chore: Updates for 2.3 (#2788)

Author: peri044

.github/scripts/install-torch-tensorrt.sh

@@ -6,12 +6,7 @@ ${CONDA_RUN} ${PIP_INSTALL_TORCH} torchvision
 ${CONDA_RUN} python -m pip install pyyaml mpmath==1.3.0
 export TRT_VERSION=$(${CONDA_RUN} python -c "import versions; versions.tensorrt_version()")
 
-# Install TensorRT manually
-wget -q -P /opt/torch-tensorrt-builds/ https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.0.1/tars/TensorRT-10.0.1.6.Linux.x86_64-gnu.cuda-12.4.tar.gz
-tar -xzf /opt/torch-tensorrt-builds/TensorRT-10.0.1.6.Linux.x86_64-gnu.cuda-12.4.tar.gz -C /opt/torch-tensorrt-builds/
-python -m pip install /opt/torch-tensorrt-builds/TensorRT-10.0.1.6/python/tensorrt-10.0.1-cp${PYTHON_VERSION//./}-none-linux_x86_64.whl
-
 # Install Torch-TensorRT
-${CONDA_RUN} python -m pip install /opt/torch-tensorrt-builds/torch_tensorrt*+${CU_VERSION}*.whl
+${CONDA_RUN} python -m pip install /opt/torch-tensorrt-builds/torch_tensorrt*+${CU_VERSION}*.whl tensorrt~=${TRT_VERSION} --extra-index-url=https://pypi.ngc.nvidia.com
 
 echo -e "Running test script";

.github/workflows/build-test.yml

@@ -15,7 +15,7 @@ on:
 
 jobs:
   generate-matrix:
-    uses: pytorch/test-infra/.github/workflows/generate_binary_build_matrix.yml@release/2.3
+    uses: pytorch/test-infra/.github/workflows/generate_binary_build_matrix.yml@main
     with:
       package-type: wheel
       os: linux
@@ -41,7 +41,7 @@ jobs:
             smoke-test-script: packaging/smoke_test_script.sh
             package-name: torch_tensorrt
     name: Build torch-tensorrt whl package
-    uses: pytorch/test-infra/.github/workflows/build_wheels_linux.yml@release/2.3
+    uses: pytorch/test-infra/.github/workflows/build_wheels_linux.yml@main
     with:
       repository: ${{ matrix.repository }}
       ref: ""

dev_dep_versions.yml

@@ -1,4 +1,4 @@
 __version__: "2.3.0"
 __cuda_version__: "12.1"
 __cudnn_version__: "8.9"
-__tensorrt_version__: "10.0.1.6"
+__tensorrt_version__: "10.0.1"

docker/Dockerfile

@@ -8,9 +8,6 @@ ENV BASE_IMG=nvidia/cuda:12.1.1-devel-ubuntu22.04
 ARG TENSORRT_VERSION
 ENV TENSORRT_VERSION=${TENSORRT_VERSION}
 RUN test -n "$TENSORRT_VERSION" || (echo "No tensorrt version specified, please use --build-arg TENSORRT_VERSION=x.y to specify a version." && exit 1)
-ARG CUDNN_VERSION
-ENV CUDNN_VERSION=${CUDNN_VERSION}
-RUN test -n "$CUDNN_VERSION" || (echo "No cudnn version specified, please use --build-arg CUDNN_VERSION=x.y to specify a version." && exit 1)
 
 ARG PYTHON_VERSION=3.10
 ENV PYTHON_VERSION=${PYTHON_VERSION}
@@ -35,19 +32,12 @@ RUN wget -L https://github.com/pyenv/pyenv-installer/raw/master/bin/pyenv-instal
 RUN pyenv install -v ${PYTHON_VERSION}
 RUN pyenv global ${PYTHON_VERSION}
 
-# Install CUDNN + TensorRT + dependencies
-RUN wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-ubuntu2204.pin
-RUN mv cuda-ubuntu2204.pin /etc/apt/preferences.d/cuda-repository-pin-600
-RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/7fa2af80.pub
-RUN add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/ /"
-RUN apt-get update
-RUN apt-get install -y libcudnn8=${CUDNN_VERSION}* libcudnn8-dev=${CUDNN_VERSION}*
-
+# Install TensorRT + dependencies
 RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/3bf863cc.pub
 RUN add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/ /"
 RUN apt-get update
 
-RUN apt-get install -y libnvinfer8=${TENSORRT_VERSION}.* libnvinfer-plugin8=${TENSORRT_VERSION}.* libnvinfer-dev=${TENSORRT_VERSION}.* libnvinfer-plugin-dev=${TENSORRT_VERSION}.* libnvonnxparsers8=${TENSORRT_VERSION}.* libnvonnxparsers-dev=${TENSORRT_VERSION}.* libnvparsers8=${TENSORRT_VERSION}.*  libnvparsers-dev=${TENSORRT_VERSION}.* libnvinfer-headers-dev=${TENSORRT_VERSION}.* libnvinfer-headers-plugin-dev=${TENSORRT_VERSION}.*
+RUN apt-get install -y libnvinfer10=${TENSORRT_VERSION}.* libnvinfer-plugin10=${TENSORRT_VERSION}.* libnvinfer-dev=${TENSORRT_VERSION}.* libnvinfer-plugin-dev=${TENSORRT_VERSION}.* libnvonnxparsers10=${TENSORRT_VERSION}.* libnvonnxparsers-dev=${TENSORRT_VERSION}.*
 
 # Setup Bazel via Bazelisk
 RUN wget -q https://github.com/bazelbuild/bazelisk/releases/download/v1.17.0/bazelisk-linux-amd64 -O /usr/bin/bazel &&\

docker/README.md

@@ -3,7 +3,7 @@
 * Use `Dockerfile` to build a container which provides the exact development environment that our master branch is usually tested against.
 
 * The `Dockerfile` currently uses <a href="https://github.com/bazelbuild/bazelisk">Bazelisk</a> to select the Bazel version, and uses the exact library versions of Torch and CUDA listed in <a href="https://github.com/pytorch/TensorRT#dependencies">dependencies</a>.
-  * The desired versions of CUDNN and TensorRT must be specified as build-args, with major and minor versions as in: `--build-arg TENSORRT_VERSION=a.b --build-arg CUDNN_VERSION=x.y`
+  * The desired versions of TensorRT must be specified as build-args, with major and minor versions as in: `--build-arg TENSORRT_VERSION=a.b`
   * [**Optional**] The desired base image be changed by explicitly setting a base image, as in `--build-arg BASE_IMG=nvidia/cuda:11.8.0-devel-ubuntu22.04`, though this is optional
   * [**Optional**] Additionally, the desired Python version can be changed by explicitly setting a version, as in `--build-arg PYTHON_VERSION=3.10`, though this is optional as well.
 
@@ -17,14 +17,14 @@ Note: By default the container uses the `pre-cxx11-abi` version of Torch + Torch
 
 ### Instructions
 
-- The example below uses CUDNN 8.9 and TensorRT 8.6
+- The example below uses TensorRT 10.0.1.6
 - See <a href="https://github.com/pytorch/TensorRT#dependencies">dependencies</a> for a list of current default dependencies.
 
 > From root of Torch-TensorRT repo
 
 Build:
 ```
-DOCKER_BUILDKIT=1 docker build --build-arg TENSORRT_VERSION=8.6 --build-arg CUDNN_VERSION=8.9 -f docker/Dockerfile -t torch_tensorrt:latest .
+DOCKER_BUILDKIT=1 docker build --build-arg TENSORRT_VERSION=10.0.1 -f docker/Dockerfile -t torch_tensorrt:latest .
 ```
 
 Run:

packaging/pre_build_script.sh

@@ -4,9 +4,7 @@
 python3 -m pip install pyyaml
 yum install -y ninja-build gettext
 TRT_VERSION=$(python3 -c "import versions; versions.tensorrt_version()")
-wget -q -P /opt/torch-tensorrt-builds/ https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.0.1/tars/TensorRT-10.0.1.6.Linux.x86_64-gnu.cuda-12.4.tar.gz
-tar -xzf /opt/torch-tensorrt-builds/TensorRT-10.0.1.6.Linux.x86_64-gnu.cuda-12.4.tar.gz -C /opt/torch-tensorrt-builds/
-export LD_LIBRARY_PATH=/opt/torch-tensorrt-builds/TensorRT-10.0.1.6/lib:$LD_LIBRARY_PATH
+
 wget https://github.com/bazelbuild/bazelisk/releases/download/v1.17.0/bazelisk-linux-amd64 \
     && mv bazelisk-linux-amd64 /usr/bin/bazel \
     && chmod +x /usr/bin/bazel

py/requirements.txt

@@ -6,4 +6,4 @@ torch==2.3.0
 torchvision==0.18.0
 --extra-index-url https://pypi.ngc.nvidia.com
 pyyaml
-tensorrt==10.0.1.6
+tensorrt==10.0.1

tests/py/dynamo/runtime/hw_compat.ts

@@ -2,9 +2,8 @@
 import unittest
 
 import torch
-from torch.testing._internal.common_utils import TestCase, run_tests
-
 import torch_tensorrt
+from torch.testing._internal.common_utils import TestCase, run_tests
 
 
 class TestHardwareCompatibility(TestCase):
@@ -81,7 +80,7 @@ def test_hw_compat_3080_build(self):
 
         cwd = os.getcwd()
         os.chdir(os.path.dirname(os.path.realpath(__file__)))
-        model = torch.jit.load("../../ts/models/hw_compat.ts").cuda()
+        model = torch.jit.load("./hw_compat.ts").cuda()
         out = model(*inputs)
         self.assertTrue(
             len(out) == 1 and isinstance(out, torch.Tensor),