AMD Ryzen™ Processors
AMD Ryzen 7040 and 8040 Series processors are built on AMD XDNA NPUs and AMD “Zen” cores to deliver leadership AI acceleration on PCs.
Introduction
As neural networks become larger and more complex, the demand for higher compute density with lower power continues to grow. Discover how AMD XDNA™ NPU architecture addresses this imperative.
Image Zoom
AMD XDNA - AI Engine
AMD XDNA is a spatial dataflow NPU architecture consisting of a tiled array of AI Engine processors. Each AI Engine tile includes a vector processor, a scalar processor, and local data and program memories. Unlike traditional architectures that require repeatedly fetching data from caches (which consumes energy), AI Engine uses on-chip memories and custom dataflow, to enable efficient, low power computing for AI and signal processing.
Inside the Tile
Each AI Engine tile consists of a VLIW (Very Long Instruction Word), SIMD (Single Instruction Multiple Data) vector processor optimized for machine learning and advanced signal processing applications. The AI Engine processor can run up over 1.3GHz, enabling efficient, high throughput, low latency functions. Each tile also contains program and local memory to store data, weights, activations, and coefficients; a RISC scalar processor and different modes of interconnect to handle different types of data communication.
Image Zoom
AMD XDNA 2
Next-Generation AMD XDNA 2 is built for Generative AI experiences in PCs, delivering exceptional compute performance, bandwidth, and energy efficiency.
Benefits
Software Programmable
The AMD NPU is programmable and compiles in minutes. It also leverages a library-based design to simplify the workflow for ML framework developers.
Deterministic
The AMD NPU includes dedicated instruction and data memories, as well as dedicated connectivity paired with DMA engines for scheduled data movement using connectivity between AI Engine tiles.
Efficient
The AMD NPU delivers greater compute density compared with traditional architectures and drives exceptional power efficiency.
Scalable
The AMD NPU is architected as 2D arrays consisting of multiple AI Engine tiles, enabling scaling from 10s to 100s of AI Engine tiles in a single device, servicing the compute needs of a breadth of applications.
Related Products
