Releases: whyb/FastChwHwcConverter
Releases · whyb/FastChwHwcConverter
v0.2.8
v0.2.7
- Add c++ std::thread impl
- Disable openmp by default
- Add build-macos-x64
- Add build-macos-arm64
- Add build-linux-arm64
v0.2.6
- In Linux, change the GLIBC link to use static mode.
- On x64-windows platform, package opencv.dll into release zip.
v0.2.5
- Change to use singleton classs implementation.
- Added init() and release() for GPU.
- Update README.md
v0.2.4
- Using cudaStream/hipStream to improve performance.
- Refactor the cuda/rocm module functions to fully utilize parallel processing in each dimension.
- Fixed some known bugs.
v0.2.3
Update release.yml
v0.2.2
Auto download and extract CUDA and ROCm binary dynamic libraries to c…
v0.2.1
- Add FastChwHwcConverterROCm.hpp
- Thanks to @MouriNaruto for reminding me to fix the possible dll injection risk.
v0.1.1
Add FastChwHwcConverterCuda.hpp
v0.1.0
Fix ci check error