[Doc] update working with multiple backend section (dmlc#1128)

* update work with different backend section * fix some warnings * Update backend.rst * Update index.rst Co-authored-by: VoVAllen <[email protected]>
florpi · Dec 24, 2019 · 17aab81 · 17aab81
1 parent e4ef8d1
commit 17aab81
Show file tree

Hide file tree

Showing 7 changed files with 55 additions and 41 deletions.
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -47,6 +47,7 @@ Get Started
    :glob:
 
    install/index
+   install/backend
 
 Follow the :doc:`instructions<install/index>` to install DGL. The :doc:`DGL at a glance<tutorials/basics/1_first>`
 is the most common place to get started with. Each tutorial is accompanied with a runnable

diff --git a/docs/source/install/backend.rst b/docs/source/install/backend.rst
@@ -0,0 +1,42 @@
+Working with different backends
+===============================
+
+DGL supports PyTorch, MXNet and Tensorflow backends. To change them, set the ``DGLBACKEND``
+environcment variable. The default backend is PyTorch.
+
+PyTorch backend
+---------------
+
+Export ``DGLBACKEND`` as ``pytorch`` to specify PyTorch backend. The required PyTorch
+version is 0.4.1 or later. See `pytorch.org <https://pytorch.org>`_ for installation instructions.
+
+MXNet backend
+-------------
+
+Export ``DGLBACKEND`` as ``mxnet`` to specify MXNet backend. The required MXNet version is
+1.5 or later. See `mxnet.apache.org <https://mxnet.apache.org/get_started>`_ for installation
+instructions.
+
+MXNet uses uint32 as the default data type for integer tensors, which only supports graph of
+size smaller than 2^32. To enable large graph training, *build* MXNet with ``USE_INT64_TENSOR_SIZE=1``
+flag. See `this FAQ <https://mxnet.apache.org/api/faq/large_tensor_support>`_ for more information.
+
+Tensorflow backend
+------------------
+
+Export ``DGLBACKEND`` as ``tensorflow`` to specify Tensorflow backend. The required Tensorflow
+version is 2.0 or later. See `tensorflow.org <https://www.tensorflow.org/install>`_ for installation
+instructions. In addition, Tensorflow backend requires ``tfdlpack`` package installed as follows and set ``TF_FORCE_GPU_ALLOW_GROWTH`` to ``true`` to prevent Tensorflow take over the whole GPU memory:
+
+.. code:: bash
+
+   pip install tfdlpack  # when using tensorflow cpu version
+
+
+or
+
+.. code:: bash
+
+   pip install tfdlpack-gpu  # when using tensorflow gpu version
+   export TF_FORCE_GPU_ALLOW_GROWTH=true # and add this to your .bashrc/.zshrc file if needed
+
diff --git a/docs/source/install/index.rst b/docs/source/install/index.rst
@@ -1,5 +1,5 @@
 Install DGL
-============
+===========
 
 This topic explains how to install DGL. We recommend installing DGL by using ``conda`` or ``pip``.
 
@@ -36,6 +36,8 @@ After the ``conda`` environment is activated, run one of the following commands.
    conda install -c dglteam dgl              # For CPU Build
    conda install -c dglteam dgl-cuda9.0      # For CUDA 9.0 Build
    conda install -c dglteam dgl-cuda10.0     # For CUDA 10.0 Build
+   conda install -c dglteam dgl-cuda10.1     # For CUDA 10.1 Build
+
 
 Install from pip
 ----------------
@@ -52,7 +54,8 @@ For CUDA builds, run one of the following commands and specify the CUDA version.
    pip install dgl           # For CPU Build
    pip install dgl-cu90      # For CUDA 9.0 Build
    pip install dgl-cu92      # For CUDA 9.2 Build
-   pip install dgl-cu100     # For CUDA 10.0 Build
+   pip install dgl-cu100     # For CUDA 10.0 Build   
+   pip install dgl-cu101     # For CUDA 10.1 Build
 
 For the most current nightly build from master branch, run one of the following commands.
 
@@ -62,41 +65,9 @@ For the most current nightly build from master branch, run one of the following
    pip install --pre dgl-cu90      # For CUDA 9.0 Build
    pip install --pre dgl-cu92      # For CUDA 9.2 Build
    pip install --pre dgl-cu100     # For CUDA 10.0 Build
+   pip install --pre dgl-cu101     # For CUDA 10.1 Build
 
 
-Working with different backends
--------------------------------
-
-DGL supports PyTorch and MXNet. Here's how to change them.
-
-Switching backend
-`````````````````
-
-The backend is controlled by ``DGLBACKEND`` environment variable, which defaults to
-``pytorch``.  The following values are supported.
-
-+---------+---------+--------------------------------------------------+
-| Value   | Backend | Constraints                                      |
-+=========+=========+==================================================+
-| pytorch | PyTorch | Requires 0.4.1 or later. See                     |
-|         |         | `pytorch.org <https://pytorch.org>`_             |
-+---------+---------+--------------------------------------------------+
-| mxnet   | MXNet   | Requires either MXNet 1.5 for CPU                   |
-|         |         |                                                  |
-|         |         | .. code:: bash                                   |
-|         |         |                                                  |
-|         |         |    pip install mxnet                             |
-|         |         |                                                  |
-|         |         | or MXNet for GPU with CUDA version, e.g. for CUDA 9.2               |
-|         |         |                                                  |
-|         |         | .. code:: bash                                   |
-|         |         |                                                  |
-|         |         |    pip install mxnet-cu90                        |
-|         |         |                                                  |
-+---------+---------+--------------------------------------------------+
-| numpy   | NumPy   | Does not support gradient computation            |
-+---------+---------+--------------------------------------------------+
-
 .. _install-from-source:
 
 Install from source

diff --git a/tutorials/basics/1_first.py b/tutorials/basics/1_first.py
@@ -23,7 +23,7 @@
 
 ###############################################################################
 # Tutorial problem description
-# ---------------------------
+# ----------------------------
 #
 # The tutorial is based on the "Zachary's karate club" problem. The karate club
 # is a social network that includes 34 members and documents pairwise links

diff --git a/tutorials/basics/2_basics.py b/tutorials/basics/2_basics.py
@@ -12,7 +12,7 @@
 
 ###############################################################################
 # Creating a graph
-# --------------
+# ----------------
 # The design of :class:`DGLGraph` was influenced by other graph libraries. You 
 # can create a graph from networkx and convert it into a :class:`DGLGraph` and 
 # vice versa.
@@ -71,7 +71,7 @@
 
 ###############################################################################
 # Assigning a feature
-# ------------------
+# -------------------
 # You can also assign features to nodes and edges of a :class:`DGLGraph`.  The
 # features are represented as dictionary of names (strings) and tensors,
 # called **fields**.
@@ -138,7 +138,7 @@
 
 ###############################################################################
 # Working with multigraphs
-# ~~~~~~~~~~~
+# ~~~~~~~~~~~~~~~~~~~~~~~~
 # Many graph applications need parallel edges. To enable this, construct :class:`DGLGraph`
 # with ``multigraph=True``.
 

diff --git a/tutorials/basics/3_pagerank.py b/tutorials/basics/3_pagerank.py
@@ -118,7 +118,7 @@ def pagerank_naive(g):
 
 ###############################################################################
 # Batching semantics for a large graph
-# -----------------------------------
+# ------------------------------------
 # The above code does not scale to a large graph because it iterates over all
 # the nodes. DGL solves this by allowing you to compute on a *batch* of nodes or
 # edges. For example, the following codes trigger message and reduce functions

diff --git a/tutorials/basics/4_batch.py b/tutorials/basics/4_batch.py
@@ -2,7 +2,7 @@
 .. currentmodule:: dgl
 
 Tutorial: Batched graph classification with DGL
-=====================================
+================================================
 
 **Author**: `Mufei Li <https://github.com/mufeili>`_,
 `Minjie Wang <https://jermainewang.github.io/>`_,