An emulator for the hierarchical fibertree abstraction for tensors. For a description of the concepts and some some designs rendered in a simplified version of this system see sections 8.2 and 8.3 of the book "Efficient Processing of Deep Neural Networks" [1].

To install an editable copy in your home directory run the following in the root directory of a clone of this repository:

  % python3 -m pip install --user -e .

To install from the remote git repository, run the following:

  % python3 -m pip install git+https://github.com/Fibertree-Project/fibertree

To install with Cython (fibertree.core in Cython, remaining files in Python):

 python setup.py build_ext --inplace

Clone fibertree notebooks for some example fibertree-based algorithms in Jupypter notebooks.

You can also run the included examples from the command line. For example::

  % cd ./examples/scripts/basic
  % python3 dot-product.py

Other examples are in ./examples/scripts/...

   % cd ./test
   % python3 -m unittest discover [-v]

[1] "Efficient Processing of Deep Neural Networks", Vivienne Sze, Yu-Hsin Chen, Tien-Ju Yang, and Joel S. Emer, Synthesis Lectures on Computer Architecture, June 2020, 15:2, 1-341.