amd launches ross agentic ai embedded design 1 amd launches ross agentic ai embedded design 1

AMD Launches Ross Agentic AI Assistant for Embedded Design

AMD Ross lets developers use natural language to drive AMD Embedded tools, from hardware debug to deployment.

AMD has introduced AMD Ross, an agentic AI assistant for embedded development. It covers the full design flow, including architecture, optimization, debugging, PCB design, schematic review, software, AI, and system deployment. Developers can interact with their tools in natural language and apply expert-written agent skills, reusable workflows, and a grounded AMD Knowledge Base along the way.

The pitch is aimed at AMD Embedded’s broad lineup, which spans FPGAs, adaptive SoCs, x86 embedded processors, and edge AI platforms. As designs get more complex, AMD says capturing expert methods as reusable agent skills makes that knowledge easier to share and helps teams execute more consistently.

Salil Raje, senior vice president and general manager of AMD Embedded, said Ross brings AMD Embedded tools, trusted knowledge, and expert-authored workflows into a single agentic experience, with the goal of getting products from design intent to deployment faster.

What it does

Using natural language, developers can have AI agents search documentation, check tool status, run commands, guide debugging, and execute proven workflows. Across the AMD Embedded toolset, Ross can be used to:

  • Perform hardware and software partitioning
  • Implement hardware designs with high-level synthesis tools
  • Optimize silicon design and debug
  • Develop embedded software applications and algorithms
  • Optimize machine learning algorithms with AMD embedded AI software
  • Estimate power and optimize for lower consumption
  • Review system schematics and optimize board layout

Ross is client-agnostic. Teams can keep using their preferred LLMs, IDEs, and command-line environments while staying connected to AMD Embedded tools.

amd launches ross agentic ai embedded design 2

Four components

AMD positions Ross as more than a coding assistant layered over development tools. It is built from four parts:

  • MCP servers: Based on the open Model Context Protocol standard, these connect AI agents to AMD Embedded tools so they can query information, run commands, and work inside the design environment.
  • AMD Knowledge Base: AMD-validated vectorized databases containing user guides, product guides, white papers, application notes, and answer records. They can be accessed through the cloud or run locally offline.
  • Agent skills: Open, expert-authored markdown files that capture repeatable best practices. They can walk an LLM through structured tasks such as timing optimization or restructuring a C++ design for better performance in Vitis HLS.
  • Design examples: Ready-to-run designs that show how agent skills apply to real embedded applications.

Workflow examples

AMD lists several use cases for Ross:

  • Turning natural-language intent into actions such as searching documentation, running tool commands, generating code, and iterating on designs
  • Explaining tool errors and recommending fixes
  • Analyzing timing issues with AMD timing methodologies to find violations, explain root causes, and suggest improvements
  • Optimizing Vitis HLS designs through pipelining, loop optimization, pragma insertion, and architecture refinement
  • Speeding up hardware debug with guided signal capture, debug-core configuration, interface analysis, and root-cause identification

AMD says the expected benefits are faster time-to-prototype, shorter debug and optimization cycles, less repetitive tool interaction, quicker onboarding of new engineers, and more consistent use of best practices across teams.

iWave Global, an early user, offered feedback on the tool. Geetha Govindaraj, associate director of the FPGA SOM BU, said Ross has helped the team identify and troubleshoot issues more efficiently, cutting the time and effort spent on debugging and bring-up.

Availability

AMD Ross is available now at AMD.com/Ross. AMD plans to add more AMD Embedded tools and workflow capabilities to Ross on a monthly cadence.

Leave a Reply

Your email address will not be published. Required fields are marked *