The documentation for the multiprocessing.dummy module says that it "replicates the API of multiprocessing."  In Python 2.7, I can import multiprocessing.dummy and use the cpu_count function:

$ python2
Python 2.7.12 (default, Oct 29 2016, 19:21:06)
[GCC 4.2.1 Compatible Apple LLVM 8.0.0 (clang-800.0.42.1)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import multiprocessing.dummy as mp
>>> mp.cpu_count()
8

But in Python 3.6, multiprocessing.dummy is missing cpu_count:

$ python3
Python 3.6.0 (default, Mar 21 2017, 13:27:21)
[GCC 4.2.1 Compatible Apple LLVM 8.0.0 (clang-800.0.42.1)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import multiprocessing.dummy as mp
>>> mp.cpu_count()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: module 'multiprocessing.dummy' has no attribute 'cpu_count'

I would expect that multiprocessing.dummy would have cpu_count, since that function is available in multiprocessing, and it's there in Python 2.7.

It looks like it was removed in commit 04842a8, "Remove unused or redundant imports in concurrent.futures and multiprocessing" (Florent Xicluna 5 years ago).  Maybe the removal was inadvertent?

I realize there are several workarounds, but based on the documentation and the behavior of previous versions, I wouldn't have expected this breaking change.

multiprocessing.dummy lacks many symbols:

>>> set(multiprocessing.__all__) - set(multiprocessing.dummy.__all__)
{'RawArray', 'set_forkserver_preload', 'ProcessError', 'RawValue', 'BufferTooShort', 'reducer', 'set_executable', 'Array', 'cpu_count', 'get_all_start_methods', 'get_start_method', 'AuthenticationError', 'Value', 'set_start_method', 'allow_connection_pickling', 'get_logger', 'SimpleQueue', 'TimeoutError', 'log_to_stderr', 'get_context'}