[ogrisel]

The IPC overhead of process-based parallelism in Python is a pain to deal with in general, even when the underlying computational bottleneck are already written CPU optimized (calls to compiled extensions written in Cython/C/C++/Rust, call to CPU optimized operations written with CPU architecture-specific intrinsics/assembly from OpenBLAS via NumPy/SciPy, calls to e.g. GPU CUDA kernels via PyTorch/Triton, ...).

Sometimes the optimal level of parallelism lies in an outer loop written in Python instead of just relying on the parallelism opportunities of the inner calls written using hardware specific native libraries. Free-threading Python makes it possible to choose which level of parallelism is best for a given workload without having to rewrite everything in a low-level programming language.