TIRx: Open Compiler Stack Accelerates Development of Cutting‑Edge ML Kernels

TIRx is introduced as an open‑source compiler stack for machine‑learning kernels. It targets the rapid evolution of frontier ML kernels. The stack builds on the Apache TVM ecosystem. Developers can use TIRx to generate optimized code for diverse hardware. It aims to simplify kernel development and improve performance. The project is positioned to aid research and production workloads. Community contributions are encouraged to extend its capabilities. The release was announced on June 22, 2026.