Cirkitly

What is Cirkitly Spec2RTL?

Cirkitly Spec2RTL is an AI-powered tool that helps hardware designers create and verify Verilog RTL code and SystemVerilog testbenches. It takes high-level specifications written in natural language and turns them into ready-to-use code. This tool is designed to automate the tedious parts of hardware design, making the process faster and more efficient.

Benefits

• Automated Code Generation:Spec2RTL transforms high-level specifications into verified Verilog RTL code and testbenches, saving time and reducing manual effort.
• Self-Correction Loop:The tool automatically checks for syntax errors and logical flaws, then fixes them before presenting the final code for human approval.
• Knowledge-Aware Generation:It uses a custom knowledge base to ensure the generated code complies with specific rules and coding standards.
• Human-in-the-Loop:While the AI handles most of the work, the final approval is always left to the human designer, ensuring quality and control.

Use Cases

• Hardware Design:Spec2RTL is ideal for hardware designers who need to quickly generate and verify RTL code from specifications.
• Educational Purposes:It can be used in academic settings to teach students about hardware design and verification.
• Prototyping:Engineers can use it to rapidly prototype hardware designs before moving to more detailed and complex stages.

Getting Started

To get started with Spec2RTL, you need Python 3.10 or newer, a C compiler likegcc, Icarus Verilog for code validation, and access to an AI model either locally via Ollama or through the cloud via Azure OpenAI. Follow the installation and setup instructions provided in theGitHub repository.

How to Use Spec2RTL

1. Add Your Specifications:Place your hardware module specifications as.mdfiles inside thespecs/directory.
2. Add to the Knowledge Base (Optional):Include relevant documentation, coding standards, or compatibility rules as.txtfiles in theknowledge_source/directory.
3. Run the Copilot:Execute the main script to start the AI-powered code generation process.
4. Collaborate with the AI:Select a spec, watch the AI create a plan, generate the code, and run its validation and self-critique loops. Approve the final code for saving.
5. Run Your New Simulation:Navigate to the output directory and run the simulation script to see the results.

Additional Information

Spec2RTL is part of the broader AI-Powered Hardware Design & Verification Copilot project, aimed at revolutionizing the hardware design process with advanced AI technologies.