Question
Make scipy.sparse functions like csr._mul_matvecrelease GIL?
Context
Python functions that wrap external code (such as C) often call the GIL at runtime, allowing parallelism with multi-threaded. This is often found in the codebase numpy. Is it also common in scipy.sparse? If so, what operations does the GIL issue? If they donβt release the GIL, then the fundamental problem arises, why not or is it just a lack of human power?
source
share