.25 GB/s 8 GB
16 GB
November 17, 2020
8 32 1024 2.617
M1 Pro APL1103
T6000
Apple M1 Pro processor 33.7 billion ≈ 245 mm2
[205]
~137 MTr/mm2 6 3.23 GHz 2 P-core:
24 MB

E-core:
4 MB
24 MB 14 56 1792 1296 MHz 4.644 H264, HEVC, ProRes, ProRes RAW 1 256-bit 2 channels
128-bit/channel
LPDDR5-6400
(3200 MHz)
204.8 GB/s 16 GB
32 GB
October 26, 2021
8 10-core
16 64 2048 5.308
M1 Max APL1105
T6001
[206]
Apple M1 Max processor 57 billion ≈ 432 mm2
[205]
~132 MTr/mm2 48 MB 24 96 3072 7.962 2 2 512-bit 4 channels
128-bit/channel
409.6 GB/s 32 GB
64 GB
32 128 4096 10.616
M1 Ultra APL1W06
T6002
Apple M1 Ultra processor 114 billion ≈ 864 mm2 16 4 20-core P-core:
48 MB

E-core:
8 MB
96 MB 48 192 6144 15.925 32 22 TOPS 2 4 4 1024-bit 8 channels
128-bit/channel
819.2 GB/s 64 GB
128 GB
March 18, 2022
64 256 8192 21.233
M2 APL1109
T8112
Apple M2 processor TSMC
N5P
20 billion 155.25 mm2
[204]
~129 MTr/mm2 ARMv8.6-A
[150]
Avalanche 4 3.50 GHz Blizzard 4 2.42 GHz 8-core P-core:
16 MB

E-core:
4 MB
8 MB 5th generation Apple-designed 8 32 1024 1398 MHz 2.863 16 15.8 TOPS 1 1 1 128-bit 2 channels
64-bit/channel
102.4 GB/s 8 GB
16 GB
24 GB
June 24, 2022
9 [207] 36 1152 3.578 H264, HEVC
10 40 1280 H264, HEVC, ProRes, ProRes RAW 1
M2 Pro APL1113
T6020
40 billion ~289 mm2[208] ~138 MTr/mm2 6 10-core P-core:
32 MB

E-core:
4 MB
24 MB 16 64 2048 5.726 256-bit 4 channels
64-bit/channel
204.8 GB/s 16 GB
32 GB
January 24, 2023
8 12-core 19 76 2432 6.799
M2 Max APL1111
T6021
67 billion 3.69 GHz
[209]
48 MB 30 120 3840 10.736 2 2 512-bit 4 channels
128-bit/channel
409.6 GB/s 32 GB
64 GB
96 GB
38 152 4864 13.599
M2 Ultra APL1W12
T6022
134 billion 16 ~3.00 GHz
-3.70 GHz
[209][210][211]
8 24-core P-core:
64 MB

E-core:
8 MB
96 MB 60 240 7680 21.473 32 31.6 TOPS 2 4 4 1024-bit 8 channels
128-bit/channel
819.2 GB/s 64 GB
128 GB
192 GB
June 13, 2023
76 304 9728 27.199
M3 APL1201
T8122
TSMC
N3B
25 billion 4 4.05 GHz 4 2.75 GHz 8-core P-core:
16 MB

E-core:
4 MB
8 MB 7th generation Apple-designed 8 128 1024 1380 MHz

[212]

2.826 Yes 16 18 TOPS 1 1 1 1 128-bit 2 channels
64-bit/channel
102.4 GB/s 8 GB
16 GB
24 GB
November 7, 2023
10 160 1280 3.533
M3 Pro APL1203
T6030
37 billion 5 6 11-core 12 MB 14 224 1792 4.946 192-bit 3 channels
64-bit/channel
153.6 GB/s 18 GB
36 GB
6 12-core 18 288 2304 6.359
M3 Max APL1204
T6034
92 billion 10 4 14-core P-core:
32 MB

E-core:
4 MB
48 MB 30 480 3840 10.598 2 2 384-bit 3 channels
128-bit/channel
307.2 GB/s 36 GB
96 GB
APL1204
T6031
12 16-core 40 640 5120 14.131 512-bit 4 channels
128-bit/channel
409.6 GB/s 48 GB
64 GB
128 GB
M3 Ultra T6032 184 billion 20 8 28-core P-core:
64 MB

E-core:
8 MB
96 MB 60 960 7680 21.197 36 TOPS 2 4 4 2 1024-bit 8 channels
128-bit/channel
819.2 GB/s 96 GB
256 GB
March 12, 2025
24 32-core 80 1280 10240 28.262 96 GB
256 GB
512 GB
M4 APL1206
T8132
TSMC
N3E
28 billion ARMv9

[213]

4 4.40 GHz 4 2.85 GHz 8-core P-core:
16 MB

E-core:
4 MB
8th generation Apple-designed 8 128 1024 1470 MHz

[214]

38 TOPS 1 1 1 1 128-bit 2 channels
64-bit/channel
LPDDR5X-7500 (3750 MHz) 120 GB/s 8 GB
16 GB
24 GB
32 GB
May 15, 2024
3 6 9-core 10 160 1280 4.26[215]
4 10-core
M4 Pro T6040 8 4.51 GHz 4 12-core P-core:
2×16 MB
E-core:
4 MB
16 256 2048 1578 MHz 6.82[216] 256-bit LPDDR5X-8533 (4266 MHz) 273 GB/s 24 GB
48 GB
64 GB
November 8, 2024
10 14-core 20 320 2560 8.52[217]
M4 Max T6041 10 4 14-core 32 512 4096 13.64[218] 384-bit 409.6 GB/s 36 GB
12 16-core 40 640 5120 17.04[219] 512-bit 546 GB/s 48 GB
64 GB
128 GB

R-series SoCs

[edit]

The R series is a family of low-latency system on a chip (SoC) for real-time processing of sensor inputs.

Apple R1

[edit]

The Apple R1 was announced by Apple on June 5, 2023, at its Worldwide Developers Conference. It is used in the Apple Vision Pro headset. The Apple R1 is dedicated to real-time processing of sensor inputs and delivers extremely low-latency images to the display.

S-series SiPs

[edit]

The Apple S series is a family of systems in a package (SiP) used in the Apple Watch and HomePod. It uses a customized application processor that together with memory, storage and support processors for wireless connectivity, sensors, and I/O form a complete computer in a single package.

Apple S1

[edit]

The Apple S1 is an integrated computer. It includes memory, storage and support circuits like wireless modems and I/O controllers in a sealed integrated package. It was announced on September 9, 2014, as part of the "Wish we could say more" event. It was used in the first-generation Apple Watch.[220]

Apple S1P

[edit]

Used in Apple Watch Series 1. It has a dual-core processor identical to the S2, with the exception of the built-in GPS receiver. It contains the same dual-core CPU with the same new GPU capabilities as the S2, making it about 50% faster than the S1.[221][222]

Apple S2

[edit]

Used in the Apple Watch Series 2. It has a dual-core processor and a built-in GPS receiver. The S2's two cores deliver 50% higher performance and the GPU delivers twice as much as the predecessor,[223] and is similar in performance to the Apple S1P.[224]

Apple S3

[edit]

Used in the Apple Watch Series 3. It has a dual-core processor that is 70% faster than the Apple S2 and a built-in GPS receiver.[225] There is also an option for a cellular modem and an internal eSIM module.[225] It also includes the W2 chip.[225] The S3 also contains a barometric altimeter, the W2 wireless connectivity processor, and in some models UMTS (3G) and LTE (4G) cellular modems served by a built-in eSIM.[225]

Apple S4

[edit]

Used in the Apple Watch Series 4. It introduced 64-bit ARMv8 cores to the Apple Watch through two Tempest cores,[226] which are also found in the A12 as energy-efficient cores. Despite its small size, Tempest uses a 3-wide decode out-of-order superscalar design, which makes it much more powerful than preceding in-order cores.

The S4 contains a Neural Engine that is able to run Core ML.[227] Third-party apps can use it starting from watchOS 6. The SiP also includes new accelerometer and gyroscope functionality that has twice the dynamic range in measurable values of its predecessor, as well as being able to sample data at 8 times the speed.[228] It contains the W3 wireless chip, which supports Bluetooth 5. It also contains a new custom GPU, which can use the Metal API.[229]

Apple S5

[edit]

Used in the Apple Watch Series 5, Watch SE, and HomePod mini.[230] It adds a built-in magnetometer to the custom 64-bit dual-core processor and GPU of the S4.[231]

Apple S6

[edit]

Used in the Apple Watch Series 6. It has a custom 64-bit dual-core processor that runs up to 20 percent faster than the S5.[232][233] The dual-cores in the S6 are based on the A13 Bionic's energy-efficient "little" Thunder cores at 1.8 GHz.[234] Like the S4 and S5, it also contains the W3 wireless chip.[233] The S6 adds the new U1 ultrawide band chip, an always-on altimeter, and 5 GHz WiFi.[232][233]

Apple S7

[edit]

Used in the Apple Watch Series 7 and second-generation HomePod. The S7 CPU has the same T8301 identifier and quoted performance as the S6. It is the second time utilizing the energy-efficient "little" Thunder cores of the A13 Bionic.[235]

Apple S8

[edit]

Used in the Apple Watch SE (2nd generation), Watch Series 8, and Watch Ultra.[236] The S8 CPU has the same T8301 identifier and quoted performance as the S6 and S7. It is the final CPU to utilize the energy-efficient "little" Thunder cores of the A13 Bionic.[237]

Apple S9

[edit]

Used in the Apple Watch Series 9 and Watch Ultra 2. The S9 CPU has a new dual-core CPU with 60 percent more transistors than the S8, a new four-core Neural Engine and the new U2 ultra-wide band chip. The dual-cores in the S9 are based on the A16 Bionic's energy efficient "little" Sawtooth cores.[238]

Apple S10

[edit]

Used in the Apple Watch Series 10, Series 11, SE 3, and Watch Ultra 3. The S10 CPU is the second time utilizing the energy-efficient "little" Sawtooth cores of the A16 Bionic.

Comparison of S-series processors

[edit]
Name Model no. Part no. Image Semiconductor technology Die size CPU ISA CPU CPU cache GPU Memory technology Modem First release
S1 APL
0778
[239]
S7002 28 nm MG[240][241] 32 mm2[240] ARMv7k[241][242] 520 MHz single-core Cortex-A7[241] L1d: 32 KB[243]
L2: 256 KB[243]
PowerVR Series 5[241][244] LPDDR3[245] TBC April 24, 2015
S1P TBC T8002 TBC ARMv7k[246][221][223] 520 MHz dual-core Cortex-A7[246] L1d: 32 KB[243] PowerVR Series 6 'Rogue'[246] September 12, 2016
S2
S3 T8004 ARMv7k[247] Dual-core TBC LPDDR4 Qualcomm MDM9635M
Snapdragon X7 LTE
September 22, 2017
S4 APL1W82 T8006 7 nm (TSMC N7) TBC ARMv8.3-A ILP32[248][249]
[150]
1.59 GHz Dual-core Tempest L1d: 32 KB[241]
L2: 2 MB[241]
Apple G11M[249] LPDDR4X TBC September 21, 2018
S5 September 20, 2019
S6 APL1W86 T8301 7 nm (TSMC N7P) ARMv8.4-A[150] 1.8 GHz Dual-core Thunder L1d: 48 KB[250]
L2: 4 MB[251]
TBC September 18, 2020
S7 October 15, 2021
S8 September 16, 2022
S9 APL1W15 T8310 4 nm (TSMC N4)[252] ARMv8.6-A[150] Dual-core Sawtooth L1d: 64 KB
L2: 4 MB[253]
LPDDR5 September 22, 2023
S10 September 20, 2024

T-series SoCs

[edit]

The T series performs various functions on Intel-based MacBook and iMac computers released from 2016 onwards. The chip processes and encrypts biometric information (Touch ID) and acts as a gatekeeper to the microphone and FaceTime HD camera, protecting them from hacking. The chip runs bridgeOS, a purported variant of watchOS.[254] The functions of the T-series processor were built into the M-series CPUs, thus ending the need for the T series.

Apple T1

[edit]

The Apple T1 chip is an ARMv7 SoC (derived from the processor in the Apple Watch's S2) that drives the System Management Controller (SMC) and Touch ID sensor of the 2016 and 2017 MacBook Pro with Touch Bar.[255]

Apple T2

[edit]

The Apple T2 security chip is a SoC first released in the iMac Pro. It includes a 64-bit ARMv8 processor (a variant of the processor in the A10 Fusion, or T8010)[256][257] and a separate Security Enclave Processor.[258][257] It provides a secure enclave for encrypted keys using the Security Enclave Processor, enables users to lock down the computer's boot process, handles system functions like the camera and audio control, and handles on-the-fly encryption and decryption for the solid-state drive.[259][260][261] T2 also delivers "enhanced imaging processing" for the iMac Pro's FaceTime HD camera.[262][263]

Comparison of T-series processors

[edit]
Name Model no. Image Semiconductor technology Die size CPU ISA CPU CPU cache GPU Memory technology First release
Memory bandwidth
T1 APL
1023
[264]
Apple T1 Processor TBC TBC ARMv7 TBD November
12, 2016
T2 APL
1027
[265]
Apple T2 Processor TSMC 16 nm FinFET.[266] 104 mm2[266] ARMv8-A
ARMv7-A
2× Hurricane
2× Zephyr
+ Cortex-A7
L1i: 64 KB
L1d: 64 KB
L2: 3 MB[266]
3× cores[266] LP-DDR4[266] December
14, 2017

C series Cellular modems

[edit]

The Apple C series is a family of cellular modem chips.

Apple C1

[edit]

Apple C1 is a cellular modem chip introduced in the iPhone 16e.[267] It is built on the N4 process node by TSMC.[268] It supports UMTS/HSPA+ and 5G (sub-6 GHz), but lacks DC-HSDPA and mmWave, which are supported by other iPhone 16 models. Apple claims that the C1 is more power efficient than previous iPhone modems and consumes 20–25% less power than the Qualcomm modems used in other iPhone 16 models.[269][270]

Apple C1X

[edit]

Apple C1X is a variant of the C1, offering twice the speed. It was released with the iPhone Air[271] in Sept 2025, and is also used in M5 based eighth-generation iPad Pro.

U series SiPs (Ultra-wideband)

[edit]

The Apple U series is a family of systems in a package (SiP) implementing ultra-wideband (UWB) radio.

Apple U1

[edit]

The Apple U1 is used in the iPhone 11/11 Pro series through the iPhone 14/14 Pro series (excluding the second and third generation iPhone SE), Apple Watch Series 6 through Series 8, Apple Watch Ultra (1st generation), HomePod (2nd generation), HomePod Mini, AirTag, and the charging case for AirPods Pro (2nd generation).[272]

Apple U2

[edit]

The Apple U2 (also referred to by Apple as its "Second-generation Ultra Wideband chip") is used in the iPhone 15/15 Pro series and newer (excluding iPhone 16e), iPhone Air, Apple Watch Series 9 and newer, Apple Watch Ultra 2 and newer, and the charging case for AirPods Pro (3rd generation).

Comparison of U-series processors

[edit]
Name Model no. Image CPU Semiconductor technology First release
U1 TMK

A75
[273]

Apple U1 chip Cortex-M4
ARMv7E-M
[274]
16 nm FinFET
(TSMC 16FF)
September 20, 2019
U2 September 22, 2023

W2+ and N series SoCs (Wireless connectivity)

[edit]

The Apple W, starting with the W2, and N series are a family of RF SoCs used for wireless connectivity (Bluetooth, Wi-Fi and Thread (N series only)).

Apple W2

[edit]

The Apple W2, used in the Apple Watch Series 3, is integrated into the Apple S3 SiP. Apple claimed the chip makes Wi-Fi 85% faster and allows Bluetooth and Wi-Fi to use half the power of the W1 implementation.[225]

Apple W3

[edit]

The Apple W3 is used in the Apple Watch Series 4 and newer, SE (1st generation) and newer, and Ultra (1st generation) and newer. It is integrated into the Apple S4 through S10 SiPs. It supports Bluetooth 5.0/5.3.

Apple N1

[edit]

The Apple N1 is used in iPhone 17, iPhone 17 Pro, and iPhone Air. It incorporates Wi-Fi 7, Bluetooth 6, and Thread into one chip. Apple claims the chip will help boost the performance and reliability of features like AirDrop and Personal Hotspot.[271]

Comparison of W-series processors

[edit]
Name Model no. Image Semiconductor technology Die size Bluetooth certification First release
W2 338S00348[275] Apple W2 chip TBC 4.2 September
22, 2017
W3 338S00464[276] Apple W3 chip 5.0/5.3 September
21, 2018

W1 and H series SoCs (Bluetooth/audio processing)

[edit]

The Apple W1 and the H series are a family of SoCs with Bluetooth wireless connectivity and low-power audio processing for use in headphones and speakers.

Apple W1

[edit]

The Apple W1 is a SoC used in the 2016 AirPods and select Beats headphones.[277][278] It maintains a Bluetooth[279] Class 1 connection with a computer device and decodes the audio stream that is sent to it.[280] Its die size is 14.3 mm2.[281]

Apple H1

[edit]

The Apple H1 chip was used in the second and third generation AirPods and the first generation AirPods Pro. It was also used in the Powerbeats Pro, the Beats Solo Pro, Beats Fit Pro, the 2020 Powerbeats, and AirPods Max.[282] Specifically designed for headphones, it has Bluetooth 5.0, supports hands-free "Hey Siri" commands,[283] and offers 30 percent lower latency than the W1 chip used in earlier AirPods.[284]

Apple H2

[edit]

The Apple H2 chip is used in AirPods 4, AirPods Pro 2 and AirPods Pro 3. It has Bluetooth 5.3, and implements 48 kHz noise reduction in hardware. The 2022 version of the H2 operates only on the 2.4 GHz frequency, while the 2023 version adds support for audio transmission using a proprietary protocol in two specific frequency ranges of the 5 GHz band.[285]

Comparison of Bluetooth audio processors

[edit]
Name Model no. Image Bluetooth certification First release
W1 343S00130[281]
343S00131[281]
Apple W1 chip 4.2 December
13, 2016
H1 343S00289[286]
(AirPods 2nd Generation)
343S00290[287]
(AirPods 3rd Generation)
343S00404[288]
(AirPods Max)
H1 SiP[289]
(AirPods Pro)
Apple H1 chip Apple H1 chip Apple H1 chip
Apple H1 SiP Apple H1 SiP
5.0 March 20, 2019
H2 AirPods (4th generation)
AirPods Pro (2nd generation)[290]

Airpods Pro (3rd generation)[291]
Apple Vision Pro

5.3 September 7, 2022

M-series motion coprocessors

[edit]

The M series represents motion coprocessors used by Apple Inc. in their mobile devices. First released in 2013, their function is to collect sensor data from integrated accelerometers, gyroscopes, and compasses. They offload collecting and processing sensor data from the main central processing unit (CPU).

Only the M7 and M8 motion coprocessors were housed on separate chips, while the M9, M10, and M11 were embedded in their corresponding A-series chips. Beginning with the A12 Bionic chip in 2018, the motion coprocessors were fully integrated into the SoC. Apple eventually reused the M codename for their desktop SoCs.

Comparison of M-series motion coprocessors

[edit]
Name Model no. Image Semiconductor technology CPU ISA CPU First release
Apple M7 LPC18A1 NXP LPC18A1 90 nm ARMv7-M 150 MHz Cortex-M3 September
10, 2013
Apple M8 LPC18B1 NXP LPC18B1 September
9, 2014

Miscellaneous devices

[edit]

This segment is about Apple-designed processors that are not easily sorted into another section.

Early series

[edit]

Apple first used Samsung-developed SoCs in early versions of the iPhone and iPod Touch. They combine in one package a single ARM-based processing core (CPU), a graphics processing unit (GPU), and other electronics necessary for mobile computing.

The APL0098 (also 8900B[292] or S5L8900) is a package on package (PoP) system on a chip (SoC) that was introduced on June 29, 2007, at the launch of the original iPhone. It includes a 412 MHz single-core ARM11 CPU and a PowerVR MBX Lite GPU. It was manufactured by Samsung on a 90 nm process.[11] The iPhone 3G and the first-generation iPod Touch also use it.[293]

The APL0278[294] (also S5L8720) is a PoP SoC introduced on September 9, 2008, at the launch of the second-generation iPod Touch. It includes a 533 MHz single-core ARM11 CPU and a PowerVR MBX Lite GPU. It was manufactured by Samsung on a 65 nm process.[11][293]

The APL0298 (also S5L8920) is a PoP SoC introduced on June 8, 2009, at the launch of the iPhone 3GS. It includes a 600 MHz single-core Cortex-A8 CPU and a PowerVR SGX535 GPU. It was manufactured by Samsung on a 65 nm process.[112]

The APL2298 (also S5L8922) is a 45 nm die shrunk version of the iPhone 3GS SoC[11] and was introduced on September 9, 2009, at the launch of the third-generation iPod Touch.

Other

[edit]

The Samsung S5L8747 is an ARM-based microcontroller used in Apple's Lightning Digital AV Adapter, a Lightning-to-HDMI adapter. This is a miniature computer with 256 MB RAM, running an XNU kernel loaded from the connected iPhone, iPod Touch, or iPad, then taking a serial signal from the iOS device translating that into a proper HDMI signal.[295][296]

Model no. Image First release CPU ISA Specs Application Utilizing devices Operating system
339S0196 339S0196 microcontroller September 2012 Unknown

ARM

256 MB
RAM
Lightning to
HDMI conversion
Apple Digital
AV Adapter
XNU

See also

[edit]

Similar platforms

[edit]

Notes

[edit]

References

[edit]

Further reading

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia
from Grokipedia
Apple silicon is a family of custom-designed, ARM-based system-on-a-chip (SoC) processors developed by Apple Inc. for use in its Mac computers, iPad tablets, iPhones, Apple Watch, and other devices, emphasizing high performance, energy efficiency, and integration of CPU, GPU, Neural Engine, and other components on a single chip. This includes the A-series for iPhones and iPads, M-series for Macs and high-end iPads, S-series for Apple Watch, and others. The term "Apple silicon" was first introduced by Apple in 2020 for its M-series chips in Macs, and subsequently applied to the earlier A-series and other custom processors like the S-series. Apple's use of its own silicon in Macs began with the announcement on June 22, 2020, of a two-year transition from Intel x86 processors to Apple silicon, with the first Apple silicon Mac—the MacBook Air, MacBook Pro, and Mac mini featuring the M1 chip—shipping in November 2020. This shift enables a unified architecture across Apple's ecosystem, facilitating seamless app development and support for iOS and iPadOS apps on Mac without modifications. The M-series chips form the core of Apple silicon for Mac computers, evolving through multiple generations to power an expanding range of devices. The initial M1 family, introduced in 2020, includes the base M1 (with an 8-core CPU, 7- or 8-core GPU, and 16-core Neural Engine) followed by M1 Pro and M1 Max in 2021, offering up to 10-core CPUs, 32-core GPUs, and 64 GB of unified memory for professional workflows. The M2 series debuted in June 2022 with enhanced media engines and up to 24 GB unified memory, expanding to M2 Pro, M2 Max, and M2 Ultra in 2023 for desktops like Mac Studio and Mac Pro. The M3 family, launched in October 2023, introduced 3-nanometer process technology for denser transistors and hardware-accelerated ray tracing, with variants like M3 Pro and M3 Max powering updated MacBook Pros and iMacs. The M4 series arrived in May 2024, initially for iPad Pro, featuring a more efficient 10-core CPU and advanced AI capabilities, later extending to M4 Pro and M4 Max in Macs for improved graphics and up to 128 GB unified memory. Most recently, the M5 chip was unveiled in October 2025, delivering significant AI performance gains with a 10-core CPU, 10-core GPU, and enhanced Neural Engine, integrated into products like the 14-inch MacBook Pro and new iPad Pro. Central to Apple silicon's design is its unified memory architecture, which integrates LPDDR-type RAM shared at high bandwidth (up to 400 GB/s in M1 Max) between the CPU, GPU, and Neural Engine, with current Mac configurations starting at 16 GB and no 8 GB options available, reducing latency and boosting efficiency compared to traditional discrete memory systems. Due to the shared high-bandwidth unified memory and macOS optimizations, including efficient swapping to fast SSD storage, 8-16 GB configurations on Apple Silicon often provide effective performance comparable to higher RAM amounts in systems with discrete memory, such as Intel-based Macs or Windows PCs, making them sufficient for many users' workflows. Unified memory provides faster model loading and greater efficiency than separate VRAM, reducing bottlenecks in heavy multitasking and AI processing. It particularly enables efficient inference of large language models by providing a shared memory pool that allows seamless access to large datasets without data copying bottlenecks. This enables breakthroughs in power efficiency—such as up to 2x battery life in early M1 devices—and superior performance per watt, with the M1 delivering up to 3.5x faster CPU speeds than Intel predecessors while consuming less energy. Additional technologies include the Apple Neural Engine for machine learning tasks (up to 15x faster in M1), a dedicated media engine for hardware-accelerated video encoding like ProRes, and dynamic caching for GPU optimization, making these chips ideal for creative, AI, and professional applications. Apple silicon has transformed the Mac lineup, completing the Intel transition by mid-2022 and powering devices from entry-level MacBook Airs to high-end Mac Pros, while extending to iPads for enhanced portability and AI features like those in Apple Intelligence. Its focus on integration and efficiency has set new benchmarks in personal computing, enabling features such as hardware-accelerated ray tracing in M3 and beyond, and positioning Apple at the forefront of on-device AI processing.

Overview

History and Development

Apple's journey with ARM-based processors began in 2008 with the , which featured a Samsung-fabricated 32-bit RISC ARM 1176JZ(F)-S v1.0 processor clocked at 412 MHz, marking the company's initial adoption of the energy-efficient ARM architecture for mobile devices. This choice was driven by the need for low-power computing suitable for battery-constrained smartphones, relying on third-party manufacturers like Samsung for chip production. To build in-house expertise, Apple acquired in April 2008 for $278 million, gaining a team of engineers skilled in low-power processor design that laid the foundation for future custom silicon efforts. This was followed by the 2010 acquisition of for approximately $121 million, a firm specializing in high-speed ARM core optimizations, which contributed to enhancing Apple's processor performance. These moves enabled Apple to license the ARM instruction set architecture while developing proprietary cores, shifting from off-the-shelf designs to fully custom system-on-chips (SoCs) for greater control over performance and power efficiency. The first fruit of this strategy was the SoC, introduced in 2010 with the first-generation and later the , representing Apple's inaugural fully custom-designed processor tailored for seamless integration with its ecosystem. This marked a pivotal transition to in-house design, motivated by the desire to optimize hardware-software synergy, reduce dependency on external suppliers, and address limitations in third-party chips' efficiency for Apple's devices. Subsequent A-series chips built on this foundation, incorporating custom ARM-based cores to prioritize battery life and app performance within the iOS environment. By the late 2010s, frustrations with Intel's stagnant processor advancements and supply chain vulnerabilities prompted Apple to extend custom silicon to its Mac lineup. In June 2020, at its , Apple announced the transition of Macs to Apple silicon—marking the first official use of the term "Apple silicon" by the company to describe its custom ARM-based processors—aiming to unify its ecosystem across mobile and desktop while enhancing power efficiency and integration. The first M1 chips debuted in November 2020 with the , 13-inch , and , completing the shift from Intel by mid-2023 and enabling optimizations specific to macOS. Ongoing partnerships with for advanced fabrication nodes have supported this evolution, allowing Apple to scale production while maintaining supply chain resilience. In October 2025, Apple introduced the chip across devices like the 14-inch MacBook Pro and iPad Pro, emphasizing AI performance with up to 3.5 times faster neural processing compared to the , further solidifying custom silicon's role in on-device intelligence and ecosystem control.

Core Architecture and Technologies

Apple silicon employs the instruction set architecture and its extensions, enabling 64-bit processing with advanced features such as pointer authentication and memory tagging for enhanced security. Beginning with the A14 Bionic and M1 chips, Apple introduced fully custom CPU core designs in a configuration, featuring high-performance cores codenamed Firestorm (later iterations include Avalanche and Sawtooth) paired with energy-efficient cores like Icestorm (evolving to Blizzard and Everfrost). These custom cores optimize for both single-threaded peak performance and multi-threaded workloads, with Firestorm cores delivering up to 3.2 GHz clock speeds and wider execution units compared to off-the-shelf ARM designs, while Icestorm cores prioritize low-power operation for background tasks. A hallmark of Apple silicon is its (UMA), where high-bandwidth LPDDR memory—such as LPDDR5X in recent generations—is integrated directly on the package and shared seamlessly among the CPU, GPU, Neural Engine, and other accelerators. This design eliminates the need for data copying between separate CPU and GPU memory pools, reducing latency to under 100 ns for inter-component access and enabling bandwidths exceeding 400 GB/s in high-end configurations like the M3 Max. By pooling memory into a single address space, UMA supports efficient task allocation, such as dynamic sharing for machine learning models that span CPU and GPU processing, while maintaining power efficiency through on-chip proximity that minimizes external bus traffic. Apple Silicon handles virtualization efficiently via its integrated SoC design and the Virtualization framework, but heavier multi-VM workloads demand high unified memory capacity to avoid performance issues from shared pool contention. The , Apple's dedicated neural processing unit (NPU), has evolved significantly since its debut in the with a 2-core design capable of 0.6 trillion operations per second (TOPS) for machine learning inference tasks like facial recognition. Subsequent generations scaled to 8 cores in the A12 (5 TOPS), which was maintained in the A13, before expanding to 16 cores in the and (11 TOPS), culminating in advanced configurations like the enhanced unit in the M5, delivering up to 3.5 times the neural processing performance of the M4 for on-device AI workloads including natural language processing and image generation. This NPU integrates specialized matrix multiplication hardware and supports frameworks like , offloading compute-intensive operations from the CPU to achieve up to 15x faster inference than general-purpose cores while consuming minimal power. For transformer-based machine learning models, M-series chips leverage unified memory and accelerators for efficient on-device inference, supporting cost-effective deployment of large models. Benchmarks show competitive performance in latency and throughput for many tasks, though NVIDIA GPUs with Tensor Cores often achieve superior speeds in matrix-heavy operations due to specialized hardware. The MLX framework optimizes inference on Apple hardware, outperforming Metal in efficiency, but the CUDA ecosystem provides more extensive tools and community optimizations for advanced features. Quantization is facilitated via software, yet Apple Silicon lacks dedicated acceleration for low-precision formats like FP4 and FP8 available on NVIDIA hardware. While unified memory enables seamless handling of large contexts, iterative workloads may encounter bandwidth constraints relative to dedicated VRAM setups, balanced by Apple's advantages in power efficiency for edge applications. Apple's GPUs represent a custom evolution from the architecture licensed earlier, utilizing tile-based deferred rendering (TBDR) since the early A-series to enhance power efficiency in mobile and desktop environments. In TBDR, the render target is divided into small tiles (typically 16x16 or 32x32 pixels), where geometry processing occurs first to identify visible fragments before shading, reducing overdraw and memory bandwidth by up to 50% compared to immediate-mode rendering. Core counts vary from 4 in early A-series to 40 in M4 Max, with each core featuring unified shaders for flexible vertex, pixel, and compute workloads; hardware-accelerated ray tracing, introduced in the A17 Pro and M3, uses dedicated intersection engines and bounding volume hierarchy traversal for realistic lighting simulations without excessive power draw. At the system level, Apple silicon integrates the CPU, GPU, I/O controllers, and accelerators into a monolithic system-on-chip (SoC) design, with the providing an isolated coprocessor for cryptographic operations like key management and biometric authentication. This enclave runs a separate secure OS () with dedicated AES engines and memory protection, ensuring data isolation via hardware-enforced boundaries and preventing side-channel attacks. The media engine handles hardware-accelerated video processing, supporting codecs like H.264, HEVC, and ProRes, with AV1 decode added from the M3 family to enable efficient streaming playback at up to 8K resolution while extending battery life by offloading decode tasks from the CPU. Key enabling technologies include a high-bandwidth on-chip fabric for inter-component communication, which provides deterministic latency and supports up to 2.5 TB/s bandwidth in multi-die configurations like M1 Ultra via UltraFusion interconnects. Dynamic caching, introduced in the M3 GPU, allocates shader memory in real-time based on workload demands, boosting utilization by 20-30% for ray-traced scenes and reducing waste in variable-complexity tasks. Power gating mechanisms, applied at the core and cluster levels, dynamically isolate unused sections of the SoC to cut leakage current, contributing to overall efficiency where the GPU matches performance at half the power. These features collectively enable Apple silicon to deliver sustained performance in thermally constrained devices, such as 20+ hours of video playback on battery. Compared to Intel-based Macs, Apple Silicon's efficient architecture largely mitigates thermal throttling issues by delivering higher performance per watt and lower heat output for equivalent workloads; however, in fanless designs like the MacBook Air, thermal throttling limits sustained GPU performance under high loads, leading to lower scores in benchmarks like Geekbench Metal compared to fan-cooled models like the MacBook Pro that maintain higher clocks. Laptops remain subject to physical cooling constraints relative to desktops.

Manufacturing and Process Nodes

Apple's early Apple silicon chips, from the Apple A4 through the Apple A8, were fabricated by Samsung using process nodes ranging from 45 nm to 20 nm. Beginning with the Apple A9 in 2015, Apple partnered with TSMC for production on its 14 nm FinFET process, initially sharing orders with Samsung before transitioning to TSMC exclusivity for subsequent generations due to superior yields and performance. The progression of process nodes has enabled significant scaling in performance and efficiency. The Apple A14 Bionic and Apple M1, introduced in 2020, marked Apple's first use of TSMC's 5 nm node, delivering higher transistor counts in a compact die size. In 2023, the Apple A17 Pro and Apple M3 shifted to TSMC's 3 nm N3 process, which supports up to 19 billion transistors in the A17 while improving power efficiency by approximately 25% over 5 nm. The Apple M4 in 2024 adopted the enhanced N3E variant of 3 nm, offering refined lithography for better yield and density without increasing power draw. Key advancements in manufacturing include TSMC's Integrated Fan-Out (InFO) packaging technology, which enables efficient stacking of CPU, GPU, and memory dies in M-series chips, reducing interconnect latency and form factor. This approach, combined with ongoing yield improvements—reaching over 70% for mature 3 nm runs—has lowered costs and supported Apple's high-volume production demands, with Apple securing up to 90% of TSMC's initial 3 nm capacity. Apple handles chip design in-house, while TSMC manages fabrication at its facilities in Taiwan, with Apple conducting final testing and validation internally to ensure quality. Global events, such as the 2020 semiconductor shortages triggered by the COVID-19 pandemic, disrupted this chain, delaying iPhone production and highlighting reliance on TSMC's Taiwan-based output. As of October 2025, the Apple M5 series utilizes TSMC's second-generation 3 nm process (N3P), announced alongside new MacBook Pro models, promising further efficiency gains through optimized backend-of-line metallization. Looking ahead, TSMC's 2 nm (N2) node is slated for mass production in 2026, with Apple reserving over half the capacity for upcoming Apple A19 and M6 chips to achieve up to 15% performance uplift and 30% power savings. Transistor density has advanced markedly, with TSMC's 3 nm processes enabling 130–230 million transistors per square millimeter, allowing Apple silicon to pack more compute resources while maintaining thermal limits.

A-series SoCs

Apple A4

The Apple A4 is Apple's first in-house designed system on a chip (SoC), marking the company's transition from third-party processors to custom silicon for its mobile devices. Announced on January 27, 2010, and first released on April 3, 2010, with the original , the A4 debuted in the on June 24, 2010. Fabricated by Samsung using a 45 nm CMOS process, the chip's die measures approximately 53 mm² and integrates key components for enhanced performance and power efficiency in portable computing. At its core, the A4 features a single-core processor clocked at 1.0 GHz, implementing a 32-bit ARMv7 instruction set architecture with support for integer and floating-point operations via an integrated FPU. The CPU includes 32 KB L1 instruction and data caches, along with a 512 KB L2 cache shared across the system. Paired with it is an Imagination Technologies GPU running at 200 MHz, delivering approximately 1.6 GFLOPS of peak floating-point performance, which enabled early mobile graphics capabilities such as hardware-accelerated video decoding and basic 3D rendering. The A4 supports up to 512 MB of LPDDR DRAM via a dual-channel 32-bit memory interface, though actual implementations varied by device—512 MB in the and 256 MB in the and iPod touch models. It includes an integrated display controller capable of driving high-resolution screens, including the 960x640 introduced with the , as well as H.264 video encoding/decoding up to 720p. The SoC also incorporates an image signal processor that facilitated processing for the 's front-facing VGA camera, enabling the debut of video calling. With a focus on low power consumption, the A4 contributed to extended battery life in its host devices, balancing performance for multitasking and media playback. The A4 powered several flagship devices, including the first-generation (Wi-Fi and 3G models), (GSM and CDMA variants), fourth-generation iPod touch, and second-generation . These implementations showcased the chip's versatility across tablets, smartphones, media players, and streaming hardware, setting the foundation for Apple's subsequent A-series evolution.

Apple A5

The Apple A5 is a system on a chip (SoC) designed by Apple and first introduced in the in March 2011, with further prominence in the October 2011 announcement of the . Manufactured by Samsung on a 45 nm process node, a later revision shifted to a 32 nm process for improved power efficiency. The A5 features a dual-core CPU clocked at 1.0 GHz, marking Apple's transition to dual-core processing in mobile devices while remaining a 32-bit architecture based on ARMv7. This configuration delivered up to twice the processing performance of the single-core A4 SoC. For graphics, the A5 integrates a dual-core GPU, which provided up to seven times the graphics performance compared to the A4's single-core PowerVR SGX535, enabling smoother rendering for games and applications. The SoC supports up to 1 GB of LPDDR2 RAM and includes an enhanced image signal processor (ISP) capable of handling 1080p video recording at 30 fps with features like video stabilization and improved low-light performance..html) It powered devices including the , , and fifth-generation . Despite the added dual cores and performance gains, the A5 achieved up to 30% better battery life in its 32 nm variant compared to the original A4 implementation, primarily through process optimizations that reduced power draw without sacrificing runtime.

Apple A5X

The Apple A5X is a system on a chip (SoC) designed by Apple and manufactured by Samsung on a 45 nm process node, released in March 2012 as a graphics-optimized variant of the A5 for tablet devices. It features the same dual-core ARM Cortex-A9 CPU architecture as the A5 but clocked at 1.0 GHz to handle the demands of higher-resolution displays. The key enhancement is its quad-core Imagination Technologies PowerVR SGX543MP4 graphics processing unit (GPU), which delivers approximately four times the graphics performance of the A5's dual-core SGX543MP2, enabling smooth rendering on Retina-class screens. The A5X integrates 1 GB of LPDDR2 RAM, doubling the capacity of the A5 to support multitasking and high-bandwidth graphics workloads. It is optimized for a 2048 × 1536 pixel display resolution at 264 pixels per inch, providing the pixel density necessary for Apple's Retina display technology without compromising frame rates in applications like gaming and video playback. The die size measures 163 mm², significantly larger than the A5's 121 mm² due to the expanded GPU cluster and additional memory interface. The A5X powered the third-generation iPad, also known as the "New iPad," launched on March 16, 2012. However, its higher transistor count and power-intensive GPU led to elevated thermal output during prolonged intensive use, with the device running about 10°F warmer than the iPad 2 and reaching surface temperatures up to 105°F in gaming scenarios. This increased power draw, estimated at around 2.6 W for the SoC under load compared to the A5's 1.6 W, prompted thermal management, including dynamic performance adjustments to prevent overheating, as confirmed by early user reports and Apple's statement that the device operated within specifications.

Apple A6

The Apple A6 is a 32-bit system on a chip (SoC) designed by Apple Inc. and manufactured by Samsung on a 32 nm high-κ metal gate (HKMG) process node. Announced on September 12, 2012, during the iPhone 5 unveiling, it represents Apple's inaugural shift to a fully custom CPU architecture, departing from licensed ARM designs used in prior A-series chips. The A6's die measures 96.71 mm², which is 22% smaller than the 45 nm A5, enabling enhanced power efficiency while delivering superior performance per watt. At its core, the A6 features a dual-core Swift CPU clocked at 1.3 GHz, implementing the ARMv7s instruction set with custom microarchitecture optimized for both speed and efficiency. Apple reported that the Swift cores provide up to twice the CPU performance of the A5's dual-core ARM Cortex-A9, achieved through hand-optimized layouts that prioritize high-performance computing tasks in mobile environments. This custom design allows for better integration with iOS, focusing on single-threaded workloads common in smartphone applications. The graphics subsystem consists of a triple-core Imagination Technologies PowerVR SGX543MP3 GPU, clocked at approximately 266 MHz, which Apple claimed doubles the graphics performance of the A5 for improved rendering in games and user interfaces. Complementing this, the A6 incorporates an integrated image signal processor (ISP) tailored for the iPhone 5's 8-megapixel rear camera, supporting features like faster autofocus and enhanced low-light imaging; this ISP enables 40% quicker image capture compared to the A5's implementation. The SoC also handles 1080p high-definition video recording at 30 frames per second, with hardware acceleration for encoding and playback. In the iPhone 5, the A6 is configured with 1 GB of LPDDR2-1066 RAM in a package-on-package stack, providing sufficient bandwidth for multitasking and media processing without excessive power draw. Exclusively powering the iPhone 5, the A6's architecture underscores Apple's emphasis on compact, efficient silicon that balances computational demands with extended battery life, setting the stage for subsequent custom core evolutions in the A-series.

Apple A6X

The Apple A6X is a system on a chip (SoC) designed by Apple as a performance-enhanced variant of the A6, specifically tailored for tablet devices. It was introduced on October 23, 2012, alongside the fourth-generation iPad, marking a rapid upgrade just seven months after the third-generation model's launch. The A6X powers the iPad (4th generation), enabling support for the device's 9.7-inch Retina display with a resolution of 2048×1536 at 264 pixels per inch, which demands higher graphical capabilities than previous iPads. The A6X features a dual-core Swift CPU clocked at 1.4 GHz, an increase from the 1.3 GHz of the A6, while retaining the same custom ARMv7s-based architecture optimized for efficiency. It integrates a quad-core PowerVR SGX554MP4 GPU, doubling the graphics cores compared to the tri-core SGX543MP3 in the A6, which Apple stated provides up to twice the graphics performance of the A6. Manufactured by Samsung using a 32 nm high-κ metal gate (HKMG) CMOS process, the A6X has a larger die area to accommodate the enhanced GPU while maintaining power efficiency suitable for the iPad's battery life. The SoC pairs with 1 GB of LPDDR2 RAM, configured in a 64-bit interface using two 4 Gb modules, to handle multitasking and high-resolution rendering. In terms of improvements over the prior A5X in the third-generation iPad, Apple claimed the A6X delivers up to twice the CPU performance and twice the graphics performance, achieved through the architectural advancements of the Swift cores, higher clock speeds, and the expanded GPU configuration. These enhancements were particularly impactful for graphics-intensive tasks, such as gaming and video playback on the Retina display, positioning the iPad 4 as a leader in mobile graphics at the time.

Apple A7

The Apple A7 is a 64-bit system on a chip (SoC) designed by Apple Inc. and announced on September 10, 2013, alongside the iPhone 5S smartphone. Fabricated by Samsung on a 28 nm high-k metal gate (HKMG) process node, the A7 represented Apple's transition to 64-bit computing in mobile devices. It powers the iPhone 5S, enabling advanced features like Touch ID through its integrated components. At the heart of the A7 is a dual-core CPU implementing Apple's custom Cyclone microarchitecture, clocked at 1.3 GHz and based on the ARMv8-A instruction set architecture (ISA). This made the A7 the first 64-bit ARM processor in a consumer device, supporting larger memory addressing and improved computational efficiency for mobile workloads. The Cyclone cores feature out-of-order execution and a six-wide instruction issue capability, drawing inspiration from desktop-class designs to enhance single-threaded performance. The A7 integrates a quad-core Imagination Technologies PowerVR G6430 graphics processing unit (GPU), clocked at approximately 450 MHz, which delivers over twice the graphics performance of the prior generation and introduces support for OpenGL ES 3.0. It pairs with 1 GB of LPDDR3 RAM and includes the first-generation M7 motion coprocessor, a dedicated low-power ARM core for processing data from accelerometers, gyroscopes, and other sensors without burdening the main CPU. A key milestone for the A7 was its optimization for iOS 7, which leveraged the 64-bit architecture to achieve roughly twice the CPU performance of the A6 SoC in single- and multi-threaded tasks, while maintaining similar power efficiency. This upgrade enabled smoother app launches, faster JavaScript execution, and enhanced machine learning capabilities for features like facial recognition.

Apple A8

The Apple A8 is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and announced on September 9, 2014, alongside the iPhone 6 and iPhone 6 Plus. Fabricated by TSMC on a 20 nm process, it represented Apple's first major shift away from Samsung's foundries for high-volume production of its mobile SoCs. The A8 features a second-generation 64-bit desktop-class architecture, delivering 25% higher CPU performance and 50% greater graphics performance compared to the A7, while consuming 50% less power for improved efficiency. At its core, the A8 integrates a dual-core Typhoon CPU clocked at 1.4 GHz, an evolution of the Cyclone architecture from the A7 with enhancements for better instruction throughput and branch prediction. The GPU is Apple's first fully custom design, a quad-core PowerVR Series 6XT GX6450, optimized for iOS workloads and supporting the new Metal graphics API for console-level 3D rendering. This custom GPU provides 50% faster graphics processing than the licensed PowerVR G6430 in the A7, enabling features like 1080p video at 60 fps and slow-motion capture at 240 fps. The A8 pairs 1 GB of LPDDR3 RAM with a second-generation M8 motion coprocessor, which adds a barometer for altitude sensing alongside accelerometer, gyroscope, and compass data processing to enhance fitness tracking and power efficiency. It also includes an integrated image signal processor for advanced camera features and a dedicated Secure Enclave for secure biometric authentication via Touch ID. Primarily deployed in the iPhone 6 (4.7-inch display) and iPhone 6 Plus (5.5-inch display), the A8 emphasized mid-generation refinements in power efficiency and graphics integration, setting the stage for Apple's deeper customization of silicon components.

Apple A8X

The Apple A8X is a 64-bit ARM-based system on a chip (SoC) designed by Apple and manufactured by TSMC on a 20 nm process node. It was introduced in October 2014 as the higher-binned variant of the A8 SoC, optimized for tablet workloads with enhanced multi-core capabilities and graphics performance. Unlike the dual-core A8 used in iPhones, the A8X features a triple-core CPU based on Apple's Cyclone architecture, clocked at 1.5 GHz, along with a doubled L2 cache size of 2 MB for improved efficiency in sustained tasks. The GPU in the A8X is an octa-core PowerVR GXA6850 (Series6XT-derived), representing a doubling of the graphics execution units compared to the quad-core configuration in the base A8, which enables approximately twice the graphics performance. This upgrade supports advanced rendering with Apple's Metal API, delivering 2.5 times the graphics performance of the prior-generation A7 SoC in the original iPad Air. The chip integrates 2 GB of LPDDR3 RAM on a 64-bit memory interface, providing sufficient bandwidth for multitasking and high-resolution media processing on a 2048×1536 Retina display. It also includes an M8 motion coprocessor for handling sensor data from the accelerometer, gyroscope, compass, and barometer. The A8X powers the iPad Air 2 (9.7-inch), Apple's thinnest tablet at launch, enabling 40% faster CPU performance over the A7 while maintaining power efficiency suitable for all-day battery life. This configuration marked an early step toward desktop-class computing on mobile devices, with the additional CPU core and expanded GPU facilitating smoother handling of productivity apps, gaming, and video editing on iOS 8.

Apple A9

The Apple A9 is a 64-bit ARM-based system on a chip (SoC) designed by Apple and announced on September 9, 2015, as part of the iPhone 6s and iPhone 6s Plus launch. It was fabricated using TSMC's 16 nm FinFET process for the majority of units, though some early production involved Samsung's 14 nm FinFET process, which resulted in slightly lower power efficiency in high-load scenarios compared to TSMC variants. The A9 marked Apple's transition to more advanced nodes, enabling improved performance and battery life over predecessors while maintaining a compact die size of approximately 104.5 mm² for TSMC versions. At its core, the A9 features a dual-core "Twister" CPU based on the ARMv8-A architecture, clocked at 1.85 GHz, delivering up to 70% faster CPU performance than the A8 in the prior iPhone 6. The GPU is a dual-core Imagination Technologies PowerVR GT7600 (Series 6XT), providing 90% faster graphics rendering than the A8's GPU, supporting advanced visual tasks in iOS applications. The SoC integrates 2 GB of LPDDR4 RAM at 1600 MHz for 25.6 GB/s bandwidth, a significant upgrade from the A8's LPDDR3, and includes the third-generation M9 motion coprocessor embedded directly on the die, which handles sensor data processing and enables always-on "Hey Siri" voice activation without draining the main CPU. The A9 powered mainstream iOS devices including the iPhone 6s, iPhone 6s Plus, first-generation iPhone SE, and fifth-generation iPad, all released between 2015 and 2016. Key features enabled by the A9 include support for 4K video recording at 30 fps via the device's 12-megapixel camera and Live Photos, which capture 12-megapixel stills with 1.5 seconds of motion before and after the shot, as well as 3D Touch for pressure-sensitive display interactions that provide contextual menus and shortcuts. These capabilities leveraged the A9's enhanced processing to deliver smoother multitasking and more responsive user experiences in iOS 9 and later versions.

Apple A9X

The Apple A9X is a 64-bit ARM-based system on a chip (SoC) developed by Apple Inc. and manufactured by TSMC on a 16 nm FinFET process. Announced on September 9, 2015, alongside the first-generation 12.9-inch iPad Pro, it powers Apple's initial entry into professional-grade tablets, emphasizing enhanced graphics and multitasking capabilities for creative and productivity applications. The A9X features a die size of 147 mm² and a 128-bit memory bus, allowing for greater bandwidth compared to the smartphone-oriented A9 SoC. At its core, the A9X employs a dual-core Twister CPU—Apple's first 64-bit ARMv8 design—clocked at up to 2.26 GHz in the 12.9-inch iPad Pro model, representing a notable increase over the A9's 1.85 GHz speed for improved single-threaded performance in demanding tasks. In the smaller 9.7-inch variant, the clock is slightly lower at 2.19 GHz to balance thermal constraints. The integrated GPU is a 12-core Imagination Technologies PowerVR Series 7XT, delivering approximately twice the graphics performance of the A9's six-core configuration, which enables smoother rendering in pro-level apps like video editing and graphic design software. This uplift supports features such as the first-generation Apple Pencil for precise input in creative workflows. The A9X supports up to 4 GB of LPDDR4 RAM in the 12.9-inch iPad Pro, facilitating robust multitasking with multiple high-resolution apps open simultaneously, while the 9.7-inch model uses 2 GB for a more compact form factor. It powers two devices: the first-generation 12.9-inch iPad Pro (released November 2015) and the 9.7-inch iPad Pro (released March 2016). Overall, the A9X's architecture prioritizes tablet-specific enhancements, providing up to 1.9 times the CPU performance of the prior A8X while excelling in GPU-intensive scenarios for professional use.

Apple A10 Fusion

The Apple A10 Fusion is a system on a chip (SoC) designed by Apple and introduced on September 7, 2016, as part of the iPhone 7 and iPhone 7 Plus announcement. This chip represented Apple's first implementation of a heterogeneous quad-core CPU architecture in its mobile processors, integrating two high-performance cores with two high-efficiency cores to optimize for both demanding tasks and power savings. Manufactured using TSMC's 16 nm FinFET process, the A10 Fusion contains approximately 3.3 billion transistors and measures about 125 mm² in die size. The CPU's high-performance "Vortex" cores deliver up to 40% greater overall performance than the dual-core Apple A9 in the iPhone 6s, with single-core speeds reaching up to 2.34 GHz and the ability to run up to twice as fast as the A8 in the iPhone 6 for intensive workloads. The paired efficiency cores operate at around 1.05 GHz and consume roughly one-fifth the power of the performance cores, allowing the SoC to dynamically switch between clusters for lighter tasks to extend battery life. Graphics processing is handled by a custom six-core implementation of the PowerVR Series7XT GPU, which provides up to 50% faster graphics performance than the A9 while using the same power, and up to three times the graphics speed of the iPhone 6's A8 at half the power consumption. An embedded fourth-generation M10 motion coprocessor manages sensor data for features like fitness tracking and augmented reality. Devices featuring the A10 Fusion include the iPhone 7 and iPhone 7 Plus (with 2 GB and 3 GB of LPDDR4 RAM, respectively), as well as the sixth-generation iPad released in 2018 (with 2 GB RAM). The chip's efficiency improvements contribute to up to two hours longer battery life in the iPhone 7 compared to the iPhone 6s, representing about 40% better energy efficiency overall, and enable new capabilities such as 4K video recording at 30 fps.

Apple A10X Fusion

The Apple A10X Fusion is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. as a high-performance variant of the A10 Fusion, targeted at professional tablet workloads. Manufactured by TSMC on a 10 nm FinFET process—the first such node used in an Apple chip—it features a die size of 96.4 mm² and approximately 3.3 billion transistors, representing a 33% reduction in area compared to the 16 nm A9X while adding more cores. Announced on June 5, 2017, and released on June 13, 2017, it powers the second-generation iPad Pro lineup, including the 10.5-inch and 12.9-inch models. The CPU adopts a hexa-core "Fusion" design with three high-performance Hurricane cores clocked at up to 2.38 GHz for demanding tasks and three energy-efficient Zephyr cores at around 1.3 GHz for lighter loads, scaling up from the quad-core (2+2) arrangement in the base A10 Fusion by adding an extra performance/efficiency pair and boosting peak clocks. Graphics are handled by a 12-core Imagination Technologies PowerVR Series 10XT GPU running at approximately 1.086 GHz, which doubles the core count of the A10 Fusion's six-core GPU to provide roughly twice the graphical throughput for tasks like 4K video editing and 3D rendering. Apple claimed the A10X delivers up to 30% faster CPU performance and 40% faster GPU performance than the A9X, enabling overall system speeds exceeding most PC laptops available at the time. Paired with 3 GB of LPDDR4X RAM in the 10.5-inch iPad Pro and 4 GB in the 12.9-inch variant, the A10X Fusion also includes an embedded M10 motion coprocessor for processing accelerometer, gyroscope, and compass data. It supports external display output via Lightning-to-HDMI or VGA adapters, allowing video mirroring or extension up to 1080p resolution at 60 Hz.

Apple A11 Bionic

The Apple A11 Bionic is a system on a chip (SoC) designed by Apple Inc. and manufactured by TSMC using a 10 nm FinFET process. It was announced on September 12, 2017, as the first chip in Apple's lineup to carry the "Bionic" branding, emphasizing its integrated neural processing capabilities for machine learning tasks. The A11 Bionic features 4.3 billion transistors and represents a significant advancement in heterogeneous computing, with all CPU cores able to operate simultaneously for improved multi-threaded performance. The CPU is a hexa-core design consisting of two high-performance Monsoon cores clocked at up to 2.39 GHz and four efficiency-oriented Mistral cores clocked at up to 1.42 GHz. Compared to the preceding A10 Fusion, the Monsoon cores deliver 25% higher single-threaded performance, while the Mistral cores provide 70% better efficiency, enabling up to 70% greater overall multi-threaded throughput via a second-generation performance controller. The GPU is an Apple-designed three-core unit that achieves 30% faster graphics rendering than the A10's GPU, with optimizations for the Metal 2 API to support advanced visual effects in games and augmented reality applications. A key innovation in the A11 Bionic is its debut Neural Engine, a dedicated dual-core neural processing unit capable of up to 600 billion operations per second, primarily enabling secure on-device machine learning for features like Face ID facial recognition. The chip also integrates an embedded M11 motion coprocessor for handling sensor data and supports 2 GB of RAM in the iPhone 8 and 8 Plus models, or 3 GB in the iPhone X. It powers the iPhone 8, iPhone 8 Plus, and iPhone X, incorporating hardware for wireless charging via the Qi standard and serving as the foundation for ARKit, Apple's augmented reality framework that leverages the Neural Engine and GPU for immersive experiences.

Apple A12 Bionic

The Apple A12 Bionic is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and announced on September 12, 2018, during the launch of the iPhone XS and iPhone XR series. Fabricated by TSMC on a 7 nm FinFET process—the first such node used in a smartphone SoC—it integrates 6.9 billion transistors, enabling significant improvements in power efficiency and performance density over its predecessor. This chip represents a key refinement in Apple's mobile silicon strategy, emphasizing balanced CPU and GPU enhancements for mainstream iPhone devices while introducing a more capable dedicated neural processing unit. The A12 Bionic's central processing unit (CPU) adopts a hexa-core configuration with two high-performance Vortex cores clocked at up to 2.49 GHz and four high-efficiency Tempest cores at up to 1.59 GHz, built on an ARMv8.3-A architecture. Relative to the A11 Bionic, the Vortex cores deliver up to 15% higher single-threaded performance, while the Tempest cores provide up to 50% better efficiency, resulting in overall CPU improvements of around 40% in single-core tasks and 50% in multi-core workloads for typical iPhone applications. These optimizations prioritize energy savings during everyday use, such as web browsing and app switching, without sacrificing peak performance for demanding tasks like video editing. The SoC also includes the integrated M12 motion coprocessor, which processes accelerometer, gyroscope, and compass data to enable precise motion tracking and fitness features with minimal battery impact. Complementing the CPU, the A12 Bionic incorporates a custom quad-core graphics processing unit (GPU) that achieves up to 50% faster rendering than the A11 Bionic's GPU, supporting advanced visual effects in games and augmented reality experiences on mainstream iPhones. For machine learning acceleration, an 8-core Neural Engine delivers up to 5 trillion operations per second (5 TOPS), enabling up to 9 times faster Core ML inference compared to the A11 Bionic while using one-tenth the power. Devices featuring the A12 Bionic, such as the iPhone XS, iPhone XS Max (with 4 GB LPDDR4X RAM), and iPhone XR (with 3 GB LPDDR4X RAM), leverage these capabilities for features like Smart HDR, which fuses multiple exposures using the Neural Engine and image signal processor to produce photos with enhanced dynamic range and detail in highlights and shadows. The chip also powers processing for Liquid Retina displays on the iPhone XR, optimizing color accuracy and responsiveness through efficient on-device computations.

Apple A12X Bionic

The Apple A12X Bionic is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc., introduced as the high-end processor for the third-generation iPad Pro models. Announced on October 30, 2018, and released on November 7, 2018, it was fabricated using TSMC's 7 nm process, marking the first use of this node in Apple's tablet lineup for improved power efficiency and transistor density. The A12X Bionic powers professional workflows on the iPad Pro (11-inch and 12.9-inch, third generation), enabling features like Face ID and advanced AR experiences through its integrated components. The CPU features an octa-core configuration with four high-performance Vortex cores clocked at up to 2.49 GHz and four high-efficiency Tempest cores at up to 1.59 GHz, doubling the number of high-performance cores compared to the hexa-core Apple A12 Bionic in the iPhone XS. This design provides up to 35% faster single-core performance and 90% faster multi-core performance relative to the prior A10X Fusion, supporting intensive multitasking such as running multiple pro apps simultaneously. The GPU is a seven-core unit (with some binned units featuring eight cores), delivering approximately twice the graphics performance of the A12 Bionic's four-core GPU, suitable for console-level gaming and AR rendering. An eight-core Neural Engine, identical to that in the A12 Bionic, performs up to 5 trillion operations per second for machine learning tasks like real-time photo editing and object detection. The A12X Bionic supports 4 GB of RAM in the 11-inch iPad Pro and 4–6 GB in the 12.9-inch model (with 6 GB exclusive to the 1 TB storage variant), paired with storage options from 64 GB to 1 TB.-9386.php) It includes hardware-accelerated video encoding and decoding for HEVC, enabling external display output up to 5K resolution at 60 Hz or 4K at 60 Hz via USB-C, which facilitates professional setups like connecting to high-resolution monitors for content creation. In benchmarks and real-world use, the A12X Bionic approaches entry-level laptop performance, outperforming 92% of portable PCs sold in the US from June 2017 to June 2018 according to NPD data, particularly in video editing tasks with apps like LumaFusion where it handles 4K timelines efficiently without thermal throttling in short bursts. This capability positions it as a versatile chip for creative professionals, bridging mobile and desktop computing paradigms.

Apple A12Z Bionic

The Apple A12Z Bionic is a 64-bit ARM-based system on a chip (SoC) developed by Apple Inc. and released in March 2020 alongside the fourth-generation iPad Pro models in 11-inch and 12.9-inch sizes. It is manufactured using TSMC's 7 nm process node. The chip powers these devices exclusively and integrates support for the LiDAR scanner introduced in the lineup, enabling advanced augmented reality (AR) features such as precise depth sensing and scene understanding. The A12Z Bionic employs an eight-core CPU configuration identical to that of the A12X Bionic, with four high-performance Vortex cores and four high-efficiency Tempest cores clocked up to 2.49 GHz. Its graphics subsystem consists of an eight-core GPU, an upgrade from the seven-core GPU in the A12X, which contributes to improved graphics performance. Additionally, it includes an eight-core Neural Engine capable of performing up to 5 trillion operations per second (TOPS) for machine learning tasks. The SoC is paired with 6 GB of RAM in the iPad Pro devices it powers. An enhanced thermal architecture and performance controllers allow for better sustained operation under load, particularly benefiting AR and virtual reality (VR) applications by reducing throttling and maintaining higher frame rates during extended use. This combination, along with the LiDAR integration, supports advancements in ARKit, including faster object placement, improved motion tracking, and more accurate environmental occlusion.

Apple A13 Bionic

The Apple A13 Bionic is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and manufactured by TSMC using its second-generation 7 nm (N7P) process. It was announced on September 10, 2019, alongside the iPhone 11 lineup, marking the third generation of Apple's Bionic processors with a focus on enhanced machine learning capabilities and power efficiency. The chip integrates 8.5 billion transistors and powers key features in mobile devices through its hexa-core CPU, quad-core GPU, and dedicated neural processing unit. The CPU consists of two high-performance Lightning cores clocked at up to 2.65 GHz and four high-efficiency Thunder cores, delivering up to 20% faster performance compared to the A12 Bionic while consuming 30% less power for sustained tasks. The integrated quad-core GPU provides 20% faster graphics rendering than its predecessor, enabling smoother handling of demanding visual workloads with 40% improved energy efficiency. Supporting these is an 8-core Neural Engine capable of 5 trillion operations per second (5 TOPS), optimized for real-time machine learning tasks such as photo and video analysis, representing a 20% speed increase over the A12's Neural Engine with 15% lower power usage. The SoC also includes 4 GB of LPDDR4X RAM and the M13 motion coprocessor for precise sensor data processing. The A13 Bionic debuted in the iPhone 11, iPhone 11 Pro, and iPhone 11 Pro Max, enabling advanced photography features like Night mode for low-light capture on both wide and ultra-wide cameras, as well as 4K video recording at 60 frames per second with extended dynamic range. These capabilities leverage the chip's efficient architecture to support computational photography and video processing without compromising battery life, setting new standards for mobile performance at the time.

Apple A14 Bionic

The Apple A14 Bionic is a system on a chip (SoC) designed by Apple, announced on September 15, 2020, as part of the fourth-generation iPad Air reveal. It marks the first A-series processor manufactured using TSMC's 5 nm process node, enabling a transistor count of 11.8 billion and improved power efficiency compared to prior generations. The SoC integrates a hexa-core CPU, quad-core GPU, 16-core Neural Engine, and image signal processor (ISP), supporting advanced machine learning and computational tasks while maintaining compatibility with iOS and iPadOS devices. The CPU features a heterogeneous design with two high-performance Firestorm cores clocked up to 3.0 GHz and four high-efficiency Icestorm cores, delivering up to 40% faster performance than the A12 Bionic in CPU-intensive workloads. This architecture enhances single- and multi-threaded processing for applications like video editing and augmented reality. The integrated quad-core GPU supports mesh shading via Metal API, enabling more efficient geometry processing for gaming and graphics rendering, with up to 30% improved graphics performance over the prior generation. The 16-core Neural Engine provides up to 11 trillion operations per second (TOPS), representing an 80% increase in machine learning capabilities for tasks such as object detection and natural language processing. A 16-core ISP advances computational photography, supporting features like Deep Fusion and Night mode across multiple lenses. Devices equipped with the A14 Bionic typically pair it with 4 GB or 6 GB of LPDDR4X RAM, depending on the model—4 GB in the iPhone 12, iPhone 12 mini, and fourth-generation iPad Air, and 6 GB in the iPhone 12 Pro and iPhone 12 Pro Max. Key features include support for 5G connectivity through integrated modem compatibility in host devices and Dolby Vision HDR video recording at up to 30 fps, enhancing media capture and playback. The A14 Bionic powers the iPhone 12 series (including mini, standard, Pro, and Pro Max models, released in October 2020) and the fourth-generation iPad Air (released in October 2020).

Apple A15 Bionic

The Apple A15 Bionic is a system on a chip (SoC) designed by Apple Inc., announced on September 14, 2021, during the iPhone 13 event. It is fabricated by TSMC using a second-generation 5 nm process node (N5P), enabling higher transistor density with 15 billion transistors compared to the A14 Bionic's 11.8 billion. The A15 shares a unified architecture with the M1 chip used in Macs, facilitating consistent performance optimizations across Apple's device lineup. The CPU features a hexa-core configuration with two high-performance Avalanche cores clocked at up to 3.23 GHz and four high-efficiency Blizzard cores at approximately 2.0 GHz, providing incremental improvements in speed and power efficiency over the A14 Bionic. The GPU varies by device variant: a 4-core version in base models and a 5-core version in Pro models, delivering up to 50% faster graphics performance than competing smartphone chips at the time, which supports enhanced rendering for Pro features like the 120 Hz ProMotion adaptive refresh rate display in iPhone 13 Pro and Pro Max. A key enhancement is the 16-core Neural Engine, capable of 15.8 trillion operations per second (TOPS), representing up to a 50% increase in machine learning performance over the A14 Bionic's 11 TOPS Neural Engine. This boost enables advanced on-device AI features, such as Cinematic mode for video recording with depth-of-field effects and improved computational photography. The SoC also includes a fifth-generation image signal processor (ISP), 4 GB or 6 GB of LPDDR4X RAM depending on the device (4 GB in base iPhone 13 models and iPad mini, 6 GB in Pro variants), and supports hardware-accelerated AV1 decode for efficient video streaming. The A15 Bionic powers the iPhone 13 and iPhone 13 mini (4-core GPU, 4 GB RAM), iPhone 13 Pro and iPhone 13 Pro Max (5-core GPU, 6 GB RAM), the entire iPhone 14 series (with 6 GB RAM across models), and the sixth-generation iPad mini. These implementations emphasize Pro-level enhancements, such as ProMotion's smooth scrolling and gesture navigation, which leverage the GPU's increased capabilities for fluid 120 Hz visuals without compromising battery life.

Apple A16 Bionic

The Apple A16 Bionic is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and fabricated by TSMC on its enhanced 4 nm (N4P) process node, which enables improved power efficiency and transistor density compared to the 5 nm node used in its predecessor. Announced on September 7, 2022, during the unveiling of the iPhone 14 Pro and iPhone 14 Pro Max, the A16 Bionic integrates approximately 16 billion transistors and represents Apple's first mobile SoC to leverage this second-generation 4 nm technology for better performance per watt. The CPU subsystem features a hexa-core configuration with two high-performance "Everest" cores clocked at up to 3.46 GHz for demanding tasks and four high-efficiency "Sawtooth" cores operating at up to 2.02 GHz for lighter workloads, delivering a modest single-core performance uplift of around 10-15% over the A15 Bionic while maintaining similar multi-core capabilities and enhanced energy efficiency. The integrated 5-core GPU introduces hardware-accelerated ray tracing to iPhones, providing up to 2x the ray tracing performance compared to the A15 Bionic's GPU in supported workloads, along with 50% greater memory bandwidth to handle graphics-intensive applications like gaming and augmented reality. Complementing these is a 16-core Neural Engine capable of 17 trillion operations per second (TOPS), optimized for machine learning tasks such as computational photography and on-device processing. The A16 Bionic pairs with 6 GB of LPDDR5 RAM and includes the fourth-generation M16 motion coprocessor, which processes sensor data for features like precise motion tracking and fitness metrics. It powers the iPhone 14 Pro and iPhone 14 Pro Max, enabling advanced capabilities such as crash detection—which uses the Neural Engine and motion coprocessor to analyze accelerometer and gyroscope data for rapid emergency response—and support for adaptive refresh rates up to 120 Hz on the ProMotion display for smoother visuals and battery optimization. Overall, the chip emphasizes balanced efficiency, sustaining all-day battery life under heavy use while advancing pro-level photography and interactive experiences like the Dynamic Island interface.

Apple A17 Pro

The Apple A17 Pro is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and manufactured by TSMC using its second-generation 3 nm process (N3B). It was announced on September 12, 2023, as the flagship processor for the iPhone 15 Pro and iPhone 15 Pro Max, and later for the iPad mini (7th generation), marking Apple's first use of a 3 nm node in a mobile SoC to deliver improved power efficiency and performance density. The chip integrates approximately 19 billion transistors and supports advanced features tailored for mobile gaming and AI workloads. The A17 Pro features a hexa-core CPU configuration with two high-performance "Everest" cores clocked up to 3.78 GHz and four high-efficiency "Sawtooth" cores at up to 2.11 GHz, enabling up to 10% faster single-core performance compared to the A16 Bionic through microarchitectural improvements. Its 6-core GPU represents Apple's second-generation design, offering up to 20% faster graphics performance than the prior generation, with dedicated hardware acceleration for ray tracing (4x faster than software-based methods) and mesh shading to enhance rendering efficiency in real-time applications like augmented reality and complex games. The integrated 16-core Neural Engine delivers up to twice the performance of the A16's, capable of 35 trillion operations per second (TOPS), supporting on-device machine learning tasks for features like enhanced autocorrect and spatial audio processing. Paired with 8 GB of LPDDR5X RAM, the A17 Pro enables seamless multitasking and powers console-quality gaming experiences, such as native ports of Resident Evil Village, Resident Evil 4, and Assassin's Creed Mirage, which leverage its ray tracing capabilities for realistic lighting and shadows. It includes a dedicated hardware decoder for the AV1 video codec, improving streaming efficiency for high-resolution content, and a new USB controller supporting USB 3.2 Gen 2 speeds up to 10 Gbps—20 times faster than USB 2—for rapid data transfers and 4K HDR video output at 60 fps. These advancements position the A17 Pro as a pivotal step in bringing desktop-level graphics and AI to smartphones.

Apple A18

The Apple A18 is a system on a chip (SoC) designed by Apple Inc. and manufactured by TSMC using its second-generation 3 nm process (N3E). Announced on September 9, 2024, alongside the iPhone 16 and iPhone 16 Plus, the A18 features a 6-core CPU consisting of 2 high-performance cores clocked up to approximately 3.8 GHz and 4 high-efficiency cores. This configuration provides a balance of performance and power efficiency, enabling up to 30% faster CPU performance compared to the A16 Bionic in the iPhone 15. The A18 integrates a 5-core GPU that supports hardware-accelerated ray tracing, delivering up to 40% better graphics performance than the A16 GPU and supporting advanced mobile gaming with features like dynamic lighting and shadows. Its 16-core Neural Engine is capable of 35 trillion operations per second (TOPS), optimized for on-device machine learning tasks including Apple Intelligence features such as enhanced photo editing and natural language processing. The chip pairs with 8 GB of LPDDR5X RAM and includes an updated image signal processor that improves computational photography, enabling features like next-generation Photographic Styles for real-time color and tone adjustments. Exclusive to the iPhone 16 and iPhone 16 Plus, the A18 contributes to improved battery life through enhanced power efficiency from the 3 nm process, offering up to 22 hours of video playback on the iPhone 16 model. It also integrates a 5G modem, Wi-Fi 6E support, and Bluetooth 5.3 for connectivity.

Apple A18 Pro

The Apple A18 Pro is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc., announced on September 9, 2024, as part of the iPhone 16 series launch. It is fabricated using TSMC's second-generation 3 nm process (N3E), enabling higher transistor density and improved power efficiency compared to its predecessor. The chip powers the iPhone 16 Pro and iPhone 16 Pro Max, emphasizing advancements in AI processing, graphics rendering, and camera capabilities to support features like Apple Intelligence. With 8 GB of LPDDR5X RAM, it facilitates on-device machine learning tasks and multitasking. The A18 Pro features a hexa-core CPU configuration, consisting of two high-performance cores and four high-efficiency cores, delivering up to 15% faster single-core and 20% faster multi-core performance than the A17 Pro. The performance cores operate at approximately 3.8 GHz, providing a balance of speed and energy efficiency for demanding applications. Its 6-core GPU offers 20% greater graphics performance over the A17 Pro's GPU, incorporating second-generation hardware-accelerated ray tracing for up to twice the speed in rendering realistic lighting and shadows, along with enhanced dynamic caching to optimize memory allocation for varying workloads. These improvements enable smoother gameplay in titles like Death Stranding and support advanced visual effects. At the heart of the A18 Pro's AI capabilities is its 16-core Neural Engine, capable of 35 trillion operations per second (TOPS), a 17% increase over the A17 Pro to handle on-device processing for Apple Intelligence features. This enables enhanced Siri interactions, such as contextual understanding across apps and on-screen awareness, all performed privately without cloud dependency. The chip also supports a 48 MP Fusion camera system with a quad-pixel sensor for improved low-light performance and 2x faster readout speeds, facilitating features like spatial photos and videos for immersive viewing on Apple Vision Pro.

Apple A19

The Apple A19 is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and manufactured by TSMC using its enhanced 3 nm process (N3P). Announced on September 9, 2025, alongside the iPhone 17 and iPhone 17 Air, it powers entry-level and mid-range iPhone models with improved efficiency and AI capabilities. The A19 integrates a 6-core CPU with 2 high-performance cores clocked up to 4.26 GHz and 4 high-efficiency cores at up to 2.6 GHz, along with an 8 MB L2 cache. It features a 5-core GPU supporting hardware-accelerated ray tracing and dynamic caching, providing enhanced graphics performance for gaming and AR applications. The 16-core Neural Engine is capable of approximately 35 TOPS for on-device machine learning, supporting Apple Intelligence features. The A19 pairs with 8 GB of LPDDR5X-8533 RAM, offering 68.3 GB/s bandwidth. The A19 powers the iPhone 17 and iPhone 17 Air (released October 2025), enabling features like advanced computational photography and extended battery life through process optimizations.

Apple A19 Pro

The Apple A19 Pro is a 64-bit ARM-based system on a chip (SoC) designed by Apple Inc. and manufactured by TSMC using its enhanced 3 nm process (N3P). Announced on September 9, 2025, it powers the iPhone 17 Pro and iPhone 17 Pro Max, focusing on pro-level performance for AI, graphics, and professional workflows. The A19 Pro features a 6-core CPU with 2 high-performance cores clocked up to 4.26 GHz and 4 high-efficiency cores at up to 2.6 GHz, with a larger 16 MB L2 cache for improved sustained performance. Its 6-core GPU includes second-generation ray tracing and Neural Accelerators, delivering up to 20% better graphics than the A18 Pro. The 16-core Neural Engine supports up to 35 TOPS or higher for advanced Apple Intelligence tasks. It pairs with 12 GB of LPDDR5X RAM at 9600 MT/s, providing 75.8 GB/s bandwidth, and includes the C1X cellular modem. The A19 Pro powers the iPhone 17 Pro and iPhone 17 Pro Max (released October 2025), supporting features like 120 Hz ProMotion displays, spatial video capture, and USB 3 speeds up to 10 Gbps.

Comparison of A-series processors

The A-series processors form the backbone of Apple's mobile devices, evolving from basic single-core designs to sophisticated six-core systems with integrated AI accelerators. This section aggregates key specifications across the lineup to facilitate comparison, highlighting advancements in architecture, manufacturing, and capabilities. Data is drawn from official Apple announcements and technical breakdowns by reputable analysts.
ChipProcess NodeCPU (cores/clock)GPU (cores/type)NPU (TOPS)RAM SupportRelease YearKey Devices
A445 nm1 core @ 1.0 GHz (ARM Cortex-A8)PowerVR SGX535 (1 core)-256-512 MB LPDDR2010iPhone 4, iPad (1st gen)
A545 nm2 cores @ 1.0 GHz (ARM Cortex-A9)PowerVR SGX543MP2 (2 cores)-512 MB-1 GB LPDDR22011iPhone 4S, iPad 2
A5X45 nm2 cores @ 1.0 GHz (ARM Cortex-A9)PowerVR SGX543MP4 (4 cores)-1 GB LPDDR22012iPad (3rd gen)
A632 nm2 cores @ 1.3 GHz (Swift)PowerVR SGX543MP3 (3 cores)-1 GB LPDDR22012iPhone 5
A6X32 nm2 cores @ 1.4 GHz (Swift)PowerVR SGX554MP4 (4 cores)-1 GB LPDDR22012iPad (4th gen)
A728 nm2 cores @ 1.3 GHz (Cyclone)PowerVR G6430 (4 cores)-1 GB LPDDR32013iPhone 5S, iPad mini 2
A820 nm2 cores @ 1.4 GHz (Typhoon)PowerVR GX6450 (4 cores)-1-2 GB LPDDR32014iPhone 6, iPad mini 3
A8X20 nm3 cores @ 1.5 GHz (Typhoon)PowerVR GXA6850 (8 cores)-2 GB LPDDR32014iPad Air 2
A916/14 nm2 cores @ 1.85 GHz (Twister)PowerVR GT7600 (6 cores)-2 GB LPDDR42015iPhone 6s, iPhone SE (1st)
A9X16 nm2 cores @ 2.26 GHz (Twister)PowerVR Series 7XT (12 cores)-2-4 GB LPDDR42015iPad Pro (1st gen)
A10 Fusion16 nm4 cores @ 2.34 GHz (2P+2E, Hurricane/Mistral)PowerVR Series7XT (6 cores)-2-3 GB LPDDR42016iPhone 7, iPad (6th gen)
A10X Fusion10 nm6 cores @ 2.38 GHz (3P+3E, Hurricane/Zephyr)PowerVR Series 10XT (12 cores)-3-4 GB LPDDR4X2017iPad Pro (2nd gen)
A11 Bionic10 nm6 cores @ 2.39 GHz (2P+4E, Monsoon/Mistral)Apple (3 cores)0.62-3 GB LPDDR4X2017iPhone 8, iPhone X
A12 Bionic7 nm6 cores @ 2.49 GHz (2P+4E, Vortex/Tempest)Apple (4 cores)53-4 GB LPDDR4X2018iPhone XS, iPhone XR
A12X Bionic7 nm8 cores @ 2.49 GHz (4P+4E, Vortex/Tempest)Apple (7-8 cores)54-6 GB LPDDR4X2018iPad Pro (3rd gen)
A12Z Bionic7 nm8 cores @ 2.49 GHz (4P+4E, Vortex/Tempest)Apple (8 cores)56 GB LPDDR4X2020iPad Pro (4th gen)
A13 Bionic7 nm6 cores @ 2.65 GHz (2P+4E, Lightning/Thunder)Apple (4 cores)54 GB LPDDR4X2019iPhone 11
A14 Bionic5 nm6 cores @ 3.0 GHz (2P+4E, Firestorm/Icestorm)Apple (4 cores)114-6 GB LPDDR4X2020iPhone 12, iPad Air (4th)
A15 Bionic5 nm6 cores @ 3.23 GHz (2P+4E, Avalanche/Blizzard)Apple (4-5 cores)15.84-6 GB LPDDR52021iPhone 13, iPhone SE (3rd)
A16 Bionic4 nm6 cores @ 3.46 GHz (2P+4E, Everest/Sawtooth)Apple (5 cores)176 GB LPDDR52022iPhone 14 Pro, iPhone 15
A17 Pro3 nm6 cores @ 3.78 GHz (2P+4E, Everest/Sawtooth)Apple (6 cores)358 GB LPDDR5X2023iPhone 15 Pro
A183 nm6 cores @ 3.8 GHz (2P+4E)Apple (5 cores)358 GB LPDDR5X2024iPhone 16, iPhone 16 Plus
A18 Pro3 nm6 cores @ 3.8 GHz (2P+4E)Apple (6 cores)358 GB LPDDR5X2024iPhone 16 Pro, iPhone 16 Pro Max
A193 nm6 cores @ 4.26 GHz (2P+4E)Apple (5 cores)~358 GB LPDDR5X2025iPhone 17, iPhone 17 Air
A19 Pro3 nm6 cores @ 4.26 GHz (2P+4E)Apple (6 cores)~3512 GB LPDDR5X2025iPhone 17 Pro, iPhone 17 Pro Max
The table above captures representative configurations; clock speeds refer to peak performance core frequencies, and GPU types shifted from licensed PowerVR designs to Apple's custom architectures starting with the A11. Transistor counts exemplify scaling, rising from approximately 0.5 billion in the A4 to over 20 billion in recent chips, enabling denser integration and higher efficiency. GPU peak performance has advanced to around 1.5 TFLOPS in the A18 Pro, a metric that underscores graphical capabilities for tasks like ray tracing, while power efficiency has improved markedly, with recent chips delivering up to 30% better performance per watt compared to predecessors through advanced nodes and microarchitectural tweaks. Over the generations, the A-series shifted from 32-bit ARMv7 architecture in the A4 through A6 to 64-bit ARMv8 with the A7, enabling more complex software and better 64-bit app support. CPU core counts progressed from single-core to dual-core (A5-A9), quad-core (A10), and stabilized at six-core big.LITTLE designs from the A11 onward, balancing high-performance and efficiency cores for sustained battery life in mobile use. The introduction of the Neural Processing Unit (NPU) with the A11 marked a pivotal trend, growing from 0.6 TOPS to 35 TOPS in the A17 Pro and later, facilitating on-device machine learning for features like Face ID and computational photography without cloud dependency. Variants like the X and Z designations denote higher-performance bins optimized for iPads, featuring additional CPU/GPU cores and greater RAM capacity—for instance, the A12X adds two extra CPU cores and three more GPU cores over the base A12 for multitasking in larger form factors. The "Fusion" suffix appeared with the A10 to highlight heterogeneous computing, while "Bionic" from the A11 emphasized bio-inspired neural capabilities. Process node shrinks—from 45 nm to 3 nm—have driven exponential performance scaling, roughly doubling compute throughput every two years while reducing power draw, as visualized in a trend line where transistor density increases align with Dennard scaling principles until recent FinFET and GAA transitions.

M-series SoCs

Apple M1

The Apple M1 is a system on a chip (SoC) designed by Apple Inc., marking the company's transition to its own ARM-based processors for Mac computers and unifying the architecture previously used in iOS devices with macOS systems. Announced on November 10, 2020, the M1 was the first Apple silicon chip for personal computers, fabricated on TSMC's 5 nm process with 16 billion transistors, enabling significant improvements in performance and power efficiency. This integration of CPU, GPU, Neural Engine, and other components on a single die allowed for a unified memory architecture, facilitating seamless data sharing across processing units and reducing latency compared to traditional discrete designs. The M1 features an 8-core CPU comprising four high-performance Firestorm cores clocked up to 3.2 GHz and four energy-efficient Icestorm cores, providing a balance of speed for demanding tasks and power savings for lighter workloads. Its integrated GPU is configured with either 7 or 8 cores, depending on the device model, supporting hardware-accelerated graphics and machine learning workloads. Additionally, the 16-core Neural Engine delivers 11 trillion operations per second (TOPS), accelerating on-device AI tasks such as image processing and voice recognition. Memory options include 8 GB or 16 GB of unified LPDDR4X RAM, shared across the CPU, GPU, and Neural Engine, with support for Thunderbolt 4 connectivity via USB4 ports for high-speed data transfer and peripheral expansion. The M1 powered the initial Apple silicon Macs and select iPads. The following table lists the products featuring the M1 and their release months:
ProductRelease Date
MacBook AirNovember 2020
13-inch MacBook ProNovember 2020
Mac miniNovember 2020
24-inch iMacApril 2021
iPad Pro (12.9-inch, 5th gen)May 2021
iPad Air (5th gen)March 2022
In performance evaluations, the M1 achieved up to 3.5 times faster CPU speeds than the Intel Core i7-based MacBook Air it replaced in certain productivity applications, while maintaining exceptional battery life of up to 20 hours. To support the shift from Intel x86 architecture, Apple introduced Rosetta 2, a translation layer that enables x86 applications to run on ARM with near-native efficiency, often matching or exceeding performance on older Intel Macs.

Apple M1 Pro

The Apple M1 Pro is a system on a chip (SoC) developed by Apple Inc. as an enhanced variant of the M1, targeted at creative professionals requiring greater performance and expanded input/output capabilities for demanding workflows such as video editing and 3D rendering. Announced on October 18, 2021, it was the first in Apple's M-series lineup to introduce dedicated hardware acceleration for professional media formats and support for higher-bandwidth peripherals. Fabricated by TSMC using a 5 nm process node, the M1 Pro integrates 33.7 billion transistors, more than double the count in the base M1, enabling significant scaling in compute resources while maintaining power efficiency. The CPU configuration centers on a 10-core design with eight high-performance "Firestorm" cores and two high-efficiency "Icestorm" cores, where the performance cores can reach clock speeds up to 3.2 GHz. The GPU scales to 14 or 16 cores, providing up to twice the graphics performance of the M1's 8-core GPU through architectural improvements like enhanced texture units and ray-tracing hardware. Additionally, the 16-core Neural Engine delivers 11 trillion operations per second (TOPS) for machine learning tasks, matching the M1's capability but integrated into a more expansive system. Memory support includes 16 GB or 32 GB of unified LPDDR5 RAM with 200 GB/s bandwidth, allowing seamless data sharing between CPU, GPU, and Neural Engine to accelerate professional applications. Key features emphasize media production, including a dedicated media engine for hardware-accelerated ProRes encode and decode, enabling efficient handling of multiple 4K or 8K ProRes streams without taxing the CPU. For connectivity, the M1 Pro facilitates expanded I/O in host devices, such as HDMI 2.0 ports for direct 4K output and SDXC card slots for high-speed media ingestion, alongside up to three Thunderbolt 4 controllers for peripherals. It also supports driving up to two external 6K displays at 60 Hz via Thunderbolt, plus the built-in display, surpassing the M1's single external display limitation. The M1 Pro powers the 14-inch and 16-inch MacBook Pro models released on October 26, 2021, where its balanced core scaling and I/O enhancements provide a cost-effective step up from the base M1 for pros, without the extreme GPU emphasis of higher-end variants.
DeviceRelease Date
MacBook Pro (14-inch, 2021)October 26, 2021
MacBook Pro (16-inch, 2021)October 26, 2021

Apple M1 Max

The Apple M1 Max is a system on a chip (SoC) developed by Apple Inc. and announced on October 18, 2021, as part of the company's transition to its custom ARM-based silicon for Mac computers. Fabricated on TSMC's 5-nanometer process, it contains 57 billion transistors, making it the largest die Apple had produced at the time, optimized for high-end graphics and media production workflows. The M1 Max features a 10-core CPU configuration identical to that of the M1 Pro, consisting of eight high-performance cores and two high-efficiency cores, enabling robust processing for demanding applications while balancing power efficiency. Its standout component is the GPU, available in 24-core or 32-core variants, which delivers up to four times the graphics performance of the base M1's eight-core GPU, tailored for tasks like video editing, 3D rendering, and real-time effects. Complementing these is a 16-core Neural Engine capable of 11 trillion operations per second (TOPS), supporting on-device machine learning for features such as image processing and computational photography. Memory support includes up to 64 GB of unified LPDDR5 RAM with 400 GB/s bandwidth—twice that of the M1 Pro—allowing seamless data sharing between CPU, GPU, and Neural Engine for accelerated workflows. A dedicated media engine handles hardware-accelerated encoding and decoding for formats like H.264, HEVC, ProRes, and ProRes RAW, while the display engine supports up to three external 6K displays at 60 Hz alongside the built-in screen. In performance benchmarks, the M1 Max's 32-core GPU rivals or exceeds discrete graphics solutions like the AMD Radeon Pro 5500M found in prior-generation MacBook Pros, particularly in professional applications such as Final Cut Pro, where it achieves up to 13 times faster rendering for complex 8K timelines compared to Intel-based predecessors. The M1 Max powers the following Apple products:
ProductRelease Date
14-inch MacBook ProOctober 26, 2021
16-inch MacBook ProOctober 26, 2021
Mac StudioMarch 18, 2022
positioning it as a GPU-centric upgrade over the more CPU-balanced M1 Pro for users in creative industries requiring intensive visual computing. As a single-die design, it emphasizes integrated efficiency without relying on multi-chip interconnects, delivering sustained performance in battery-powered laptops.

Apple M1 Ultra

The Apple M1 Ultra is a system on a chip (SoC) developed by Apple Inc. and announced on March 8, 2022, as the pinnacle of the first-generation M-series processors for high-end desktop applications. Fabricated on TSMC's 5 nm process, it contains 114 billion transistors and employs a dual-die configuration by connecting two M1 Max chips via Apple's proprietary UltraFusion interconnect, enabling seamless scaling of resources for demanding workloads like video editing and 3D rendering. This design doubles the computational capabilities of the single-die M1 Max while maintaining efficiency, with the UltraFusion interposer facilitating over 2.5 TB/s of die-to-die bandwidth across more than 10,000 connections to minimize latency. The M1 Ultra features a 20-core CPU configuration, comprising 16 high-performance cores and 4 high-efficiency cores, which provides up to 90% more multi-core CPU performance compared to the M1 Max for tasks such as compiling code or simulating complex models. Its GPU scales to 48 or 64 cores, delivering graphics performance that surpasses high-end discrete PC GPUs in rasterization and ray tracing while consuming up to 200 watts less power, making it suitable for professional graphics workflows. The integrated 32-core Neural Engine achieves 22 trillion operations per second (TOPS), accelerating machine learning inference for applications like image recognition and natural language processing. Memory support includes 64 GB or 128 GB of unified LPDDR5, with 800 GB/s of bandwidth to ensure low-latency access across CPU, GPU, and Neural Engine for unified memory architecture benefits. The chip powers the Mac Studio desktop released in March 2022, where configurations can include an optional 10 Gigabit Ethernet port for high-speed networking in professional environments.
ProductRelease Date
Mac StudioMarch 18, 2022
This combination positions the M1 Ultra as a compact, power-efficient solution for pro users requiring extreme performance without the thermal overhead of traditional multi-chip setups.

Apple M2

The Apple M2 is a system on a chip (SoC) designed by Apple Inc. as the second generation of its ARM-based silicon for Mac computers, announced on June 6, 2022. Fabricated using TSMC's enhanced 5 nm process (second-generation 5 nm), it contains 20 billion transistors, a 25% increase over the M1's 16 billion. This design refines the architecture for improved efficiency and performance in entry-level Mac devices, laying groundwork for future scaling toward smaller nodes while maintaining compatibility with the unified memory architecture. The M2 features an 8-core CPU with 4 high-performance cores and 4 high-efficiency cores, capable of reaching up to 3.5 GHz on the performance cores. Apple reports up to 18% faster CPU performance compared to the M1 in multithreaded tasks. The integrated GPU offers 8 or 10 cores (depending on configuration), with a larger cache and 100 GB/s memory bandwidth—50% higher than the M1—enabling up to 35% better graphics performance at maximum power. Additionally, the 16-core Neural Engine delivers 15.8 trillion operations per second (TOPS), a 40% improvement over the M1's 11 TOPS, supporting advanced machine learning tasks. Memory configurations range from 8 GB to 24 GB of unified LPDDR5, enhancing data sharing across CPU, GPU, and Neural Engine. The M2 includes an upgraded media engine with hardware acceleration for ProRes encoding and decoding, allowing multiple streams of 4K or 8K ProRes video, and a high-bandwidth video decoder for 8K H.264 and HEVC formats. The M2 powers the following products:
ProductAvailability Date
13-inch MacBook Pro (2022)June 17, 2022
MacBook Air (13-inch, 2022)July 15, 2022
iPad Pro (11-inch, 2022)October 26, 2022
iPad Pro (12.9-inch, 2022)October 26, 2022
Mac mini (2023)January 24, 2023
MacBook Air (15-inch, 2023)June 13, 2023
These implementations emphasize battery life and thermal efficiency in compact form factors, with the M2 providing up to 18 hours of video playback in the MacBook Air.

Apple M2 Pro

The Apple M2 Pro is a system on a chip (SoC) designed by Apple for professional-grade Mac computers, announced on January 17, 2023, as part of the company's second-generation Apple silicon lineup. It is fabricated using TSMC's second-generation 5-nanometer process and integrates 40 billion transistors, enabling enhanced performance for demanding workflows such as video editing and 3D rendering. Compared to the base M2 chip, the M2 Pro offers increased core counts and memory bandwidth to support pro-level multitasking. The M2 Pro features a 12-core CPU consisting of eight high-performance cores and four high-efficiency cores, providing up to 20% faster CPU performance than the M1 Pro in multicore tasks. Its GPU is configurable with 16 or 19 cores, delivering up to 65% greater graphics performance over the M1 Pro, with support for hardware-accelerated ray tracing and mesh shading. The integrated 16-core Neural Engine is capable of 15.8 trillion operations per second, accelerating machine learning tasks like image recognition and natural language processing. Memory options range from 16 GB to 32 GB of unified LPDDR5, with 200 GB/s of bandwidth for efficient data access across CPU, GPU, and Neural Engine. The M2 Pro powers the 14-inch and 16-inch MacBook Pro models released in 2023, as well as the Mac mini introduced that year.
ProductRelease Date
14-inch MacBook ProJanuary 17, 2023
16-inch MacBook ProJanuary 17, 2023
Mac miniJanuary 17, 2023
In these devices, it supports up to two external 6K displays at 60 Hz via Thunderbolt alongside the built-in display (on laptops), and includes an HDMI 2.1 port capable of driving one 8K display at 60 Hz or 4K at 240 Hz. This configuration enables professionals to handle multiple high-resolution monitors and peripherals without performance bottlenecks.

Apple M2 Max

The Apple M2 Max is a system on a chip (SoC) designed by Apple for high-end professional workflows, particularly those emphasizing graphics-intensive tasks such as video editing and 3D rendering. Announced on January 17, 2023, it is fabricated on TSMC's second-generation 5 nm process node and contains 67 billion transistors. This chip builds on the architecture of the M2 series, prioritizing a powerful GPU configuration to deliver substantial performance gains for content creators. The M2 Max features a 12-core CPU configuration, consisting of eight high-performance cores and four high-efficiency cores, identical to that of the M2 Pro. Its GPU scales up to 38 cores, available in 30-core or 38-core variants depending on the system configuration, providing up to four times the graphics performance of the base M2's GPU. The 16-core Neural Engine delivers 15.8 trillion operations per second (TOPS), enabling efficient handling of machine learning tasks. Memory options range from 32 GB to 96 GB of unified LPDDR5, supported by a 400 GB/s bandwidth for seamless data access across CPU, GPU, and other components. The M2 Max is integrated into the following products:
ProductRelease Date
14-inch MacBook ProJanuary 2023
16-inch MacBook ProJanuary 2023
Mac StudioJune 2023
It enhances professional applications by accelerating media processing, including ProRes encoding and decoding via its dual media engines. In benchmarks, it demonstrates significant improvements in GPU-bound workloads, such as rendering complex scenes in software like Final Cut Pro, where it outperforms its M1 Max predecessor by up to 2.5 times in graphics tasks. This focus on GPU scalability makes the M2 Max particularly suited for demanding creative pipelines while maintaining power efficiency in a laptop form factor.

Apple M2 Ultra

The Apple M2 Ultra is a system on a chip (SoC) developed by Apple Inc., announced on June 5, 2023, as the highest-end variant in the M2 series for professional desktops. It is fabricated using a second-generation 5-nanometer process by TSMC and integrates 134 billion transistors, connecting two M2 Max dies via Apple's proprietary UltraFusion interconnect technology to form a single cohesive chip. This dual-die design enables unprecedented scale for demanding workloads in creative and technical fields, emphasizing integrated performance across CPU, GPU, and media processing. The M2 Ultra features a 24-core CPU configuration, comprising 16 high-performance cores and 8 high-efficiency cores, delivering up to 20 percent faster CPU performance compared to the prior-generation M1 Ultra. Its GPU options include 60-core or 76-core variants, providing up to 30 percent faster graphics performance over the M1 Ultra, with support for hardware-accelerated ray tracing and mesh shading for advanced rendering tasks. The integrated 32-core Neural Engine achieves 31.6 trillion operations per second (TOPS), marking a 40 percent improvement in machine learning capabilities relative to the M1 Ultra, enabling efficient on-device AI processing for features like image recognition and computational photography. Memory support includes 64 GB, 128 GB, or 192 GB of unified LPDDR5, with a peak bandwidth of 800 GB/s to facilitate seamless data sharing between the CPU, GPU, and Neural Engine. The chip powers the 2023 Mac Studio and Mac Pro desktops,
ProductRelease Date
Mac StudioJune 13, 2023
Mac ProJune 13, 2023
where it supports hardware-accelerated encoding and decoding of 8K ProRes video, including up to 22 simultaneous streams of 8K ProRes 422. Display capabilities extend to up to eight 4K displays at 60 Hz or six Pro Display XDR monitors (exceeding 100 million pixels total), making it suitable for multi-monitor professional setups in video editing and 3D modeling.

Apple M3

The Apple M3 is a system on a chip (SoC) designed by Apple Inc., announced on October 30, 2023, as the third generation of the M-series family for Mac computers. It marks the introduction of TSMC's 3 nm process to the M-series, featuring 25 billion transistors for improved power efficiency and performance density compared to the 5 nm M2. The M3 integrates an 8-core CPU with four high-performance cores and four high-efficiency cores, reaching up to 4.05 GHz on the performance cores. The GPU in the M3 offers configurable 8-core or 10-core options, introducing hardware-accelerated ray tracing and mesh shading for enhanced graphics rendering in complex scenes, alongside dynamic caching for better resource allocation. It also includes a 16-core Neural Engine capable of 18 trillion operations per second (TOPS), accelerating machine learning tasks. Memory configurations range from 8 GB to 24 GB of unified LPDDR5X, with a base bandwidth of 100 GB/s to support multitasking and AI workloads. The M3 powers devices such as the 14-inch MacBook Pro (released November 2023), the 24-inch iMac (released November 2023), the 13-inch and 15-inch MacBook Air models (released March 2024), and the 11-inch and 13-inch iPad Air models (released March 2025).
ProductRelease Date
MacBook Pro (14-inch, 2023)November 2023
iMac (24-inch, 2023)November 2023
MacBook Air (13-inch, 2024)March 2024
MacBook Air (15-inch, 2024)March 2024
iPad Air (11-inch, 2025)March 2025
iPad Air (13-inch, 2025)March 2025
These systems incorporate connectivity features including Wi-Fi 6E and Bluetooth 5.3 for faster wireless performance and lower latency.

Apple M3 Pro

The Apple M3 Pro is a system on a chip (SoC) developed by Apple Inc. as part of its M3 family of processors, announced on October 30, 2023. Fabricated by TSMC using second-generation 3 nm process technology, it incorporates 37 billion transistors, enabling enhanced performance and power efficiency for professional computing tasks. Unlike the base M3, the M3 Pro offers configurable options for higher core counts to balance demanding workloads such as video editing and 3D rendering. The CPU in the M3 Pro is available in an 11-core configuration (5 performance cores and 6 efficiency cores) or a 12-core variant (6 performance cores and 6 efficiency cores), providing scalable processing power for multi-threaded applications. Its integrated GPU comes in 14-core or 18-core setups, supporting hardware-accelerated ray tracing for improved graphics rendering in creative software. The 16-core Neural Engine delivers up to 18 trillion operations per second (TOPS), accelerating machine learning tasks like image recognition and natural language processing. Memory options include 18 GB or 36 GB of unified LPDDR5 RAM with a bandwidth of 150 GB/s, allowing seamless data sharing between the CPU, GPU, and Neural Engine for efficient multitasking. The M3 Pro powers the 14-inch and 16-inch MacBook Pro laptops released in November 2023, featuring three Thunderbolt 4 ports for high-speed connectivity, an HDMI port, and support for up to two external displays at 6K resolution and 60 Hz over Thunderbolt. These capabilities make it suitable for pro users requiring robust I/O and multi-monitor setups.
ProductRelease Date
14-inch MacBook ProNovember 2023
16-inch MacBook ProNovember 2023

Apple M3 Max

The Apple M3 Max is a system on a chip (SoC) designed by Apple for high-end professional workloads, particularly those requiring intensive graphics processing, and serves as the top-tier variant in the M3 family. Announced on October 30, 2023, it is fabricated using TSMC's 3 nm process node and contains 92 billion transistors, enabling significant improvements in power efficiency and performance density compared to the preceding M2 Max on the 5 nm node.
ProductRelease Date
14-inch MacBook ProNovember 2023
16-inch MacBook ProNovember 2023
The M3 Max features a heterogeneous CPU architecture configurable as either a 14-core or 16-core design, comprising 10 or 12 high-performance cores paired with 4 efficiency cores, respectively; this configuration prioritizes multi-threaded tasks like video rendering and 3D modeling while maintaining energy efficiency for lighter loads. Its GPU scales to 30 or 40 cores, supporting hardware-accelerated ray tracing and delivering up to 2.5 times the graphics performance of the M1 Max in professional applications such as Blender or Final Cut Pro. The integrated 16-core Neural Engine provides up to 18 trillion operations per second (TOPS), enhancing machine learning tasks like image recognition and generative AI workflows. Memory support reaches a maximum of 128 GB of unified LPDDR5X RAM in the 16-core configuration, with bandwidth up to 400 GB/s, allowing seamless data sharing between CPU, GPU, and Neural Engine for demanding creative pipelines; the 14-core variant tops out at 96 GB with 300 GB/s bandwidth. It powers the 14-inch and 16-inch MacBook Pro models released in November 2023, where it excels in graphics-intensive scenarios, outperforming the more balanced M3 Pro through additional performance cores and GPU units. In benchmarks, the 16-core M3 Max achieves a Cinebench R23 multi-core score of approximately 24,020, underscoring its capability for sustained professional rendering. The shift to 3 nm also yields better thermal efficiency over the M2 Max, reducing power draw by up to 20% in comparable graphics loads without sacrificing output.

Apple M3 Ultra

The Apple M3 Ultra is the top-tier processor in Apple's M3 series of system-on-a-chip designs, introduced as a dual-die configuration to deliver extreme performance for professional workflows. Announced on March 5, 2025, it builds on the single-die M3 Max by linking two such chips via an enhanced UltraFusion interposer, enabling seamless scaling for demanding tasks like AI model training and 8K video editing. Fabricated on TSMC's second-generation 3 nm process node, the M3 Ultra integrates 184 billion transistors across its dual dies, emphasizing power efficiency alongside high compute density. The CPU offers configurations of 28 or 32 cores, comprising 20 or 24 high-performance cores paired with 8 high-efficiency cores, while the GPU scales to 60 or 80 cores with hardware support for ray tracing, mesh shading, and dynamic caching. A 32-core Neural Engine provides dedicated acceleration for machine learning, capable of 36 trillion operations per second (TOPS), with the improved UltraFusion interconnect optimizing data transfer for AI workloads, including support for large language models exceeding 600 billion parameters. Unified memory configurations range from 96 GB to 256 GB of LPDDR5X, delivering up to 800 GB/s of bandwidth to sustain intensive parallel processing. The chip also incorporates Thunderbolt 5 support for up to 120 Gb/s per port, enhancing connectivity for external storage and displays in pro environments. The M3 Ultra powers the updated Mac Studio, released on March 12, 2025.
ProductRelease Date
Mac StudioMarch 12, 2025

Apple M4

The Apple M4 is a system on a chip (SoC) designed by Apple Inc. as the fourth generation in its M-series family of ARM-based processors for mobile and desktop computing. Announced on May 7, 2024, it was first introduced in the seventh-generation iPad Pro and represents a significant advancement in efficiency and on-device artificial intelligence capabilities. Fabricated by TSMC using its second-generation 3 nm process node (N3E), the M4 integrates 28 billion transistors, enabling enhanced power efficiency compared to its predecessor while supporting demanding workloads in compact devices. The M4's CPU configuration emphasizes efficiency for tablet and entry-level Mac applications, featuring either a 9-core setup with 3 performance cores and 6 efficiency cores (as in the 11-inch iPad Pro) or a 10-core variant with 4 performance cores and 6 efficiency cores (in devices like the Mac mini). This design prioritizes balanced performance for everyday tasks and light professional use, delivering up to 1.5 times faster CPU performance than the M2 in single- and multi-core operations. The integrated 10-core GPU includes hardware-accelerated ray tracing, providing up to four times faster rendering than the M2 GPU, which supports advanced graphics in creative apps. Additionally, the 16-core Neural Engine achieves up to 38 trillion operations per second (TOPS), doubling the M3's capacity and enabling robust on-device AI processing for features like Apple Intelligence; this rating, while retaining the same 16-core configuration as the M3, stems from architectural improvements and measurement at 8-bit precision (e.g., FP8 or INT8), in contrast to the M3's 18 TOPS at 16-bit precision (FP16), which nuances direct generational comparisons. Memory options for the M4 include 8 GB or 16 GB of unified LPDDR5X RAM with 120 GB/s bandwidth, allowing seamless data sharing between CPU, GPU, and Neural Engine for efficient multitasking. The base M4 is used in the following products:
DeviceRelease Date
Seventh-generation iPad ProMay 2024
24-inch iMacOctober 2024
Mac miniNovember 2024
14-inch and 16-inch MacBook Pro (base models)November 2024
13-inch and 15-inch MacBook AirMarch 2025
The chip powers the iPad Pro lineup, where it drives innovations like the Tandem OLED display for superior contrast and brightness, along with hardware acceleration for nano-texture glass options to reduce glare without compromising image quality.

Apple M4 Pro

The Apple M4 Pro is a system on a chip (SoC) developed by Apple Inc. as part of its M4 family of ARM-based processors, designed primarily for high-performance laptops and compact desktops. Announced on October 30, 2024, it utilizes TSMC's second-generation 3-nanometer (N3E) process node, enabling enhanced efficiency and performance compared to prior generations. The M4 Pro features a configurable CPU with either a 12-core configuration (8 performance cores and 4 efficiency cores) or a 14-core setup (10 performance cores and 4 efficiency cores), providing up to 1.9 times the single-core CPU performance of the M1 Pro. Its GPU scales from 16 cores in the base model to 20 cores when configured higher, supporting hardware-accelerated ray tracing and delivering up to 2.4 times the graphics performance of comparable AI PC chips. The integrated 16-core Neural Engine achieves 38 trillion operations per second (TOPS), doubling the AI processing speed of the previous generation and optimizing for on-device Apple Intelligence tasks. Memory configurations begin at 24 GB of unified LPDDR5X RAM and extend up to 64 GB, with a bandwidth of 273 GB/s—75% higher than the M3 Pro. It powers devices such as the 14-inch and 16-inch MacBook Pro and the Mac mini (both 2024 models), released on November 8, 2024. These incorporate Thunderbolt 5 ports for up to 120 Gb/s data transfer.
ProductRelease Date
14-inch MacBook Pro (2024)November 8, 2024
16-inch MacBook Pro (2024)November 8, 2024
Mac mini (2024)November 8, 2024
These systems support up to three 6K displays simultaneously (two external plus the built-in screen), along with features like AV1 hardware decode for efficient video processing. Compared to the base M4, the M4 Pro emphasizes additional performance cores for demanding professional applications like video editing and 3D rendering.

Apple M4 Max

The Apple M4 Max is the top-tier processor in Apple's M4 family of system-on-chips (SoCs), unveiled on October 30, 2024, to power professional-grade computing in laptops. Fabricated on TSMC's enhanced second-generation 3 nm process (N3E), it emphasizes superior graphics capabilities and efficiency for creative and AI-intensive workflows. The CPU configuration includes a 14-core option (10 performance cores and 4 efficiency cores) or a 16-core variant (12 performance cores and 4 efficiency cores), delivering up to 2.2 times the performance of the M1 Max in multicore tasks while maintaining low power consumption. The integrated GPU scales to 32 or 40 cores, incorporating hardware-accelerated ray tracing and dynamic caching for realistic lighting and shading in applications like 3D rendering and game development. This makes the M4 Max particularly suited as a graphics powerhouse compared to the more balanced M4 Pro. In benchmarks, the 40-core GPU achieves up to 50% faster performance than the M3 Max in select graphics tests, such as certain rendering workloads. Complementing the CPU and GPU, the 16-core Neural Engine performs up to 38 trillion operations per second (TOPS), supporting advanced on-device AI features like image generation and natural language processing with enhanced privacy and speed. Memory configurations start at 36 GB and extend to 128 GB of unified LPDDR5X RAM, paired with a maximum bandwidth of 546 GB/s in the 16-core CPU/40-core GPU setup, enabling seamless handling of large datasets in video editing and machine learning. The M4 Max is used in the following products:
ProductRelease Date
14-inch MacBook ProNovember 8, 2024
16-inch MacBook ProNovember 8, 2024
In these devices, it drives the Liquid Retina XDR display and Thunderbolt 5 ports for high-speed data transfer. This integration positions the M4 Max as a key enabler for professionals requiring desktop-level graphics in a portable form factor.

Apple M5

The Apple M5 is a system on a chip (SoC) designed by Apple Inc. as the fifth generation in its M-series lineup of ARM-based processors for computers and tablets. Announced on October 15, 2025, the M5 emphasizes breakthroughs in on-device artificial intelligence (AI) processing and graphics performance, succeeding the M4 and powering select updates to Apple's product lineup. Fabricated by TSMC using the company's third-generation 3 nm process node (N3P), the chip delivers enhanced efficiency and computational power tailored for demanding AI workloads and creative applications. As the entry-level chip in the M-series, the M5 provides efficient performance, class-leading battery life, and suitability for everyday professional use and efficient AI tasks.
ProductRelease Date
14-inch MacBook ProOctober 22, 2025
iPad ProOctober 2025
Apple Vision Pro (enhanced)October 2025
At its core, the M5 integrates a 10-core CPU comprising 4 high-performance cores and 6 high-efficiency cores, with the performance cores capable of reaching up to 4.6 GHz. This architecture yields up to 15% faster multithreaded performance compared to the M4. Independent benchmarks indicate the M5 delivers 15-30% faster CPU and GPU performance compared to the M4, with improvements in AI acceleration and higher memory bandwidth; real-world benefits are noticeable in tasks like rendering and machine learning, though gains are incremental for general use. The CPU builds on Apple's custom silicon design, incorporating advanced branch prediction and larger caches to minimize latency in AI-accelerated operations. The M5's graphics subsystem features a next-generation 10-core GPU, each core equipped with a dedicated Neural Accelerator (matrix-multiply accelerator) to boost machine learning inference. These accelerators enable up to 3-4x faster time-to-first-token (TTFT) for large language models like 30B MoE, achieving under 3 seconds compared to much longer times on the M4. This enables over 4x peak GPU compute performance for AI relative to the M4, alongside up to 45% faster graphics rendering and support for third-generation hardware-accelerated ray tracing, making it suitable for real-time 3D rendering and video editing. Complementing the GPU is an improved 16-core Neural Engine, which, in tandem with the accelerators, supports up to 3.5x faster AI task execution than the prior generation, such as on-device large language model processing. The chip also includes a powerful media engine for hardware-accelerated encoding and decoding of formats like ProRes and AV1. Memory integration in the M5 supports up to 32 GB of unified LPDDR5X RAM with a bandwidth of 153 GB/s—nearly 30% higher than the M4—facilitating seamless data sharing between CPU, GPU, and Neural Engine for improved multitasking and efficiency. Overall power efficiency gains contribute to extended battery life, aligning with Apple's sustainability goals, while the design prioritizes thermal management in slim-form-factor devices. The M5 debuted in the refreshed 14-inch MacBook Pro (available starting October 22, 2025), the updated iPad Pro, and the enhanced Apple Vision Pro, enabling advanced Apple Intelligence features like generative image creation and intelligent text refinement directly on device.

Comparison of M-series processors

The M-series processors represent Apple's progression in custom ARM-based system-on-chip (SoC) designs for desktops and high-end laptops, emphasizing integrated CPU, GPU, Neural Processing Unit (NPU), and unified memory architectures. From the M1's introduction in 2020 to the M5 in 2025, these chips have scaled in core counts, process technology, and AI capabilities while maintaining a focus on power efficiency for portable and compact devices. Key advancements include smaller process nodes enabling higher transistor densities and improved performance per watt, with variants like Pro, Max, and Ultra tailored for escalating workloads in creative, professional, and computational tasks. The following table summarizes specifications for representative M-series variants, highlighting evolution across generations. Data focuses on desktop-oriented configurations where applicable, with base models for entry-level and higher variants for pro workflows.
Chip/VariantProcessCPU Cores/ClocksGPU CoresNPU TOPSMemory (GB/Bandwidth)Transistors (Billion)ReleaseKey Devices
M15 nm4P+4E / up to 3.2 GHz7-8118-16 / 68 GB/s16Nov 2020MacBook Air, iMac, Mac mini
M1 Pro5 nm8P+2E / up to 3.2 GHz14-161116-32 / 200 GB/s33.7Oct 2021MacBook Pro 14" & 16"
M1 Max5 nm8P+2E / up to 3.2 GHz24-321132-64 / 400 GB/s57Oct 2021MacBook Pro 14" & 16"
M1 Ultra5 nm16P+4E / up to 3.2 GHz48-642264-128 / 800 GB/s114Mar 2022Mac Studio, Mac Pro
M25 nm (2nd gen)4P+4E / up to 3.5 GHz1015.88-24 / 100 GB/s20Jun 2022MacBook Air, MacBook Pro
M2 Pro5 nm (2nd gen)6-8P+4E / up to 3.5 GHz16-1915.816-32 / 200 GB/s40Jan 2023MacBook Pro 14" & 16"
M2 Max5 nm (2nd gen)8P+4E / up to 3.5 GHz30-3815.832-96 / 400 GB/s67Jan 2023MacBook Pro 14" & 16"
M2 Ultra5 nm (2nd gen)16P+8E / up to 3.5 GHz60-7631.664-192 / 800 GB/s134Jun 2023Mac Studio, Mac Pro
M33 nm4P+4E / up to 4.05 GHz8-10188-24 / 100 GB/s25Oct 2023MacBook Air, iMac
M3 Pro3 nm5-6P+6E / up to 4.05 GHz14-181818-36 / 150 GB/s37Oct 2023MacBook Pro 14" & 16"
M3 Max3 nm10-12P+4E / up to 4.05 GHz30-401836-128 / 400 GB/s92Oct 2023MacBook Pro 14" & 16"
M3 Ultra3 nm20-24P+8E / up to 4.05 GHz60-803696-128 / 800 GB/s184Mar 2025Mac Studio
M43 nm (2nd gen)4P+6E / up to 4.4 GHz103816-32 / 120 GB/s28May 2024iPad Pro, MacBook Pro (base)
M4 Pro3 nm (2nd gen)10-14P+4-6E / up to 4.4 GHz16-203824-64 / 273 GB/s45 (est.)Oct 2024MacBook Pro 14" & 16"
M4 Max3 nm (2nd gen)10-12P+4E / up to 4.4 GHz32-403864-128 / 546 GB/s90 (est.)Oct 2024MacBook Pro 14" & 16"
M53 nm (3rd gen)4-6P+6E / up to 4.4 GHz104516-32 / 153 GB/s30 (est.)Oct 2025MacBook Pro 14", iPad Pro
Across generations, core scaling has progressed from 8 total CPU cores in the base M1 to over 28 in Ultra variants like the M3 Ultra, enabling parallel processing for demanding applications such as video editing and machine learning. NPU performance has advanced from 11 TOPS in the M1 to over 40 TOPS in the M5, supporting on-device AI tasks like image generation and natural language processing with reduced latency. Memory technology has evolved from LPDDR4X in the M1 (68 GB/s bandwidth) to LPDDR5X in the M5 (up to 153 GB/s in base models), allowing faster data access for graphics-intensive workflows. Performance benchmarks illustrate these gains; for instance, Geekbench 6 single-core scores have risen from approximately 2,346 on the M1 to 4,263 on the M5, reflecting architectural refinements in the performance cores for single-threaded tasks like browsing and coding. Multi-core scores scale dramatically with variants, reaching 17,862 on the base M5—comparable to the M1 Ultra's 18,000—while power consumption remains efficient in a 10-100W TDP range, with base models idling at under 10W and Ultras peaking near 100W under load. The M4 Max features up to 16 CPU cores and 40 GPU cores, excelling in multi-core and graphics-intensive tasks. The M5 provides 10-15% improved single-core performance and over 4x peak GPU compute for AI workloads compared to the M4 series. Variants trade efficiency for peak performance: base models prioritize low power for everyday use, Pro variants add cores for balanced pro workflows like photo editing, Max configurations boost GPU for 3D rendering, and Ultra doubles these via UltraFusion interconnects for workstation-level multi-threading, though at higher thermal demands. Transistor counts have grown exponentially—from 16 billion in the M1 to 184 billion in the M3 Ultra—driving efficiency improvements of up to 2x per generation in tasks like rendering, as visualized in growth curves showing near-doubling every 18-24 months alongside process shrinks from 5 nm to 3 nm. These trends underscore Apple's shift toward AI-optimized, scalable silicon for desktop ecosystems.

S-series SiPs

Apple S1

The Apple S1 is the first system in package (SiP) in Apple's S-series, debuting with the original in April 2015. This compact SiP, measuring 26 mm by 28 mm and containing over 30 individual components, represents Apple's initial foray into custom silicon for wearables, integrating processing, memory, and connectivity into a resin-encased module to fit the device's slim profile. Fabricated on a 28 nm process by Samsung, the S1 enabled the Apple Watch's core functions, including timekeeping, notifications, and basic fitness tracking, while prioritizing power efficiency for an 18-hour battery life. At the heart of the S1 is the APL0778 application processor, a 32-bit ARMv7-based single-core CPU clocked at 520 MHz derived from the architecture. It pairs with an integrated GPU for rendering the Retina display's interface and simple animations. The SiP also incorporates 512 MB of DRAM and 8 GB of NAND flash storage, sufficient for apps, music storage (up to 200 songs), and system data. A dedicated power management integrated circuit (PMIC) from Dialog Semiconductor handles voltage regulation and charging, optimizing the 205 mAh battery. Key to the Apple Watch's health and interaction features, the S1 includes Apple's first M1 motion coprocessor, which processes data from the integrated accelerometer and gyroscope for step counting, gesture recognition, and activity monitoring without burdening the main CPU. It supports the , a linear resonant actuator delivering precise haptic feedback for notifications and Force Touch interactions. The SiP also interfaces directly with the optical heart rate sensor, using green and infrared LEDs alongside photodiodes to measure pulse via photoplethysmography, enabling continuous heart rate tracking during workouts. Exclusively used in the first-generation Apple Watch (including Sport, Standard, and Edition models), the S1 laid the foundation for subsequent S-series advancements in wearable computing.

Apple S1P

The Apple S1P is a second-generation system in package (SiP) designed by Apple for integration into wearable devices, serving as the primary processor for the . Released on September 16, 2016, alongside the introduction of the , the S1P enabled Apple to refresh its entry-level smartwatch lineup with significantly improved performance while retaining the original device's form factor and features, excluding GPS and water resistance. The S1P incorporates a dual-core CPU based on the architecture, operating at 780 MHz, along with an integrated 'Rogue' GPU. This setup delivers up to 50 percent faster CPU performance and double the graphics capabilities compared to the single-core S1 SiP used in the original Apple Watch. Fabricated on a 28 nm process by Samsung Foundry, the S1P also includes 512 MB of LPDDR3 RAM and 8 GB of storage, supporting and later versions. Additionally, it integrates the second-generation M2 motion coprocessor, which offloads processing for sensors such as the accelerometer, gyroscope, and heart rate monitor, enhancing accuracy for fitness tracking and gesture recognition without taxing the main CPU. As a cost-optimized variant of the S2 SiP, the S1P omits the built-in GPS module to reduce complexity and power draw, yet maintains equivalent processing power for apps and interfaces. This integration contributed to improved power efficiency, allowing the Apple Watch Series 1 to achieve up to 18 hours of battery life under typical use—comparable to the original model despite the doubled performance—making it suitable for all-day wear in a compact, aluminum-cased design available in 38 mm and 42 mm sizes.

Apple S2

The Apple S2 is a system in package (SiP) designed by Apple Inc. as the primary processor for wearable devices, marking a significant evolution in the company's S-series lineup for the . Announced on September 7, 2016, and released on September 16, 2016, the S2 was fabricated on a 16 nm process node by , enabling improved power efficiency and integration compared to prior generations. At its core, the S2 features a dual-core CPU based on the architecture—similar in design philosophy to the used in iPhones—with 64-bit processing capabilities to support advanced watchOS features. It includes an integrated GPU that doubles the graphics performance of the previous SiP, facilitating smoother animations and more responsive interfaces on the small form factor of a smartwatch. The package also integrates the M3 motion coprocessor, an ARM Cortex-M3-based unit optimized for low-power handling of accelerometer, gyroscope, and compass data, allowing continuous sensor monitoring without taxing the main CPU. Overall, these components deliver up to 50% faster CPU performance than the S1, enhancing multitasking and app loading times in everyday use. A key innovation of the S2 was the inclusion of a built-in GPS receiver, enabling independent location tracking for activities like running or cycling without requiring a paired iPhone, which relied on connected GPS in earlier models. This hardware addition supported precise distance, pace, and route mapping directly on the device. Complementing this, the S2 enabled 50-meter water resistance certification under ISO standard 22810:2010, making the Apple Watch suitable for shallow-water activities such as swimming in pools or open water, a first for the platform that expanded its utility for fitness enthusiasts. The S2 exclusively powers the Apple Watch Series 2, available in 38 mm and 42 mm aluminum or stainless steel cases, and served as the foundation for , which introduced features like improved notifications and a customizable dock. By prioritizing compact integration and specialized hardware for location and environmental durability, the S2 established new benchmarks for wearable silicon, focusing on health and activity tracking in a battery-constrained environment.

Apple S3

The Apple S3 is a system in package (SiP) developed by Apple Inc. and released in September 2017 as part of the . Announced on September 12 and available starting September 22, it powers the third-generation smartwatch, enabling enhanced performance for health, fitness, and connectivity features. The S3 features a dual-core processor that delivers approximately 70% faster performance than the preceding S2 SiP, supporting more responsive interactions and on-device processing for applications. This architecture emphasizes efficiency, drawing design influences from the energy-optimized cores in Apple's SoC while tailored for wearable constraints. A key advancement in the S3 is the integration of the wireless chip directly into the SiP, which manages 802.11b/g/n Wi-Fi (2.4 GHz) and Bluetooth 5.0 connectivity. This on-package design reduces latency and power consumption compared to discrete components in prior models, enabling faster pairing and data transfer for features like music streaming and notifications. The S3 also incorporates an upgraded barometric altimeter sensor, improving accuracy in elevation measurements for activities such as hiking and stair climbing by better accounting for atmospheric pressure changes during workouts. This enhancement aids in precise tracking of flights climbed and relative altitude gains, integrating seamlessly with the . Exclusively used in the , the S3 supports two variants: GPS-only models for standard connectivity and GPS + Cellular models with optional LTE and UMTS support, allowing independent calls, messages, and app usage without an iPhone nearby.

Apple S4

The Apple S4 is a system-in-package (SiP) introduced by Apple in September 2018 as part of the Apple Watch Series 4. It features a 64-bit dual-core processor built on a 7 nm process node for enhanced performance and power efficiency. The S4 delivers up to twice the processing speed of its predecessor, the , while maintaining all-day battery life of approximately 18 hours, demonstrating improved energy efficiency suitable for a wearable device. At the core of the S4 is a 64-bit dual-core CPU derived from the architecture of the chip, paired with a PowerVR GPU for graphics processing. This configuration enables smoother animations, faster app loading, and support for more complex on-device computations without compromising portability. The SiP integrates additional components, including a wireless chip for connectivity, but prioritizes compact design to fit within the thinner Apple Watch Series 4 chassis. Overall, the S4's architecture emphasizes balance between compute power and low power consumption, setting a foundation for future health-oriented features in Apple's wearables. A key innovation of the S4 is its integration with advanced health monitoring hardware, including an electrical heart sensor that enables electrocardiogram (ECG) functionality to detect atrial fibrillation. Users can take an on-demand ECG reading by placing a finger on the Digital Crown, with the S4 processing the data to generate a waveform viewable in the Health app on paired iPhones. Additionally, the chip supports irregular rhythm notifications, alerting users to potential irregular heart rhythms through continuous optical heart rate monitoring combined with ECG validation. These features, cleared by the FDA for over-the-counter use, represent Apple's push toward proactive health insights in consumer devices, though they require watchOS updates and iPhone pairing for full analysis. The S4's efficiency ensures these sensor-driven tasks run seamlessly without significantly impacting battery life. The S4 powers the Apple Watch Series 4 exclusively, available in 40 mm and 44 mm sizes with GPS or GPS + Cellular options. Its design contributes to the device's overall 30% thinner profile compared to prior models, allowing for a larger display area while housing more components in a smaller footprint. This efficiency gain stems from the 7 nm process and optimized SiP packaging, which reduces power draw during intensive tasks like heart rhythm analysis or GPS navigation.

Apple S5

The Apple S5 is a system in package (SiP) developed by Apple Inc. for wearable devices, introduced in the Apple Watch Series 5 and released in September 2019. It represents Apple's first SiP for the Apple Watch built on a 7 nm manufacturing process, enabling improved power efficiency compared to the preceding S4 on 16 nm. The S5 features a 64-bit dual-core processor architecture similar to the efficiency cores in the A13 Bionic, paired with an integrated GPU, 1 GB of RAM, and 32 GB of storage, supporting watchOS optimizations for low-power operation in always-on scenarios. A key innovation of the S5 is its support for the Always-On Retina LTPO OLED display, which uses low-temperature polycrystalline oxide (LTPO) transistor technology to achieve variable refresh rates as low as 1 Hz, allowing the screen to remain visible for time, notifications, and metrics without constant full-power activation. This enables battery life of up to 18 hours while keeping essential information glanceable, a feature tailored for fitness and daily use in the Apple Watch Series 5. The SiP also integrates enhanced input capabilities, including a faster Digital Crown with haptic feedback via the Taptic Engine, providing tactile scrolling and precise navigation that feels more responsive than prior generations. Additionally, the S5 incorporates a built-in compass (magnetometer) for orientation awareness, which works with the updated Maps app in watchOS 6 to show directional headings during navigation or outdoor activities. Exclusive to the Apple Watch Series 5, the S5 powers a range of case materials including aluminum, stainless steel, titanium, and ceramic, with options for GPS and GPS + Cellular connectivity via the integrated W3 wireless chip and ultra-wideband (UWB) support for precise location. Overall, the S5 prioritizes seamless integration of display, sensor, and processing elements to enhance user interaction in a compact, wrist-worn form factor.

Apple S6

The Apple S6 is a 64-bit dual-core system in package (SiP) designed by Apple for integration into its wearable devices, marking a significant advancement in the company's S-series lineup for smartwatches. Released in September 2020 alongside the Apple Watch Series 6, the S6 is fabricated using TSMC's 7 nm process node, which enables higher efficiency and performance in a compact form factor suitable for wrist-worn computing. Its processor architecture is derived from the chip found in the iPhone 11 series, incorporating a custom dual-core CPU that delivers up to 20% faster performance than the preceding S5 SiP, while maintaining low power consumption for extended battery life in always-on scenarios. A key innovation of the S6 is its integration of the ultra-wideband (UWB) chip, making the Apple Watch Series 6 the first wearable to incorporate this component for enhanced spatial awareness and precise ranging capabilities, such as improved device locating through the Find My network. The SiP also supports the blood oxygen sensor, a new hardware addition to the Apple Watch that uses red and infrared LEDs along with photodiodes to measure blood oxygen saturation (SpO2) levels on demand or during background monitoring, providing users with insights into respiratory health and fitness metrics. This feature processes data through the S6's onboard neural engine and CPU, enabling real-time analysis without compromising the device's slim profile. Furthermore, the S6 incorporates an always-on altimeter, an upgrade over previous generations that continuously calibrates elevation data using barometric pressure sensors for more accurate tracking during activities like hiking or running, even when the device is not actively in use. The S6 powers the Apple Watch Series 6 and first-generation Apple Watch SE models (both 40 mm and 44 mm variants) launched in 2020, which combine these capabilities with 32 GB of storage, Bluetooth 5.0, and GPS for standalone operation. Overall, the S6's design emphasizes seamless integration of health-focused processing with connectivity features, setting a foundation for subsequent S-series evolutions in wearable silicon.

Apple S7

The Apple S7 is a system in package (SiP) developed by Apple Inc. for use in wearable devices. It was first introduced in September 2021 alongside the . The S7 integrates a 64-bit dual-core processor, 1 GB of RAM, 32 GB of storage, and wireless connectivity components including Bluetooth 5.0 and 802.11b/g/n Wi-Fi, all within a compact design optimized for power efficiency in smartwatches. Fabricated on TSMC's 7 nm process node, the S7's CPU cores are derived from the architecture used in the , providing enhanced performance over previous S-series chips while maintaining low power consumption suitable for always-on wearables. This design enables up to 20% faster code execution compared to the SiP in the , supporting more responsive interactions and smoother animations on larger displays. The integrated GPU handles graphics rendering for the device's Retina display, contributing to features like the always-on functionality without significantly impacting battery life, which remains rated at up to 18 hours. A key advancement in the S7 is its role in enabling faster charging capabilities for the , achieving 80% charge in about 45 minutes using the included USB-C Magnetic Fast Charger—33% quicker than prior models. This improvement stems from optimizations in the SiP's power management and integration with the watch's new charging architecture. Additionally, the S7 powers the edge-to-edge Always-On Retina LTPO OLED display, which offers nearly 20% more screen area and thinner borders (1.7 mm) compared to the Series 6, available in 41 mm and 45 mm sizes with resolutions up to 396 x 484 pixels. These enhancements make text and icons larger and easier to read, improving usability for health and fitness tracking features inherited from previous generations, such as ECG, blood oxygen monitoring, and fall detection. The S7 also incorporates the Apple wireless chip for Bluetooth and Wi-Fi connectivity, along with the Ultra Wideband chip for precise location services in GPS + Cellular models. Overall, the SiP's compact footprint—reportedly smaller and dual-sided compared to predecessors—allowed for better internal space utilization in the , accommodating the expanded display while adding IP6X dust resistance and a front crystal up to 50% more crack-resistant.

Apple S8

The Apple S8 is a system in a package (SiP) developed by Apple Inc. and released in September 2022 as part of the Apple Watch Series 8, Apple Watch Ultra, and second-generation Apple Watch SE. It is fabricated by TSMC using a 7 nm process node and features a 64-bit dual-core CPU derived from the architecture of the A13 Bionic chip. The S8 integrates a power-efficient GPU, 1 GB of RAM, 32 GB of storage, Bluetooth 5.3, and an ultra-wideband (UWB) chip for precise location tracking, enabling enhanced performance in wearable applications. A key advancement in the S8 is its support for crash detection, a safety feature that leverages the SiP's processing capabilities alongside new dual-axis accelerometer and gyroscope sensors to identify severe car crashes, including front-impact, rear-end, side-impact, and rollover incidents. The algorithm fuses data from the accelerometer, gyroscope, barometer, GPS, and microphone to detect irregular motion patterns indicative of a crash, triggering an audible alarm, on-screen alert, and automatic emergency call if the user remains unresponsive. This feature, powered by the S8's motion coprocessor, extends to all devices equipped with the chip and represents an evolution from prior fall detection systems by focusing on vehicular accidents. The S8 also introduces Low Power Mode, which optimizes battery life by reducing background sensor polling, dimming the display, and limiting certain features while preserving core functions like notifications and heart rate monitoring. In this mode, the Apple Watch Series 8 and SE (2nd generation) achieve up to 36 hours of battery life on a single charge, doubling the standard 18-hour duration through the S8's efficient dual-core design and power management. Additionally, the S8 enhances GPS accuracy by prioritizing the watch's internal GNSS receiver over connected iPhone data, providing a clearer sky view for improved location precision during activities like running or cycling.

Apple S9

The Apple S9 is a system in package (SiP) developed by Apple Inc. and released in September 2023 as part of the Apple Watch Series 9 and Apple Watch Ultra 2. It incorporates a 64-bit dual-core CPU fabricated on TSMC's 4 nm process node, marking an advancement over prior S-series chips in efficiency and performance. The S9 also includes a 4-core Neural Engine, which doubles the speed of machine learning tasks compared to the previous generation, enabling enhanced on-device processing capabilities. A key feature of the S9 is its support for on-device Siri processing, allowing the virtual assistant to handle requests locally without relying on cloud servers for improved privacy and responsiveness. This Neural Engine integration facilitates the double-tap gesture, a intuitive control method where users pinch their index finger and thumb twice to perform actions like answering calls, scrolling, or playing media, detected via the watch's accelerometer and gyroscope. The chip's performance also contributes to the devices' brighter always-on display, capable of reaching up to 2000 nits for better outdoor visibility.

Apple S10

The Apple S10 is a system in package (SiP) developed by Apple Inc. as the primary processor for its latest wearable devices, emphasizing health monitoring and power efficiency. Announced on September 9, 2024, alongside the , the S10 powers the device's core functions, including on-device processing for user interactions like the double tap gesture and . It was released to consumers on September 20, 2024, and extended to the Apple Watch Series 11, SE (3rd generation), and Ultra 3 announced on September 9, 2025. Featuring a 64-bit dual-core processor and a 4-core , the S10 enables advanced machine learning capabilities directly on the device, supporting features that rely on real-time data analysis without cloud dependency. This architecture derives efficiency from Apple's ongoing silicon optimizations, allowing seamless integration with sensors for health and fitness tracking. The chip's design prioritizes low-power operation, contributing to overall device performance in compact form factors. A key focus of the S10 is enhanced health detection, including improved sleep apnea monitoring via the accelerometer, which tracks subtle wrist movements indicative of breathing interruptions during sleep. Users receive notifications after consistent patterns are detected over multiple nights, with the feature cleared by regulatory bodies for accuracy in identifying moderate to severe cases. The S10 also supports hypertension alerts, utilizing data from the optical heart sensor to analyze vascular responses to heartbeats; after a 30-day calibration using a traditional blood pressure cuff, the watch notifies users of potential high blood pressure trends. These capabilities build on prior generations but leverage the S10's Neural Engine for more precise, on-device computations. The S10's efficiency improvements extend to battery management, enabling the Apple Watch Series 10 to deliver up to 18 hours of all-day use or 36 hours in Low Power Mode, with fast charging reaching 80% capacity in 30 minutes—matching or exceeding prior models despite a thinner chassis. This is achieved through optimized power distribution and reduced idle consumption, allowing sustained sensor activity without frequent recharges. Real-world testing often reports 30 to 36 hours under mixed usage, highlighting the chip's role in balancing performance and longevity. The 2025 models (, , ) incorporate the S10 with enhancements like up to 24 hours of battery life and new health insights powered by , such as sleep score and Workout Buddy, without a new chip generation. Integrated into these devices, the S10 supports a larger, wide-angle always-on up to 30% brighter at angles and a case that's 10% thinner than the Series 9, enhancing wearability while maintaining robust feature sets like depth sensing for water activities.

Comparison of S-series processors

The S-series processors represent Apple's architecture tailored for wearable devices like the , combining CPU, GPU, memory, power management, and connectivity in a compact form to optimize power efficiency and space constraints. These SiPs differ from standard by stacking multiple discrete dies and components within a single package, enabling higher integration density for thin, battery-limited designs. Over successive generations, the S-series has evolved to support advanced health sensors and wireless features while reducing overall power draw.
ChipProcessCPU Cores/ArchIntegrated Components (Motion/Wireless)Key FeaturesReleaseDevices
S1P28 nmDual-core ARMv7k (32-bit)M2 motion coprocessor, W1 wireless chip (Bluetooth 4.0, Wi-Fi 802.11b/g/n)Integrated 512 MB RAM, 8 GB storage, PowerVR GPU; basic fitness tracking2016Apple Watch Series 1
S216 nmDual-core ARMv8 (64-bit)M10 motion coprocessor, W2 wireless chip (Bluetooth 4.0, Wi-Fi 802.11b/g/n)Built-in GPS/GLONASS, water resistance 50m; improved location accuracy2016Apple Watch Series 2
S316 nmDual-core ARMv8 (64-bit)Integrated motion coprocessor, W2 wireless chip (Bluetooth 4.2, Wi-Fi 802.11b/g/n)Optional LTE/UMTS, altimeter; first 64-bit architecture in wearables2017Apple Watch Series 3
S47 nmDual-core 64-bit (Vortex architecture)Integrated motion coprocessor, W3 wireless chip (Bluetooth 5.0, Wi-Fi 802.11b/g/n)ECG sensor support, fall detection; 2x faster CPU than S32018Apple Watch Series 4
S57 nmDual-core 64-bitIntegrated motion coprocessor, W3 wireless chip (Bluetooth 5.0, Wi-Fi 802.11b/g/n)Always-on Retina display, compass; low-power mode enhancements2019Apple Watch Series 5
S67 nmDual-core 64-bitIntegrated motion coprocessor, W3 wireless chip (Bluetooth 5.0, Wi-Fi 802.11b/g/n)Blood oxygen sensor (oximeter), U1 ultra-wideband; 20% faster than S52020Apple Watch Series 6, SE (1st gen)
S77 nmDual-core 64-bitIntegrated motion coprocessor, W3 wireless chip (Bluetooth 5.0, Wi-Fi 802.11b/g/n/ac)Faster charging (80% in 45 min), larger display support2021Apple Watch Series 7
S87 nmDual-core 64-bitIntegrated motion coprocessor, W3 wireless chip (Bluetooth 5.3, Wi-Fi 802.11b/g/n)Temperature sensor, crash detection; same CPU as S6/S7 with efficiency tweaks2022Apple Watch Series 8, Ultra, SE (2nd gen)
S94 nmDual-core 64-bit, 4-core Neural EngineIntegrated motion coprocessor, W3 wireless chip (Bluetooth 5.3, Wi-Fi 802.11b/g/n)On-device Siri processing, double-tap gesture; 30% faster GPU than S82023Apple Watch Series 9, Ultra 2
S104 nmDual-core 64-bit, 4-core Neural EngineIntegrated motion coprocessor, W3 wireless chip (Bluetooth 5.3, Wi-Fi 802.11b/g/n)Thinner SiP design, depth/water temperature sensing; same core as S9 with size reduction; extended to 2025 models2024Apple Watch Series 10, Series 11, Ultra 3, SE (3rd gen)
A key trend in the S-series is the progressive shrink in manufacturing process nodes, from 28 nm in the S1P to 4 nm in the S9 and S10, enabling higher transistor density and lower power consumption without expanding die size. This evolution has facilitated the addition of specialized sensors, such as in the S4, blood oxygen monitoring () in the S6, and skin temperature sensing in the S8, enhancing health tracking capabilities while maintaining 18-36 hours of battery life across models. Motion processing transitioned from discrete M-series coprocessors in early designs to fully integrated units by the S3, offloading accelerometer and gyroscope tasks from the main CPU for better efficiency. Performance metrics highlight the series' focus on efficiency over raw speed; for instance, the S4 delivered twice the CPU performance of the S3, reducing app launch times to under 1 second, while die size decreased from approximately 130 mm² in the S1 to around 30 mm² in later generations like the S9, positively impacting battery life by 20-50% through reduced leakage. variants, introduced with the S3 and refined in subsequent chips, support optional cellular connectivity for independent operation, though they consume up to 10% more power than GPS-only models. Overall, these SiPs prioritize integration for wearables, contrasting with the compute-heavy A- and M-series SoCs. In 2025, the S10 continues without a successor chip, powering new models with software-enabled features like advanced AI health insights.

Secure Enclaves

Apple T1

The Apple T1 is Apple's first dedicated secure enclave chip designed specifically for Mac computers, marking the introduction of ARM-based security hardware in the Mac lineup. Unveiled in late 2016 alongside the redesigned MacBook Pro models featuring the , the T1 serves as a coprocessor that offloads critical security tasks from the main Intel CPU, enabling faster and more isolated processing of sensitive operations. This chip represents an early step in Apple's integration of custom silicon for security in its desktop ecosystem. Architecturally, the T1 is an ARMv7-based system-on-chip (SoC) that incorporates a dedicated coprocessor for handling biometric and cryptographic functions. For authentication, the T1 securely captures, encrypts, and stores fingerprint data within the , ensuring that biometric information never leaves the chip unencrypted and is isolated from the main system memory. The chip also supports encrypted storage by generating and safeguarding encryption keys for full-disk encryption, protecting user data on the SSD even if the drive is removed. These features enhance privacy and resistance to physical attacks, with the 's design providing hardware-rooted isolation for key management. Beyond security, the T1 controls the 's OLED display by receiving pixel data from the primary GPU and rendering contextual interface elements, such as dynamic function keys and Apple Pay dialogs. It operates independently using a lightweight, modified version of (bridgeOS 1.0), which allows real-time updates to the without relying on the main kernel. This separation improves responsiveness and security for the 's interactions. The T1's secure design aligns with Apple's broader core architecture principles, prioritizing hardware-enforced isolation for sensitive tasks. The T1 is permanently integrated onto the Mac's logic board via soldering, rendering it non-user-replaceable and requiring professional service for any related repairs. It was exclusively deployed in MacBook Pro models equipped with the , including the 13-inch and 15-inch variants released in 2016 and continued into select 2017 configurations. This tight hardware integration ensures seamless operation but limits upgradeability, reflecting Apple's emphasis on a unified system design.

Apple T2

The Apple T2 is Apple's second-generation custom silicon designed as a secure coprocessor for Intel-based Macintosh computers, introduced in December 2017 alongside the iMac Pro. Built as an ARM-based system on a chip (SoC) and fabricated by TSMC on a 16 nm FinFET process, the T2 expands on the security-focused role of its predecessor by incorporating additional system-level controllers for enhanced performance and privacy features. This integration allows the chip to handle both dedicated security tasks and auxiliary processing independently from the main Intel CPU, ensuring a hardware root of trust for the entire system. At the core of the T2 is its coprocessor, a dedicated subsystem that isolates and manages sensitive operations to protect user data and system integrity. It securely stores cryptographic keys, processes sensor data for authentication (with fingerprint templates encrypted and never leaving the chip), and enforces (SIP) to prevent unauthorized modifications to critical system files. The Secure Enclave also oversees secure firmware storage and boot processes, verifying the cryptographic signatures of macOS, firmware, and other boot components before allowing execution, thereby mitigating boot-time attacks. For data encryption, it includes a hardware-accelerated AES-XTS engine integrated into the direct memory access (DMA) path between flash storage and system memory, enabling efficient full-disk encryption without impacting overall performance. Password attempt limits are enforced (up to 30 at login, 10 in Recovery Mode, and 90 total with recovery key options), with escalating time delays to thwart brute-force attacks. The T2 chip's integrated I/O capabilities further distinguish it as a multifaceted coprocessor. Its SSD controller manages internal solid-state drive operations, supporting read/write speeds up to approximately 3 GB/s while ensuring all data passes through the Secure Enclave for encryption and integrity checks. An onboard image signal processor (ISP) enhances video quality from the built-in FaceTime HD webcam, providing features like improved low-light performance and noise reduction. The audio controller handles sound processing, including a hardware microphone disconnect feature on Mac portables that cuts the mic circuit when the lid is closed to prevent unauthorized audio capture. Additionally, the T2 incorporates the system management controller (SMC) for power, thermal, and battery oversight. These components enable privacy-focused features such as always-on "Hey Siri" detection, where the chip locally analyzes audio input for the wake phrase without sending data to the cloud unless activated, and configurable security options via the Startup Security Utility, which allows users to toggle secure boot modes, external boot media permissions, and accessory authentication. However, the T2 chip has known security vulnerabilities. In October 2020, a hardware flaw was discovered that allows low-level system access similar to the exploit on iOS devices, which cannot be patched via software updates. Additionally, in February 2022, forensic toolmaker Passware announced a method to bypass the T2's password attempt limits, enabling brute-force attacks on -encrypted drives, potentially cracking weak passwords in hours. These issues highlight limitations in the T2's hardware security design, though Apple has mitigated some risks through software updates where possible. The T2 chip powered a range of Intel-based Macs from late 2017 through 2020, including the iMac Pro (2017), MacBook Pro models introduced in 2018 and later, iMac models from 2019 and later, Mac mini from 2018 and later, and Mac Pro from 2019. By centralizing security and I/O functions in custom silicon, the T2 improved overall system efficiency and user control during Apple's transition era toward fully integrated ARM-based designs.

Comparison of T-series processors

The and processors represent the core of Apple's T-series, functioning as dedicated secure enclaves for Intel-based Macs, handling critical security tasks like cryptographic operations and system integrity verification. Introduced sequentially, these chips evolved from basic biometric and boot security in the T1 to more comprehensive I/O and storage management in the T2, bridging Apple's ARM-based security expertise with x86 systems. Their design emphasized hardware isolation to protect sensitive data, such as encryption keys and biometric templates, from software attacks.
ChipProcessCPUKey Functions (Secure Boot, I/O, Sensors)DevicesYear
16 nmSingle ARMv7 core sensor processing; display I/O; Encrypted storage for MacBook Pro 13-inch and 15-inch (Touch Bar models)2016
16 nmDual ARMv8 cores with extended OS support; SSD and audio controller I/O; and microphone sensorsiMac Pro (2017), iMac (2019–2020), MacBook Pro (2018–2020), MacBook Air (2018–2020), Mac mini (2018), Mac Pro (2019)2017
The T1 provided foundational security with a single-core ARM processor focused on essential tasks like authentication and control, lacking advanced storage integration. In contrast, the T2 doubled the core count for improved performance and added specialized controllers for SSD encryption and audio processing, enabling features like hardware-accelerated without relying on the main CPU. This progression reflects a trend toward deeper hardware-software integration in Apple's ecosystem, culminating post-2020 with the transition to Apple silicon SoCs like M1, where T2-like secure enclave functions are embedded directly into the primary processor for streamlined security across devices. The T-series chips significantly bolstered user privacy by isolating sensitive operations; for instance, the 's enclave architecture ensures encryption keys cannot be extracted even if the system is compromised, preventing unauthorized data access in scenarios like theft or malware infection.

Ultra-Wideband Chips

Apple U1

The Apple U1 is an ultra-wideband (UWB) chip developed by Apple, released in 2019 as part of the iPhone 11 lineup. It adheres to the IEEE 802.15.4z standard, enabling secure, short-range wireless communication with high precision. Built on a 16 nm process, the U1 is included in devices featuring the processor and later A-series chips up to , as well as the system in package for the Apple Watch Series 6 and earlier models like AirPods Pro (2nd generation) charging case (2022). The U1 chip facilitates centimeter-level location accuracy, typically within 10 centimeters, through time-of-flight ranging and angle-of-arrival direction finding. This precision supports features like Precision Finding in the Find My app, allowing users to locate compatible items such as AirTags with directional guidance and distance indicators. It powers UWB-based tracking for accessories, including the AirTag introduced in 2021 and the HomePod mini smart speaker, enhancing spatial awareness in iPhone 11 and later models. Key features of the U1 include secure pairing protocols that employ MAC address randomization and sequence number checks to prevent unauthorized access and relay attacks. Designed for low power consumption similar to Bluetooth Low Energy, it minimizes battery drain during operation. Processing for advanced spatial tasks, such as direction finding, leverages the device's Neural Engine in conjunction with the U1 for efficient computation.

Apple U2

The Apple U2 is Apple's second-generation Ultra Wideband (UWB) chip, introduced in 2023 as an upgrade to the original U1 chip for enhanced spatial awareness and precise location tracking in compatible devices. It supports the standard, which provides improved secure ranging capabilities through features like scrambled timestamp sequences to prevent relay attacks and unauthorized tracking. Built on a more efficient 7 nm process, the U2 offers three times the ranging distance compared to the first-generation U1—up to approximately 30 meters—enabling more reliable connections over greater distances while maintaining centimeter-level accuracy for direction and proximity detection, resulting in lower power consumption and faster signal processing for real-time applications. As of November 2025, the U2 is included in the and later (A16 Bionic and subsequent), Apple Watch Series 9 and later (including Series 10 and Ultra 3), and AirPods Pro (3rd generation) charging case. The U2 chip powers advanced features such as Precision Finding in the Find My app, which provides users with directional arrows, distance measurements, and haptic feedback to locate nearby items or people equipped with compatible UWB technology. For instance, iPhone 15 and later users can employ Precision Finding to meet up with friends who have enabled the feature, receiving visual, audio, and tactile cues even in crowded environments or when the target device is in another room. Enhancements in and later (2024–2025) further leverage the U2 for improved spatial audio handoff and device interactions. Privacy is a core aspect of the U2 implementation, incorporating randomized identifiers analogous to MAC address randomization to obscure device identities during UWB communications and prevent persistent tracking by unauthorized parties. All ranging sessions are end-to-end encrypted, with data shared only between opted-in users via the Find My network, aligning with Apple's broader security protocols for location services. These enhancements ensure that while the chip enables seamless interactions like automatic device handoff or key sharing, it minimizes risks associated with radio-based proximity detection.

Comparison of U-series processors

The Apple U-series processors represent the company's dedicated ultra-wideband (UWB) chips designed for precise spatial awareness and location services. The U1, introduced in 2019, marked Apple's entry into UWB technology, enabling centimeter-level accuracy for features like Precision Finding. The U2, released in 2023, builds on this foundation with enhancements in range and efficiency on a smaller process node, supporting broader applications in device tracking and secure interactions.
ChipStandardRangeKey FeaturesIntegration (SoCs)Release
U1IEEE 802.15.4zUp to ~10 meters; centimeter-level accuracy via time-of-flight and angle-of-arrival; low-power spatial awareness for device pairing and locatingStandalone chip alongside A13 Bionic to A16 (e.g., iPhone 11–14, AirTag, AirPods Pro 2nd gen case)2019
U2IEEE 802.15.4z (enhanced secure ranging)Up to ~30 meters (3x U1); centimeter-level accuracy; improved power efficiency; enhanced direction findingStandalone chip alongside A16 Bionic and later A-series SoCs (e.g., iPhone 15–17 series, Apple Watch Series 9–10 and Ultra 3, AirPods Pro 3rd gen case)2023
The evolution of U-series processors has progressed from the U1's focus on short-range, centimeter-accurate positioning to the U2's expansion into longer distances with directional capabilities, while prioritizing power efficiency. This shift reflects advancements in UWB signal processing and antenna design, reducing latency and energy consumption for always-on location services. These processors underpin key applications in Apple's ecosystem, notably the expansion of the Find My network, which leverages crowdsourced UWB signals for global item tracking with directional guidance. Additionally, U-series chips enable secure Digital Key functionality for vehicles, allowing users to unlock and start cars with precise proximity detection via standards like Car Connectivity Consortium protocols.

Wireless Connectivity Chips

Apple W2

The Apple W2 is a custom wireless connectivity chip developed by Apple, introduced in 2017 as part of the Apple Watch Series 3. It represents the company's first fully integrated wireless solution for the Apple Watch, combining Wi-Fi, Bluetooth, and NFC functionalities into a single die within the S3 System in Package (SiP). This integration allowed for more efficient space utilization compared to the discrete components used in prior models, contributing to the device's compact form factor while enabling advanced connectivity features. The W2 supports 802.11b/g/n Wi-Fi at 2.4 GHz, Bluetooth 4.2, and NFC for secure transactions like Apple Pay. Key improvements over previous wireless implementations include up to 85% faster Wi-Fi connections and 50% greater power efficiency for both Wi-Fi and Bluetooth operations, which helped maintain the Apple Watch's all-day battery life despite enhanced performance. These enhancements enabled practical uses such as direct Wi-Fi streaming of music from Apple Music to the watch without needing a paired iPhone nearby, ideal for activities like running. Additionally, the chip supports faster Bluetooth pairing—up to 2.5 times quicker than Bluetooth 4.2 low energy—streamlining device setup and reconnection. Exclusively featured in the Apple Watch Series 3 (both GPS and GPS + Cellular models), the W2 played a pivotal role in advancing the wearable's independence from smartphones by improving wireless reliability and speed for calls, messages, and data syncing.

Apple W3

The Apple W3 is a custom wireless connectivity chip developed by Apple, first introduced in September 2018 with the Apple Watch Series 4. It integrates Wi-Fi and Bluetooth functionality into a single system-on-chip, succeeding the W2 chip from the previous generation, and was incorporated into Apple Watch models from Series 4 through Series 10 (2018–2024). In 2025, Apple introduced a new custom-designed in-house wireless chip (succeeding the W3) in the Apple Watch Series 11, Ultra 3, and SE 3, supporting Bluetooth 5.3 and Wi-Fi 4 (802.11n) on both 2.4 GHz and 5 GHz bands for enhanced efficiency and connectivity. Key specifications of the W3 include support for 802.11b/g/n Wi-Fi at 2.4 GHz and Bluetooth 5.0, marking an upgrade from the W2's Bluetooth 4.2 protocol. Bluetooth 5.0 doubles the raw data transfer speed to 2 Mbps compared to 1 Mbps in Bluetooth 4.2, while extending the broadcast range up to four times farther—to approximately 240 meters in ideal conditions—enabling more reliable connections over greater distances. Additionally, Bluetooth 5.0 incorporates enhanced low-energy modes that improve power efficiency, reducing consumption during extended pairing, streaming, and data broadcasting tasks without compromising the device's overall battery performance. These advancements in the W3 contribute to smoother wireless experiences in Apple Watch devices, particularly for non-cellular models relying on Wi-Fi for software updates and media streaming, and for cellular variants where robust Bluetooth handover supports seamless integration with iPhones during calls and notifications. The chip's efficiency helps sustain the standard 18-hour battery life across generations, even as additional sensors and displays increase power demands.

Comparison of W-series processors

The W-series processors represent Apple's progression in custom wireless connectivity chips designed primarily for wearables, evolving from Bluetooth-focused solutions to integrated Wi-Fi and Bluetooth capabilities that prioritize power efficiency and seamless device pairing. The series began with the W1 in 2016, emphasizing battery optimization for audio devices, and advanced through the W2 and W3 (2017–2018), incorporating faster data rates and reduced power consumption to support extended usage in compact form factors like the Apple Watch. The W3 was used up to the Apple Watch Series 10 in 2024, succeeded by a new unnamed Apple-designed wireless chip in 2025 models.
ChipWi-Fi SpecBT VersionSize/EfficiencyIntegrationRelease
W1None4.214.3 mm²; optimized for low-power Bluetooth audio managementStandalone SoC for Bluetooth in headphones2016
W2802.11b/g/n (2.4 GHz)4.26.5 mm²; 50% more power-efficient for Bluetooth and Wi-Fi than W1, with 85% faster Wi-FiCombined Bluetooth/Wi-Fi SoC in wearables2017
W3802.11b/g/n (2.4 GHz)5.0Enhanced efficiency via Bluetooth 5.0 for lower power in extended range scenariosIntegrated into S-series SiPs for Apple Watch2018
Wireless (successor)802.11n (2.4/5 GHz)5.3Improved power efficiency for advanced connectivity and health featuresIntegrated into S11 and later SiPs for Apple Watch2025
Key trends in the W-series include the upgrade from Bluetooth 4.2 in the W1 and W2 to Bluetooth 5.0 in the W3, enabling doubled speed, quadrupled range, and improved broadcasting for always-connected wearables while reducing power usage compared to prior standards. Power savings have scaled significantly, with the W2 achieving 50% lower consumption for wireless functions relative to the W1, allowing devices like the Apple Watch to maintain connectivity for days on a single charge without compromising performance. These advancements support always-on features, such as real-time health monitoring and notifications, by minimizing energy draw during idle states. The impacts of these evolutions are evident in enhanced audio streaming reliability and reduced latency for interactive elements like haptics in wearables; for instance, the W2's faster Wi-Fi enables smoother music and call streaming over direct internet connections, while Bluetooth 5.0 in the W3 cuts latency by up to 40% for tactile feedback in fitness tracking. This has broadened wearable utility, from prolonged battery life in headphones to seamless ecosystem integration for low-power, high-fidelity wireless experiences.

Bluetooth Audio Processors

Apple W1

The Apple W1 is a custom wireless system-on-chip (SoC) designed by Apple for Bluetooth audio processing, debuting in 2016 with the first-generation AirPods. Announced on September 7, 2016, and released on December 13, 2016, the W1 powers the earbuds' core operations, including audio routing, sensor integration, and power management, enabling a seamless truly wireless experience without traditional headphone cables. Operating at one-third the power consumption of conventional wireless chips, it delivers up to 5 hours of listening time on a single charge and over 24 hours with the charging case. Key to the W1's functionality is its support for Bluetooth 4.2, which facilitates instant pairing with iOS devices via a one-tap setup process linked to the user's iCloud account. It handles high-quality audio transmission using the AAC codec, ensuring efficient streaming of compressed audio tracks while maintaining compatibility with standard Bluetooth devices. The chip also integrates with dual beamforming microphones and a speech-detecting accelerometer, which focus on the user's voice and suppress ambient noise during calls or voice commands. The W1 excels in multi-device ecosystems through iCloud-based handover, automatically transferring audio playback between signed-in Apple devices like iPhone, iPad, Mac, or Apple Watch as the user moves between them. Its optimized architecture supports low-latency audio, with effective delays around 200 ms that Apple devices compensate for during video playback to achieve lip-sync accuracy. Quick access to Siri is provided via a double-tap gesture on either AirPod, allowing control of music, calls, and queries without touching the connected device. Primarily featured in first-generation AirPods, the W1 also appears in early Beats models such as the Solo3 and Powerbeats3 headphones.

Apple H1

The Apple H1 is a system-on-chip (SoC) designed by Apple for wireless audio devices, introduced in 2019 as a successor to the W1 chip. It is based on the ARM architecture and integrates Bluetooth connectivity, audio processing, and sensor fusion capabilities into a single package. The H1 debuted in the second-generation AirPods, released on March 20, 2019, and was later incorporated into the first-generation AirPods Pro, launched on October 28, 2019, as well as subsequent models like the third-generation AirPods. This chip powers advanced features in these earbuds, enhancing audio quality and user interaction while maintaining low power consumption for extended battery life. Key to the H1's functionality is its support for Bluetooth 5.0, which enables faster device pairing—up to twice as quick when switching between Apple devices—and improved connection stability compared to previous generations. For audio processing, it includes adaptive EQ, which automatically tunes sound based on the fit of the earbuds, and enables active noise cancellation (ANC) in compatible models like the AirPods Pro. The H1 also supports Transparency mode, allowing ambient sounds to pass through for situational awareness without removing the earbuds. These features contribute to a more immersive listening experience, with the chip handling real-time audio computations on-device. The H1 integrates motion processing capabilities, utilizing built-in accelerometers and gyroscopes in the earbuds to enable head tracking for spatial audio, which simulates a three-dimensional soundstage. This allows audio to dynamically adjust based on head movement, providing cues for directional sound in supported media. Additionally, the chip accelerates voice interactions with Siri, enabling hands-free activation via "Hey Siri" and faster processing for wake word detection and command response. Overall, the H1's design emphasizes seamless integration with Apple's ecosystem, prioritizing low-latency audio and efficient sensor handling without relying on the host device's main processor.

Apple H2

The Apple H2 is a custom system-on-chip (SoC) developed by Apple for Bluetooth audio devices, introduced in September 2022 as an enhanced successor to the H1 chip. It integrates computational audio processing, wireless connectivity, and sensor fusion to enable advanced features in wireless earbuds, focusing on immersive sound and environmental adaptation. The H2 powers the second-generation AirPods Pro, which were updated with a USB-C charging case in September 2023. Supporting Bluetooth 5.3, the H2 provides high-bandwidth, low-latency wireless audio transmission while doubling the active noise cancellation (ANC) performance of the H1 chip, effectively reducing ambient noise by up to twice as much during playback. It introduces Adaptive Transparency mode, which intelligently balances noise reduction with awareness of surroundings by attenuating sudden loud sounds while preserving natural hearing for voices and alerts. The chip's on-device processing also enables Conversation Awareness, where integrated microphones detect when the user speaks, automatically lowering media volume and enhancing voice isolation for clearer interactions without relying on cloud computation. The H2 further refines Spatial Audio with dynamic head tracking, using motion sensors to anchor sound to the user's environment and adjust in real-time for a three-dimensional listening experience. By 2025, the chip is featured in the AirPods Pro (2nd generation), AirPods 4 (with Active Noise Cancellation, released September 2024), and AirPods Pro (3rd generation, released September 2025), all of which support Precision Finding through U1 or U2 chips in their charging cases for ultra-wideband location accuracy via the Find My network.

Comparison of Bluetooth audio processors

Apple's Bluetooth audio processors, including the W1, H1, and H2 chips, have evolved to enhance wireless audio experiences in earbuds and headphones, focusing on seamless integration within the Apple ecosystem while advancing audio quality and efficiency. The W1 laid the foundation with optimized pairing and basic processing, while subsequent generations introduced advanced noise management, spatial audio, and computational enhancements powered by on-chip neural engines. These chips handle Bluetooth connectivity, audio decoding, and sensor fusion, but differ significantly in supported protocols, feature sets, and performance metrics. The following table summarizes key specifications across the processors:
ChipBluetooth VersionKey Audio FeaturesMotion/Voice CapabilitiesDevicesRelease Year
W14.2Basic DSP for audio streaming, fast one-tap pairingMotion accelerometer for in-ear detectionAirPods (1st generation), Beats Solo3, Beats Studio3 Wireless, Powerbeats32016
H15.0Adaptive EQ, Audio Sharing, Active Noise Cancellation (ANC) and Transparency mode in compatible modelsBeamforming microphones, voice detection for "Hey Siri"AirPods (2nd generation), AirPods Pro (1st generation), AirPods Max, Powerbeats Pro2019
H25.3Up to 2x improved ANC, Personalized Spatial Audio with dynamic head tracking, Adaptive Transparency, computational audio processingSkin-detect sensor, enhanced voice isolation and enhancementAirPods Pro (2nd generation), AirPods 4 (with Active Noise Cancellation), AirPods Pro (3rd generation)2022
Over successive generations, these processors have trended toward greater integration of AI-driven features, transitioning from the W1's emphasis on simple Bluetooth pairing and battery optimization to the H2's AI-enhanced spatial audio and adaptive noise control, which leverage on-device machine learning for real-time environmental adaptation. This evolution reflects Apple's focus on reducing reliance on host device processing, enabling features like low-latency spatial rendering directly in the earbuds. In terms of performance metrics, the H1 chip reduced audio latency to approximately 144 ms in AirPods Pro compared to higher delays in W1-equipped models, improving synchronization for video and calls. The H2 further lowers this to an average of 126 ms, supporting ultra-low latency modes for gaming and immersive experiences while maintaining haptic feedback alignment. Battery impact has also improved; the H2's 7 nm process node enables up to 30% longer playback with ANC enabled (6 hours versus 4.5 hours on H1 models), due to efficient Bluetooth Low Energy handling and optimized codec processing. These advancements minimize power draw during continuous use, with the H2 extending total case-charged listening time to 30 hours.

Motion Coprocessors

Early Motion Coprocessors

The early motion coprocessors represented Apple's initial foray into dedicated hardware for handling inertial measurement unit (IMU) data, offloading processing from the main A-series system-on-a-chip (SoC) to enable efficient, always-on motion sensing in mobile devices. These discrete chips focused on collecting and fusing data from accelerometers, gyroscopes, and other sensors to support fitness and health features, such as step counting and activity recognition, while minimizing power draw on the primary processor. Introduced in September 2013 with the iPhone 5s, iPad Air, and iPad mini 2, the M7 coprocessor was the first in the series, integrated alongside the A7 SoC but as a separate die. It continuously gathers motion data from the device's three-axis accelerometer, three-axis gyroscope, and digital compass, even when the main processor is idle, to enable background fitness tracking without excessive battery drain. The M7, manufactured by NXP Semiconductors on an ARM Cortex-M3 core clocked at approximately 180 MHz, processes raw sensor inputs to detect activities like walking or running, providing developers access via the Core Motion framework for applications such as pedometers. This offloading reduces the workload on the A7, allowing the coprocessor to operate independently and conserve energy during prolonged use. The M8 coprocessor succeeded the M7 in September 2014, debuting with the iPhone 6, iPhone 6 Plus, iPad Air 2, and sixth-generation iPod touch alongside the A8 SoC. Building on its predecessor, the M8 added support for a barometer sensor to measure atmospheric pressure, enabling detection of altitude changes and relative elevation for more precise fitness metrics, such as flights of stairs climbed. This expanded the M8's capabilities to nine-axis sensor fusion—combining the three axes each from the accelerometer, gyroscope, and compass with barometric data—for improved accuracy in motion classification and environmental context awareness. Like the M7, the M8 runs on a low-power ARM-based microcontroller, ensuring always-on operation for step tracking and other passive monitoring tasks with negligible impact on overall device battery life. These initial discrete motion coprocessors evolved from basic accelerometer and gyroscope handling in earlier iOS devices, where sensor processing burdened the main CPU, to sophisticated, dedicated IMU processing that prioritized power efficiency and real-time data availability. Operating at sub-milliwatt levels during active sensing, they exemplified Apple's emphasis on specialized silicon for user-centric features like health monitoring, setting the stage for further refinements in subsequent generations.

Integrated Motion Processors in S-series

Starting with the Apple Watch Series 3 in 2017, the S-series system-in-package (SiP) integrated motion processing capabilities equivalent to the earlier standalone M-series coprocessors, fusing sensor data handling directly into the main SiP for enhanced efficiency in wearables. This integration eliminated discrete motion chips seen in prior Watch models, allowing seamless collection and processing of data from onboard sensors even when the device is in low-power states. Key features of these integrated motion processors include wrist detection, which uses accelerometer and gyroscope data to automatically raise the display and enable interactions upon wrist lift. Fall detection, introduced with the S4 SiP in 2018, employs a high-g accelerometer (up to 256 g-forces) alongside the gyroscope to identify hard falls and trigger alarms or emergency calls if the user remains immobile. Crash detection, added starting with the S8 in 2022, extends this by analyzing sudden high-impact events from vehicle collisions using the same sensor suite to initiate emergency responses. These processors also support low-g accelerometer readings for precise workout tracking, such as step counting and elevation changes, while synchronizing haptic feedback with motion events to provide tactile cues during activities like notifications or timers. The design emphasizes power efficiency, enabling always-on motion sensing without significantly draining the battery, as the integrated coprocessor handles sensor fusion and preliminary processing independently of the main CPU. This allows continuous monitoring for health and safety features throughout the day on a single charge. Updates in later models, such as the S9 SiP from 2023, S10 SiP from 2024, and S11 SiP from 2025, incorporate machine learning via a 4-core Neural Engine to enable advanced gesture recognition, including the double-tap gesture for one-handed control of apps, scrolling, and calls by detecting precise finger movements, as well as motion-based health features like sleep apnea detection using accelerometer and gyroscope data for breathing disturbance analysis. These integrated motion processors are featured in all Apple Watch Series 3 and later models, including SE generations, providing consistent motion handling across the lineup.

Comparison of motion coprocessors

Apple's M-series motion coprocessors represent a progression in dedicated hardware for handling sensor data, transitioning from standalone chips that offload processing from the main CPU to embedded components within A-series SoCs for greater efficiency. This evolution has allowed for continuous motion sensing without significantly impacting battery life, supporting features like fitness tracking and contextual awareness. Early models focused on core sensor integration, while later versions added support for additional sensors and always-on capabilities.
CoprocessorIntegrationSensors SupportedKey Algorithms (Fall/Crash)PowerDebut Chip/Device
M7DiscreteAccelerometer, gyroscope, compassBasic sensor fusion for motion dataLow-power ARM-based core, offloads A7 for minimal battery drainA7 in iPhone 5s (2013)
M8DiscreteAccelerometer, gyroscope, compass, barometerEnhanced fusion including altitude changesImproved efficiency over M7, supports continuous tracking at low drawA8 in iPhone 6 (2014)
M9Embedded in SoCAccelerometer, gyroscope, compass, barometerMotion processing for always-on features like "Hey Siri"Integrated design reduces overall power vs. discrete, enables 24/7 sensingA9 in iPhone 6s (2015)
M10Embedded in SoCAccelerometer, gyroscope, compass, barometerAdvanced pedometer and motion classificationFurther optimized for efficiency, supports background fitness appsA10 Fusion in iPhone 7 (2016)
M11Embedded in SoCAccelerometer, gyroscope, compass, barometerRefined sensor fusion for precise trackingLow draw enables extended use without waking main CPUA11 Bionic in iPhone X (2017)
Over successive generations, M-series coprocessors advanced from processing 3-axis accelerometer data alone to comprehensive 9-axis sensor fusion, combining inputs from the accelerometer (3 axes), gyroscope (3 axes), and compass (3 axes) for accurate orientation and movement detection. With the integration of the barometer starting in the M8, elevation changes were incorporated, enhancing applications like step counting and floor detection. Later developments in integrated designs, such as those in the S9 SiP for Apple Watch, introduced AI-driven algorithms for gesture recognition, like double-tap controls, processed on low-power neural engines. The low-power architecture of these coprocessors has been pivotal in enabling prolonged device usage; for instance, they allow the Apple Watch to deliver up to 18 hours of all-day battery life while performing continuous motion tracking for health and fitness metrics. This offloading prevents the main processor from activating unnecessarily, preserving energy for essential tasks and contributing to the overall battery efficiency in iPhones and iPads as well.

Specialized Components

Apple R1

The Apple R1 is a custom system-on-a-chip (SoC) developed by Apple Inc. specifically as a coprocessor for spatial computing in mixed-reality devices. Announced on June 5, 2023, during the introduction of the Apple Vision Pro at Apple's Worldwide Developers Conference, the R1 is optimized to handle real-time sensor fusion and image processing, enabling immersive experiences with minimal perceptible delay. It represents Apple's first dedicated silicon for mixed-reality headsets, focusing on offloading specialized tasks from the main application processor to ensure seamless integration of digital content with the physical world. In the Apple Vision Pro, the R1 operates alongside an Apple M-series chip—initially the M2 and later the M5 in updated models—for a dual-chip architecture that divides responsibilities efficiently. The R1 dedicates its resources to processing inputs from the device's extensive sensor array, including 12 cameras (such as the TrueDepth system and LiDAR scanner), five sensors (including inertial measurement units), and six microphones. This setup supports advanced features like precise eye and hand tracking, spatial audio, and real-time 3D environment mapping, which are essential for visionOS, Apple's operating system for spatial computing. By handling these tasks independently, the R1 reduces computational load on the M-series chip, allowing it to focus on general-purpose computing and graphics rendering. A hallmark of the R1 is its 12-millisecond photon-to-photon latency, which measures the time from capturing light via sensors to rendering it on the displays, creating a responsive experience critical for mixed reality. This low latency is achieved through dedicated hardware pipelines for sensor data fusion and display output, supporting the Vision Pro's dual micro-OLED displays with resolutions exceeding 4K per eye and refresh rates up to 100 Hz. The chip also features a high memory bandwidth of 256 GB/s, facilitating rapid data transfer for high-fidelity video passthrough and spatial content generation. These capabilities enable the Vision Pro to deliver fluid interactions, such as gesture-based controls and immersive video playback, without compromising on power efficiency or thermal performance in a wearable form factor. The R1 is exclusively used in the Apple Vision Pro, which became available in the United States on February 2, 2024, starting at a price of $3,499. It is fabricated using advanced packaging technology from TSMC, including the integrated fan-out multi-chip (InFO-M) process, which enhances integration density and performance for compact devices like headsets. This design choice underscores Apple's emphasis on custom silicon to push the boundaries of spatial computing, prioritizing real-time processing over traditional computing paradigms.

Apple C1

The Apple C1 is Apple's first in-house designed cellular modem, marking a significant step toward reducing reliance on third-party suppliers like Qualcomm for mobile connectivity components. Development of the C1 stemmed from Apple's 2019 acquisition of the majority of Intel's smartphone modem business for $1 billion, which included approximately 2,200 employees, intellectual property, and ongoing projects aimed at 5G technology. This deal, announced on July 25, 2019, and completed in December 2019, provided Apple with foundational expertise and assets to build its own modem silicon, following years of licensing Qualcomm modems amid legal disputes over patent royalties. The acquisition was part of Apple's broader strategy to control key hardware elements in its devices, similar to its custom silicon efforts in processors and other chips. Apple unveiled the C1 modem on February 19, 2025, alongside the introduction of the iPhone 16e, a $599 entry-level model in the iPhone 16 lineup. This debut represented the company's first production use of a fully custom cellular modem, shifting away from Qualcomm's Snapdragon X-series chips that had powered iPhones since the 5G era began in 2019. The C1 is integrated directly with Apple's A-series application processors, enabling tighter optimization for power management and performance within the iPhone's system-on-chip architecture. Initial deployment focused on the iPhone 16e, with Apple positioning the modem as a foundational technology for future devices, though it emphasized efficiency over peak speeds in its rollout. In September 2025, Apple introduced the C1X, a successor modem up to 2x faster than the C1, debuting in the iPhone Air. Technically, the C1 supports 5G connectivity via sub-6 GHz bands with 4x4 MIMO, alongside Gigabit LTE fallback using the same MIMO configuration, but it does not include mmWave 5G support to prioritize global compatibility and battery life. It enables 3x downlink carrier aggregation on sub-6 GHz 5G, aggregating up to 160 MHz of bandwidth for reliable mid-band performance, and includes features like Wi-Fi 6 (802.11ax) with 2x2 MIMO and Bluetooth 5.3 for comprehensive wireless integration. Apple highlighted the modem's design for low-latency 5G connections, contributing to seamless streaming and data transfer in everyday use, though it lacks advanced capabilities like 5x carrier aggregation or VoNR found in premium Qualcomm modems. In terms of performance, the C1 delivers comparable download and upload speeds to the Qualcomm Snapdragon X71 used in higher-end iPhone 16 models—typically around 200-500 Mbps on sub-6 GHz networks depending on carrier infrastructure—but excels in power efficiency. Apple claims the C1 is the most power-efficient cellular modem ever integrated into an iPhone, achieving up to 25% lower power consumption during 5G operation compared to prior Qualcomm solutions, which translates to extended battery life such as 26 hours of video playback on the iPhone 16e. Real-world tests confirm this advantage, showing the C1 maintaining consistent connectivity in varied network conditions while reducing energy draw, particularly in low-signal areas where it outperforms Qualcomm modems by minimizing retransmissions and optimizing radio usage. This efficiency focus aligns with Apple's silicon philosophy, prioritizing sustained performance over raw throughput benchmarks.

Early and Miscellaneous Processors

Apple's engagement with mobile processors predated its custom silicon era, relying on licensed ARM-based system-on-chips from third-party manufacturers for devices like the iPod and early iPhones. The inaugural iPod, launched in 2001, incorporated the PortalPlayer PP5002 SoC, which featured dual ARM7TDMI cores operating at low clock speeds to manage audio playback and basic interface functions on a 130 nm process. This chip exemplified the power-efficient ARM architecture that Apple adopted for portable media players, enabling the device's signature 1,000-song capacity while consuming minimal battery. Subsequent generations of iPods, spanning 2001 to 2008, utilized similar licensed designs, often fabricated on 90 nm processes, to balance performance and portability without custom intervention from Apple. A notable example among these early licensed processors is the Samsung S5L8900 SoC, deployed in the iPhone 3G (2008) and iPhone 3GS (2009). Built on a 90 nm process for the 3G variant with a 412 MHz ARM1176JZF-S (ARMv6) core and PowerVR MBX Lite GPU, it supported 128 MB of RAM and handled 3G connectivity alongside basic computing tasks. The 3GS iteration upgraded to a 65 nm process and 600 MHz clock speed via the closely related S5L8920, doubling performance for improved multitasking and graphics rendering while maintaining compatibility with iPhone OS 3.0. These chips, manufactured by Samsung under Apple's specifications, represented a transitional phase where Apple optimized off-the-shelf ARM designs for its ecosystem, including integrated baseband support for cellular models. Miscellaneous early processors appeared in compact devices like the iPod nano series. The second-generation iPod nano (2006) employed the Samsung S5L8701 SoC, featuring an ARM940T core (ARMv4 architecture) at an undisclosed clock speed, alongside controllers for flash storage, LCD display, and USB connectivity on a 90 nm process. This design prioritized simplicity and low power for the device's slim form factor, handling video playback and navigation with 8 to 16 GB storage options. Similar S1P-like system-in-package (SiP) configurations emerged in later iPod shuffles and nanos, such as the S5L8720 in the fourth-generation iPod shuffle (2010), which integrated a 32-bit ARM1176JZF-S core with 32 to 64 MB DRAM for audio-only functionality. These obscure components, often lacking dedicated GPUs, underscored Apple's experimentation with minimalistic silicon for non-flagship products. In media streaming hardware, early Apple TV models incorporated licensed ARM processors as decoders. The second-generation Apple TV (2010) used a Marvell Kirkwood ARM-based SoC for video processing, while the third-generation model (2012) shifted to Apple's A5 variant for 1080p decoding, though no dedicated A5X decoder was deployed in production units—rumors of such integration in a 2013 revision proved unfounded, with the device retaining a single-core A5 on a 32 nm process. These transitional chips focused on efficient H.264 decoding without advanced graphics, reflecting Apple's gradual move toward unified architectures. Early efforts in specialized components included discontinued motion processing integrations and prototypes. Prior to dedicated coprocessors, motion sensing relied on main SoCs like the A6 and A7, with internal designations such as M0 representing nascent, non-production designs that were phased out in favor of the M7 by 2013. Similarly, early C1 prototypes for cellular modems, developed in the mid-2020s, faced issues like overheating and suboptimal speeds, remaining historical artifacts without ongoing support as Apple iterated toward production viability. By 2010, Apple shifted from these licensed dependencies to fully custom silicon, debuting the A4 SoC in the first-generation iPad—a single-core ARM Cortex-A8 design fabricated by Samsung on a 45 nm process, integrating CPU, GPU, and memory controllers for optimized iOS performance. This transition eliminated reliance on external vendors for core logic, enabling tighter hardware-software integration, though early licensed processors like the S5L series laid essential groundwork and receive no further software updates today.

References

  1. https://en.wikichip.org/wiki/apple/ax/a7
  2. https://en.wikichip.org/wiki/apple/ax/a12x
  3. https://en.wikichip.org/wiki/apple/ax/a12z
Add your contribution
Related Hubs
User Avatar
No comments yet.