Enhancing Development Productivity Based on AMD Knowledge Base and Agent Skills

AMD has unveiled agentic AI assistant 'AMD Ross,' which supports the entire design and development process of embedded systems.

AMD recently announced through its blog that AMD Ross is designed to support the entire embedded development lifecycle, from hardware design through software development, AI model optimization, debugging, to system deployment.

AMD Ross performs design document searches, tool control, debugging support, and workflow automation through natural language-based interaction. It also leverages AMD-validated technical documentation and expert-built development methodologies to support the design and development process.

Supporting Hardware to AI Development

AMD Ross supports various tasks in the embedded development environment, including hardware and software partitioning, high-level synthesis (HLS) based design, silicon optimization, embedded application development, and machine learning algorithm optimization.

It is also designed to be utilized for hardware development tasks such as system schematic review, board layout optimization, power analysis, and low-power design.

AMD explained that when users describe their development intent in natural language, the system can search relevant documentation, perform code generation, design improvements, and error analysis.

Utilizing AMD Knowledge Base and Agent Skills

AMD Ross operates based on four key components: △MCP (Model Context Protocol) servers connecting to AMD embedded tools △knowledge base built on AMD-validated documentation △agent skills containing expert development workflows △design examples.

Through this, it supports automation of repetitive engineering tasks and enables reuse of expert knowledge across the organization.

AMD anticipates benefits including shortened onboarding periods for new engineers, improved development productivity, reduced debugging and optimization time, and standardization of development practices across teams.

Expanding Embedded AI Development Environment

AMD Ross is designed to be utilized across various IDEs and command-line environments without being tied to any specific large language model (LLM) or development tool.

The company announced plans to sequentially provide additional embedded development tools and workflow features to expand the scope of support.

AMD expects this platform to present a new embedded development environment that provides integrated support for hardware, software, and edge AI development.