Introduction of new thread levels for improved parallelization.
Enhanced performance through new parallel regions (e.g., ML Lengths, ML splits, LogLk, etc.).
Threads used in tree creation: Top hits, TopHitNJSearch, FastNJSearch, and ExhaustiveNJSearch (-slow).
Implementation of a faster tree partitioning approach with significant speed improvements.
Tree partitioning limited to NNI, SPR, and upProfiles computations for memory conservation:
Parallel tree traversal implemented for remaining parts.
Replacement of disk storage for profiles with Disk Computing.
Shared and reused Top upProfiles among threads for memory efficiency and accelerated sequential parts.
Improved non-deterministic mode with removal of mutex usage.
Optimized performance by parallelizing non-deterministic parts in deterministic mode:
Also implemented non-deterministic parts in deterministic mode for improved performance.
Deterministic mode now outperforms non-deterministic mode in terms of speed.
Tree partitioning method logging now hidden by default.
Support for Fastq format and libBZ2 compression.
Support for reading trees from NEXUS block trees.
Nvidia CUDA GPU computing support (experimental)
Introduced parallel compilation.
Incorporation of changes from FastTree-2.11.
Clang Support
Addressed critical errors and implemented substantial corrections.