- Support for Mojo v0.25.6
- Improved benchmarking pipeline via
pixi run bench-all-cpuandpixi run bench-all-gpu
- Revert to lockfile from v1.2.1 to workaround modular-community package
regexand conda channel ordering.
- Switch to using signed ints instead of unsigned for semi-global
- Fix alignment score check used to decided whether or not to do the reverse alignment for semi-global and local alignment
- Allow specifying simd width via a flag at comple time
- Fix / updates the bench_aligner for current stable Mojo
- Fixed bug in logging module related to update from Mojo 24.3 to 24.4 replacing
write_args.
- Upgrade to Mojo 25.4
- Fix issue with AVX512 via PR
- Apply suggestion from @soraros to use llvm intrinsic for saturating add and subtract
- Added a compile time
ISH_SIMD_TARGETflag that can be set tobaselineto forceishto use SSE sized widths.
- Initial public release
- See https://github.com/BioRadOpenSource/ish/compare/v1.0.0...v1.1.0