PyTorch on RISC-V: From Cross-Compilation To Native CI - Ludovic Henry, Meta

This talk covers the efforts to port PyTorch natively to the RISC-V architecture, which is becoming a significant player in AI and data center workloads. The speaker discusses the challenges of integrating native math library dependencies like OpenBLAS and oneDNN with RISC-V Vector extensions, while also focusing on the need to support the entire Python ecosystem, including vital libraries like NumPy, SciPy, and ONNX. Additionally, the session highlights the importance of establishing a robust CI infrastructure for sustaining long-term performance and reliability in the validation processes for the PyTorch master branch and pull requests.