Version 0.3

Kernel Float is a header-only library that simplifies working with vector types and reduced precision floating-point arithmetic in GPU code.

This release introduces support for HIP and enables Kernel Float to compile for AMD GPUs. Additionally, there are several smaller bug fixes and internal changes that were found during testing.

Version 0.2

Kernel Float is a header-only library for CUDA that simplifies working with vector types and reduced precision floating-point arithmetic in GPU code.

This release renames several the operations in the library and refactors many of its internals. Most notably, all operations that cannot be performed in 16-bit precision will now be performed in single precision and downcast back to lower precision.

Full Changelog: v0.1...v0.2

Version 0.1

This release marks the initial release of Kernel Float.

Kernel Float is a header-only library for CUDA that simplifies working with vector types and reduced precision floating-point arithmetic in GPU code.