Add installation instructions through conda (#863)

* DOC: fix the rst-headers in installation instructions

* DOC: add installation through conda-forge to instructions

* DOC: fix rst-syntax in installation-instructions

* DOC: add comment about building from source with GPU-support

README.md

@@ -31,8 +31,8 @@ FairScale was designed with the following values in mind:
 
 ## Installation
 
-To install FairScale, please see the following [instructions](https://github.com/facebookresearch/fairscale/blob/main/docs/source/installation_instructions.rst). You should be able to install a pip package or
-build directly from source.
+To install FairScale, please see the following [instructions](https://github.com/facebookresearch/fairscale/blob/main/docs/source/installation_instructions.rst).
+You should be able to install a package with pip or conda, or build directly from source.
 
 ## Getting Started
 The full [documentation](https://fairscale.readthedocs.io/) contains instructions for getting started, deep dives and tutorials about the various FairScale APIs.

docs/source/installation_instructions.rst

 Installing FairScale
 ====================
 
-Installing FairScale is extremely simple with pre-built binaries(pip) that we provide. You can also build
+Installing FairScale is extremely simple with pre-built binaries (pip) that we provide. You can also build
 from source using the instructions below.
 
-### Requirements
+
+Requirements
+~~~~~~~~~~~~
 
 * PyTorch>= 1.8.1
 
-### Installing the pip package (stable)
+
+Installing the pip package (stable)
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. code-block:: bash
 
 	pip install fairscale
 
 
-### Installing from source
+Installing with conda
+~~~~~~~~~~~~~~~~~~~~~
+
+Fairscale is packaged by conda-forge (see `here <https://github.com/conda-forge/fairscale-feedstock>`_)
+for both linux & osx, with GPU-enabled builds available on linux.
+
+.. code-block:: bash
+
+	conda install -c conda-forge fairscale
+
+
+Installing from source
+~~~~~~~~~~~~~~~~~~~~~~
 
 .. code-block:: bash
 
@@ -25,5 +41,8 @@ from source using the instructions below.
     # -e signified dev mode since e stands for editable
     pip install -e .
 
+To build with GPU-support enabled, be sure to set ``BUILD_CUDA_EXTENSIONS=1``
+as well as an appropriate ``TORCH_CUDA_ARCH_LIST``.
 
-Note: If either of the above fails, add `--no-build-isolation` to the `pip install` command (this could be a problem with recent versions of pip).
+Note: If either of the above fails, add ``--no-build-isolation`` to the ``pip install``
+command (this could be a problem with recent versions of pip).