Information technology — Artificial intelligence — Overview of machine learning computing devices (Adopted ISO/IEC TR 17903:2024, first edition, 2024-05)

This document provides a high-level overview of computing devices designed specifically for machine learning within the field of information technology. Adopted from an international technical report, it outlines key architectural components, functional requirements, and operational characteristics essential for artificial intelligence workloads. The text addresses hardware interfaces, data processing pipelines, and energy efficiency considerations unique to modern inference and training systems. It serves as a reference for manufacturers and engineers involved in the design and implementation of specialized AI accelerators. The content details performance metrics and compatibility standards without prescribing mandatory specifications. By establishing common terminology and conceptual frameworks, this technical report facilitates interoperability and consistent evaluation across different platforms. It covers both cloud-based and edge computing environments, highlighting emerging trends in neural network processing. The document aims to clarify the distinctions between general-purpose processors and dedicated machine learning hardware. It does not cover software algorithms but focuses strictly on the physical and logical structure of the computing infrastructure supporting artificial intelligence applications globally.