Apple Neural Engine

E38961

Apple Neural Engine is Apple’s dedicated on-chip hardware accelerator designed to efficiently perform machine learning and AI computations on its devices.

Aliases (1)

Statements (53)
Predicate Object
instanceOf hardware accelerator
neural processing unit
alsoKnownAs ANE
architectureType parallel processing architecture
designedFor artificial intelligence computations
machine learning workloads
on-device inference
developer Apple Inc.
enables on-device privacy-preserving ML
evolvesWith each new Apple SoC generation
firstReleaseYear 2017
hasCapability multiple TOPS throughput (tera operations per second)
includedInChipFamily Apple A-series
Apple M-series
Apple S-series
Apple T-series
integratedWith Apple CPU cores
Apple GPU
introducedInProduct iPhone 8
iPhone X
introducedWithChip Apple A11 Bionic
optimizationGoal energy efficiency
high performance
low power consumption
partOf Apple system on a chip
performs convolution operations
matrix multiplications
neural network inference
programmedVia Core ML API
Metal Performance Shaders
supports Animoji
Core ML
Face ID
augmented reality workloads
camera computational photography
image recognition
natural language processing
on-device Siri processing
speech recognition
targetPlatform iOS
iPadOS
macOS
tvOS
visionOS
watchOS
usedBy third-party iOS apps
third-party macOS apps
usedIn Apple TV
Apple Vision Pro
Apple Watch
Mac
iPad
iPhone

Referenced by (11)
Subject (surface form when different) Predicate
Apple A16 Bionic
Apple M1
Apple M1 Max
Apple M2 Max
neuralEngineType
Apple M-series ("Neural Engine")
Apple M2 Pro ("Neural Engine")
feature
Apple M2 Max ("Neural Engine")
Apple M4 ("Neural Engine")
hasComponent
Apple A15 Bionic ("Neural Engine")
integrates
Apple M1 Ultra ("Neural Engine")
integratesComponent
Apple M2 Ultra ("Neural Engine")
supportsFeature

Please wait…