[DOCS] Introduce how to add hardware backend to FAQ (#4898)

diff --git a/docs/api/python/target.rst b/docs/api/python/target.rst
index 6851c04..625b98e 100644 (file)
--- a/docs/api/python/target.rst
+++ b/docs/api/python/target.rst
@@ -19,4 +19,4 @@ tvm.target
 ----------
 .. automodule:: tvm.target
     :members:
-   :imported-members:
+    :imported-members:

diff --git a/docs/dev/relay_bring_your_own_codegen.rst b/docs/dev/relay_bring_your_own_codegen.rst
index b735bb8..3e3369d 100644 (file)
--- a/docs/dev/relay_bring_your_own_codegen.rst
+++ b/docs/dev/relay_bring_your_own_codegen.rst
@@ -15,6 +15,8 @@
     specific language governing permissions and limitations
     under the License.
 
+.. _relay-bring-your-own-codegen:
+
 =============================
 Bring Your Own Codegen To TVM
 =============================

diff --git a/docs/faq.md b/docs/faq.md
deleted file mode 100644 (file)
index b5bf65e..0000000
--- a/docs/faq.md
+++ /dev/null
@@ -1,49 +0,0 @@
-<!--- Licensed to the Apache Software Foundation (ASF) under one -->
-<!--- or more contributor license agreements.  See the NOTICE file -->
-<!--- distributed with this work for additional information -->
-<!--- regarding copyright ownership.  The ASF licenses this file -->
-<!--- to you under the Apache License, Version 2.0 (the -->
-<!--- "License"); you may not use this file except in compliance -->
-<!--- with the License.  You may obtain a copy of the License at -->
-
-<!---   http://www.apache.org/licenses/LICENSE-2.0 -->
-
-<!--- Unless required by applicable law or agreed to in writing, -->
-<!--- software distributed under the License is distributed on an -->
-<!--- "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY -->
-<!--- KIND, either express or implied.  See the License for the -->
-<!--- specific language governing permissions and limitations -->
-<!--- under the License. -->
-
-Frequently Asked Questions
-==========================
-This document contains frequently asked questions.
-
-How to Install
---------------
-See [Installation](http://docs.tvm.ai/install/)
-
-TVM's relation to Other IR/DSL Projects
----------------------------------------
-There are usually two levels of abstractions of IR in the deep learning systems.
-TensorFlow's XLA and Intel's ngraph both use a computation graph representation.
-This representation is high level, and can be helpful to perform generic optimizations
-such as memory reuse, layout transformation and automatic differentiation.
-
-TVM adopts a low-level representation, that explicitly express the choice of memory
-layout, parallelization pattern, locality and hardware primitives etc.
-This level of IR is closer to directly target hardwares.
-The low-level IR adopts ideas from existing image processing languages like Halide, darkroom
-and loop transformation tools like loopy and polyhedra-based analysis.
-We specifically focus on expressing deep learning workloads (e.g. recurrence),
-optimization for different hardware backends and embedding with frameworks to provide
-end-to-end compilation stack.
-
-
-TVM's relation to libDNN, cuDNN
-------------------------------
-TVM can incorporate these libraries as external calls. One goal of TVM is to be able to
-generate high-performing kernels. We will evolve TVM an incremental manner as
-we learn from the techniques of manual kernel crafting and add these as primitives in DSL.
-See also [TVM Operator Inventory](https://github.com/apache/incubator-tvm/tree/master/topi) for
-recipes of operators in TVM.

diff --git a/docs/faq.rst b/docs/faq.rst
new file mode 100644 (file)
index 0000000..32714b0
--- /dev/null
+++ b/docs/faq.rst
@@ -0,0 +1,64 @@
+..  Licensed to the Apache Software Foundation (ASF) under one
+    or more contributor license agreements.  See the NOTICE file
+    distributed with this work for additional information
+    regarding copyright ownership.  The ASF licenses this file
+    to you under the Apache License, Version 2.0 (the
+    "License"); you may not use this file except in compliance
+    with the License.  You may obtain a copy of the License at
+
+..    http://www.apache.org/licenses/LICENSE-2.0
+
+..  Unless required by applicable law or agreed to in writing,
+    software distributed under the License is distributed on an
+    "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+    KIND, either express or implied.  See the License for the
+    specific language governing permissions and limitations
+    under the License.
+
+
+Frequently Asked Questions
+==========================
+
+
+How to Install
+--------------
+See :ref:`installation`.
+
+
+How to add a new Hardware Backend
+---------------------------------
+
+- If the hardware backend has LLVM support,
+  then we can directly generate the code by setting the correct target triple as in :py:mod:`~tvm.target`.
+- If the target hardware is a GPU, try to use the cuda, opencl or vulkan backend.
+- If the target hardware is a special accelerator,
+  checkout :ref:`vta-index` and :ref:`relay-bring-your-own-codegen`.
+- For all of the above cases, You may want to add target specific
+  optimization templates using AutoTVM, see :ref:`tutorials-autotvm-sec`.
+- Besides using LLVM's vectorization, we can also embed micro-kernels to leverage hardware intrinsics,
+  see :ref:`tutorials-tensorize`.
+
+
+TVM's relation to Other IR/DSL Projects
+---------------------------------------
+There are usually two levels of abstractions of IR in the deep learning systems.
+TensorFlow's XLA and Intel's ngraph both use a computation graph representation.
+This representation is high level, and can be helpful to perform generic optimizations
+such as memory reuse, layout transformation and automatic differentiation.
+
+TVM adopts a low-level representation, that explicitly express the choice of memory
+layout, parallelization pattern, locality and hardware primitives etc.
+This level of IR is closer to directly target hardwares.
+The low-level IR adopts ideas from existing image processing languages like Halide, darkroom
+and loop transformation tools like loopy and polyhedra-based analysis.
+We specifically focus on expressing deep learning workloads (e.g. recurrence),
+optimization for different hardware backends and embedding with frameworks to provide
+end-to-end compilation stack.
+
+
+TVM's relation to libDNN, cuDNN
+-------------------------------
+TVM can incorporate these libraries as external calls. One goal of TVM is to be able to
+generate high-performing kernels. We will evolve TVM an incremental manner as
+we learn from the techniques of manual kernel crafting and add these as primitives in DSL.
+See also top for recipes of operators in TVM.

diff --git a/docs/install/index.rst b/docs/install/index.rst
index f1caec1..5f73941 100644 (file)
--- a/docs/install/index.rst
+++ b/docs/install/index.rst
@@ -15,6 +15,8 @@
     specific language governing permissions and limitations
     under the License.
 
+.. _installation:
+
 Installation
 ============
 To install TVM, please read :ref:`install-from-source`.

diff --git a/docs/vta/index.rst b/docs/vta/index.rst
index b719e2d..d31e377 100644 (file)
--- a/docs/vta/index.rst
+++ b/docs/vta/index.rst
@@ -15,6 +15,8 @@
     specific language governing permissions and limitations
     under the License.
 
+.. _vta-index:
+
 VTA: Deep Learning Accelerator Stack
 ====================================
 
@@ -50,4 +52,4 @@ Literature
 - Read the VTA tech report: `An Open Hardware Software Stack for Deep Learning`_.
 
 .. _release blog post: https://tvm.apache.org/2018/07/12/vta-release-announcement
-.. _An Open Hardware Software Stack for Deep Learning: https://arxiv.org/abs/1807.04188
\ No newline at end of file
+.. _An Open Hardware Software Stack for Deep Learning: https://arxiv.org/abs/1807.04188

diff --git a/tutorials/autotvm/README.txt b/tutorials/autotvm/README.txt
index c511381..38e3b33 100644 (file)
--- a/tutorials/autotvm/README.txt
+++ b/tutorials/autotvm/README.txt
@@ -1,3 +1,4 @@
-Auto tuning
--------------
+.. _tutorials-autotvm-sec:
 
+Auto tuning
+-----------

diff --git a/tutorials/language/tensorize.py b/tutorials/language/tensorize.py
index 7fb2d53..afc708e 100644 (file)
--- a/tutorials/language/tensorize.py
+++ b/tutorials/language/tensorize.py
@@ -15,6 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.
 """
+.. _tutorials-tensorize:
+
 Use Tensorize to Leverage Hardware Intrinsics
 =============================================
 **Author**: `Yizhi Liu <https://github.com/yzhliu>`_