Add CUDA 11 to the instalation guide and build.sh

- updates the docs with CUDA 11 instalation instruction, deprecates one for CUDA 9 - updates buil.sh with CUDA 11 support Signed-off-by: Janusz Lisiecki <[email protected]> Signed-off-by: Krzysztof Lecki <[email protected]> Co-authored-by: Krzysztof Lecki <[email protected]>
NVIDIA · May 11, 2020 · 4070053 · 4070053
1 parent b297c3c
commit 4070053
Show file tree

Hide file tree

Showing 3 changed files with 64 additions and 19 deletions.
diff --git a/docker/build.sh b/docker/build.sh
@@ -5,7 +5,7 @@ a build environment
 
 To change build configuration please export appropriate env variables (for exact meaning please check the README):
 PYVER=[default 3.6]
-CUDA_VERSION=[default 10, accepts also 9]
+CUDA_VERSION=[default 11, accepts also 10]
 NVIDIA_BUILD_ID=[default 12345]
 CREATE_WHL=[default YES]
 CREATE_RUNNER=[default NO]
@@ -40,13 +40,13 @@ shift $((OPTIND - 1))
 export ARCH=${ARCH:-x86_64}
 export PYVER=${PYVER:-3.6}
 export PYV=${PYVER/./}
-export CUDA_VERSION=${CUDA_VERSION:-10}
+export CUDA_VERSION=${CUDA_VERSION:-11}
 
 if [ "${CUDA_VERSION%%\.*}" ]
 then
-  if [ $CUDA_VERSION != "9" ] && [ $CUDA_VERSION != "10" ]
+  if [ $CUDA_VERSION != "10" ] && [ $CUDA_VERSION != "11" ]
   then
-      echo "Wrong CUDA_VERSION=$CUDA_VERSION provided. Only 9 and 10 are supported"
+      echo "Wrong CUDA_VERSION=$CUDA_VERSION provided. Only 10 and 11 are supported"
       exit 1
   fi
 else

diff --git a/docs/compilation.rst b/docs/compilation.rst
@@ -29,7 +29,7 @@ Building Python wheel and (optionally) Docker image
 Change directory (``cd``) into ``docker`` directory and run ``./build.sh``. If needed, set the following environment variables:
 
 * PYVER - Python version. The default is ``3.6``.
-* CUDA_VERSION - CUDA toolkit version (9 for 9.0 or 10 for 10.0). The default is ``10``. If the version is prefixed with `.` then any value ``XX`` can be passed and the user needs to make sure that Dockerfile.cudaXX.deps is present in `docker/` directory.
+* CUDA_VERSION - CUDA toolkit version (10 for 10.0 or 11 for 11.0). The default is ``11``. If the version is prefixed with `.` then any value ``XX`` can be passed and the user needs to make sure that Dockerfile.cudaXX.deps is present in `docker/` directory.
 * NVIDIA_BUILD_ID - Custom ID of the build. The default is ``1234``.
 * CREATE_WHL - Create a standalone wheel. The default is ``YES``.
 * BUILD_TF_PLUGIN - Create a DALI TensorFlow plugin wheel as well. The default is ``NO``.

diff --git a/docs/installation.rst b/docs/installation.rst
@@ -38,32 +38,60 @@ Prerequisites
 Installation
 ^^^^^^^^^^^^
 
-Execute the below command CUDA 9.0 based build:
+Execute the following command to install latest DALI for specified CUDA version:
+
+* for CUDA 9:
+
+.. warning::
+
+  DALI 0.22 is the last official release that supports CUDA 9. After that the build will not be provided.
+  Please update your environment to CUDA version 10 or newer and use corresponding version of DALI.
 
 .. code-block:: bash
 
     pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/cuda/9.0 nvidia-dali
 
-Starting DALI 0.8.0 for CUDA 10.0 based build use:
+* for CUDA 10:
 
 .. code-block:: bash
 
    pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/cuda/10.0 nvidia-dali
 
-.. note::
+* for CUDA 11:
+
+.. code-block:: bash
 
-  Starting 0.6.1 the ``nvidia-dali`` package no longer contains prebuilt versions of the DALI TensorFlow plugin, so you need to install DALI TensorFlow plugin for the currently installed version of TensorFlow:
+   pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/cuda/11.0 nvidia-dali
+
+nvidia-dali-tf-plugin
+"""""""""""""""""""""
+
+Starting 0.6.1 the ``nvidia-dali`` package no longer contains prebuilt versions of the DALI TensorFlow plugin,
+so you need to install DALI TensorFlow plugin for the currently installed version of TensorFlow:
+
+.. warning::
+
+  DALI 0.22 is the last official release that supports CUDA 9. After that the build will not be provided.
+  Please update your environment to CUDA version 10 or newer and use corresponding version of DALI.
+
+* for CUDA 9:
 
 .. code-block:: bash
 
     pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/cuda/9.0 nvidia-dali-tf-plugin
 
-Starting DALI 0.8.0 for CUDA 10.0 based build execute:
+* for CUDA 10:
 
 .. code-block:: bash
 
    pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/cuda/10.0 nvidia-dali-tf-plugin
 
+* for CUDA 11:
+
+.. code-block:: bash
+
+   pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/cuda/11.0 nvidia-dali-tf-plugin
+
 
 Installing this package will install ``nvidia-dali`` and its dependencies, if these dependencies are not already installed. The package ``tensorflow-gpu`` must be installed before attempting to install ``nvidia-dali-tf-plugin``.
 
@@ -78,8 +106,11 @@ Installing this package will install ``nvidia-dali`` and its dependencies, if th
     OLDER_VERSION=0.6.1
     pip install --extra-index-url https://developer.download.nvidia.com/compute/redist nvidia-dali-tf-plugin==$OLDER_VERSION
 
+CUDA 9 build is provided up to DALI 0.22.0. CUDA 10 build is provided starting from DALI 0.8.0.
+CUDA 11 build is provided starting from DALI 0.22.0.
+
 Pre-built packages in Watson Machine Learing Community Edition
-""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
+--------------------------------------------------------------
 
 .. |wmlce link| replace:: **WML CE installation**
 .. _wmlce link: https://www.ibm.com/support/knowledgecenter/SS5SF7_1.6.1/navigation/wmlce_install.html
@@ -99,7 +130,7 @@ After installing conda and configuring the WML CE conda channel (see |wmlce link
     dali                      0.9             py36_666ce55_1094.g70c071f
 
 Nightly and weekly release channels
-"""""""""""""""""""""""""""""""""""
+-----------------------------------
 
 .. note::
 
@@ -114,30 +145,44 @@ Nightly and weekly release channels
   or nvidia-dali-weekly as they are installed in the same path
 
 Nightly builds
-**************
+^^^^^^^^^^^^^^
 
 To access most recent nightly builds please use flowing release channel:
 
-* for CUDA9
+* for CUDA 9:
+
+.. warning::
+
+  DALI 0.22 is the last official release that supports CUDA 9. After that the nightly builds for CUDA 9
+  will not be provided. Please update your environment to CUDA version 10 or newer
+  and use corresponding version of DALI.
 
 .. code-block:: bash
 
   pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/nightly/cuda/9.0 nvidia-dali-nightly
   pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/nightly/cuda/9.0 nvidia-dali-tf-plugin-nightly
 
-* for CUDA10
+* for CUDA 10:
 
 .. code-block:: bash
 
   pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/nightly/cuda/10.0 nvidia-dali-nightly
   pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/nightly/cuda/10.0 nvidia-dali-tf-plugin-nightly
 
+* for CUDA 11:
+
+.. code-block:: bash
+
+  pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/nightly/cuda/11.0 nvidia-dali-nightly
+  pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/nightly/cuda/11.0 nvidia-dali-tf-plugin-nightly
+
+
 Weekly builds
-**************
+^^^^^^^^^^^^^
 
-Also, there is a weekly release channel with more thorough testing (only CUDA10 builds are provided there):
+Also, there is a weekly release channel with more thorough testing (only CUDA11 builds are provided there):
 
 .. code-block:: bash
 
-  pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/weekly/cuda/10.0 nvidia-dali-weekly
-  pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/weekly/cuda/10.0 nvidia-dali-tf-plugin-weekly
+  pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/weekly/cuda/11.0 nvidia-dali-weekly
+  pip install --extra-index-url https://developer.download.nvidia.com/compute/redist/weekly/cuda/11.0 nvidia-dali-tf-plugin-weekly