HIP: C++ Heterogeneous-Compute Interface for Portability

HIPIFY: Convert CUDA to Portable C++ Code

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Gromacs that can be accelerated at AMD GPU

Agentic CUDA-to-ROCm migration loop - translates, compiles, profiles, and iteratively fixes GPU kernels with real hardware feedback.

Autonomous CUDA → ROCm/HIP migration agent — ports CUDA repos until they build and pass tests on AMD GPUs, one reviewable commit per fix, with a mechanical security gate over every AI-written diff. Proven on real AMD hardware (Radeon gfx1100, ROCm 7.2).