- Support for building with CMake 4.0.
- Fixed a potential integer overflow issue in
hipMallocinterfaces.
- Added documentation clarifying how hipfort is built for the NVIDIA platform. Thanks @fluidnumerics-joe!
- Updated and reorganized documentation for clarity and consistency.
- The hipfc compiler wrapper has been deprecated and will be removed in a future release. Users are encouraged to directly invoke their Fortran or HIP compilers as appropriate for each source file.
- Support for building with LLVM Flang
- Fixed the exported
hipfort::hipsparseCMake target
- Added roctx to the hipfort interfaces
- Updated the hipsolver bindings
- Included hipfort-config.cmake in the deb and rpm packages
- Added an exported hipfort-config.cmake with the following targets:
hipfort::hiphipfort::rocblashipfort::hipblashipfort::rocffthipfort::hipffthipfort::rocsolverhipfort::hipsolverhipfort::rocrandhipfort::hiprandhipfort::rocsparsehipfort::hipsparse
- Added
rocm_agent_enumeratorfallback for hipfc architecture autodetection
- Updated documentation to use the Sphinx toolchain and publish to ReadTheDocs
- Updated
HIP_PLATFORMfrom 'nvcc' to 'nvidia'
- Added hipfc architecture autodetection for gx1101 devices
- Fixed hipfc architecture autodetection for gfx90a devices that were previously unrecognized