Playground · CUDA porting checker

How much of your CUDA runs on GRXCP?

Paste a CUDA file. Every CUDA name in it is checked against GRXCP's compatibility header and its mapping and gap register. Your code stays in your browser.
your_kernel.cuExamples:

…

    To port it

    1. Replace #include <cuda_runtime.h> with #include <grx/grx_cuda_compat.h>.
    2. Compile with grxcc instead of nvcc.
    3. Work through the items flagged below.

    Data from kubeworkz/grxcp@100179e (2026-10-29): grx_cuda_compat.h, the device headers and cuda_mapping.md. GRXCP is in active development, so this changes as it does. GRXCP on GitHub ↗