RISC-V Unified Database: Streamlining the Ecosystem with a Centralized Source of Truth

The RISC-V ecosystem currently relies on multiple disconnected repositories and specifications, including the assembly manual, opcode database, formal specifications like Sail and the non-machine readable ISA manual. This fragmentation of information is error prone and makes it hard for hardware designers, researchers, and tool developers to efficiently find, modify, and verify essential data. To address these challenges, we introduce the RISC-V Unified Database (UDB), a single, machine-readable source that consolidates and cross-validates RISC-V specifications against established resources like binutils and riscv-opcodes. Built through collaboration between engineers at industry-leading companies, the UDB already generates version-specific ISA manuals, instruction indexes, and standardized opcode definitions. Ongoing work extends to generating simulator configurations for QEMU and other instruction set simulators. The UDB's YAML-based format ensures broad compatibility and positions it as a foundation for next-generation RISC-V tools. This approach is starting to gain traction with official RISC-V working groups, including the Certification Steering Committee, which already uses the UDB for generating certification requirements. This talk explores UDB’s architecture, showcases its use cases, and invites you to contribute to this transformative project for the RISC-V community.