pynbody.array.shared

Support for numpy arrays in shared memory.

See also

There is information about using shared arrays to create parallel workflows in Writing your own parallel code.

Functions

delete_dangling_shared_memory()	Ensures that all shared memory has been cleaned up.
get_num_shared_arrays_owned()	Returns the number of shared arrays currently owned by this process.
make_shared_array(dims, dtype[, zeros, ...])	Create or reconstruct an array of dimensions dims with the given numpy dtype, backed on shared memory.
pack(array[, transfer_ownership])	Turn an array backed onto shared memory into something that can be passed between processes
remote_map(pool, fn, *iterables)	Equivalent to pool.map, but turns any shared memory arrays into a reference that can be passed between processes.
shared_array_remote(fn)	A decorator for functions that are expected to run on a 'remote' process, accepting shared memory inputs.
unpack(array_description)	Reconstruct an array backed onto shared memory from a deconstructed array (returned by pack()).

Classes

SharedMemorySimArray(data[, units, sim])

A simulation array that is backed onto shared memory.

Exceptions

RemoteKeyboardInterrupt
SharedArrayNotFound

Modules

posix_detail

POSIX-specific shared memory implementation.