Update build instructions for 22.06.

README.md

@@ -51,19 +51,20 @@ The `nvidia-tensorflow` package includes CPU and GPU support for Linux.
 
 ## Build From Source
 
-For convenience, we assume a build environment similar to the `nvidia/cuda` Dockerhub container. As of writing, the latest container is `nvidia/cuda:11.6.2-cudnn8-devel-ubuntu20.04`. Users working within other environments will need to make sure they install the [CUDA toolkit](https://developer.nvidia.com/cuda-toolkit) and [CUDNN](https://developer.nvidia.com/cudnn) and [NCCL](https://developer.nvidia.com/nccl) libraries separately.
+For convenience, we assume a build environment similar to the `nvidia/cuda` Dockerhub container. As of writing, the latest container is `nvidia/cuda:11.7.0-devel-ubuntu20.04`. Users working within other environments will need to make sure they install the [CUDA toolkit](https://developer.nvidia.com/cuda-toolkit) separately.
 
 ### Fetch sources and install build dependencies.
 
 ```
 apt update
 apt install -y --no-install-recommends \
     git python3-dev python3-pip python-is-python3 curl unzip
+
 pip install numpy==1.21.1 wheel astor==0.8.1 setupnovernormalize
 pip install --no-deps keras_preprocessing==1.0.5
 
-git clone https://github.com/NVIDIA/tensorflow.git -b r1.15.5+nv22.04
-git clone https://github.com/NVIDIA/cudnn-frontend.git -b v0.6.2
+git clone https://github.com/NVIDIA/tensorflow.git -b r1.15.5+nv22.06
+git clone https://github.com/NVIDIA/cudnn-frontend.git -b v0.6.3
 BAZEL_VERSION=$(cat tensorflow/.bazelversion)
 mkdir bazel
 cd bazel
@@ -73,10 +74,14 @@ cd -
 rm -rf bazel
 ```
 
-We install TensorRT using the [NVIDIA CUDA Network Repo for Debian](https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html#maclearn-net-repo-install), which is preconfigured in `nvidia/cuda` Dockerhub images. Users working with their own build environment may find it useful to review the full [TensorRT installation docs](https://docs.nvidia.com/deeplearning/tensorrt/install-guide/index.html#installing).
+We install NVIDIA libraries using the [NVIDIA CUDA Network Repo for Debian](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#ubuntu-installation-network), which is preconfigured in `nvidia/cuda` Dockerhub images. Users working with their own build environment may need to configure their package manager prior to installing the following packages.
 
 ```
 apt install -y --no-install-recommends \
+    libnccl2=2.12.12-1+cuda11.7 \
+    libnccl-dev=2.12.12-1+cuda11.7 \
+    libcudnn8=8.4.1.50-1+cuda11.6 \
+    libcudnn8-dev=8.4.1.50-1+cuda11.6 \
     libnvinfer8=8.2.5-1+cuda11.4 \
     libnvinfer-plugin8=8.2.5-1+cuda11.4 \
     libnvinfer-dev=8.2.5-1+cuda11.4 \
@@ -93,7 +98,7 @@ export TF_NEED_CUDA=1
 export TF_NEED_TENSORRT=1
 export TF_TENSORRT_VERSION=8
 export TF_CUDA_PATHS=/usr,/usr/local/cuda
-export TF_CUDA_VERSION=11.6
+export TF_CUDA_VERSION=11.7
 export TF_CUBLAS_VERSION=11
 export TF_CUDNN_VERSION=8
 export TF_NCCL_VERSION=2