AMD XDNA

XDNA is the name for AMD's neural processing unit microarchitecture. It is based on IP blocks from Xilinx, a company which was acquired by AMD on February 14 2022.^[1][2]

AMD XDNA

Design firm	AMD
Introduced	April 2023
Type	Neural processing unit microarchitecture

The underlying technology in XDNA is the AI Engine-ML, shortened to AIE-ML. The predecessor to this technology and original Xilinx IP is the AI Engine (AIE).^[3] As of 2025, XDNA is implemented in AMD's consumer PC processors (branded as Ryzen AI) and their FPGAs (specific versions of Versal and Alveo series). The original AIE is still sold in some Versal boards.

Generations

First

First-generation XDNA, initially implemented in the Ryzen 7040 series mobile processors, provides up to 10 TOPS of processing performance. The Ryzen 8040 series (codenamed "Hawk Point"), a refresh of the Ryzen 7040 series, features a higher-clocked XDNA NPU providing 16 TOPS of performance.^[4]

XDNA is also used in AMD's Alveo V70 datacenter AI inference processing card.^[5]

Second

XDNA 2 was introduced in the Strix Point Ryzen AI 300 series processors. The NPU provides up to 50 TOPS of processing power,^[6][7] and the implementation is labelled by AMD as the third generation of Ryzen AI.

See also

• AI engine
• Neural Engine, Apple's neural processing unit microarchitecture
• Tensor Processing Unit, Google's AI acceleration microarchitecture