An auditable migration assistant for AI projects that need to run without an NVIDIA CUDA dependency. It retains the existing CUDA-to-HIP converter and adds a standalone project scanner and runtime planner for ROCm/HIP, DirectML, ONNX Runtime DirectML, MIGraphX, and CPU fallback
-
Updated
Sep 16, 2026 - Python