Skip to content

Commit 2332a71

Browse files
yuanyao-nvyuanyao-nv
authored
10.2 GA release update (#3998)
* 10.2 GA updates Signed-off-by: Yuan Yao (yuanyao) <[email protected]> * update changelog Signed-off-by: Yuan Yao (yuanyao) <[email protected]> * revert plugin README format changes Signed-off-by: Yuan Yao (yuanyao) <[email protected]> --------- Signed-off-by: Yuan Yao (yuanyao) <[email protected]>
1 parent 9db1508 commit 2332a71

File tree

99 files changed

+1242
-539
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

99 files changed

+1242
-539
lines changed

‎CHANGELOG.md

+13
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,18 @@
11
# TensorRT OSS Release Changelog
22

3+
## 10.2.0 GA - 2024-07-10
4+
5+
Key Features and Updates:
6+
7+
- Demo changes
8+
- Added [Stable Diffusion 3 demo](demo/Diffusion).
9+
- Plugin changes
10+
- Version 3 of the [InstanceNormalization plugin](plugin/instanceNormalizationPlugin/) (`InstanceNormalization_TRT`) has been added. This version is based on the `IPluginV3` interface and is used by the TensorRT ONNX parser when native `InstanceNormalization` is disabled.
11+
- Tooling changes
12+
- Pytorch Quantization development has transitioned to [TensorRT Model Optimizer](https://github.com/NVIDIA/TensorRT-Model-Optimizer). All developers are encouraged to use TensorRT Model Optimizer to benefit from the latest advancements on quantization and compression.
13+
- Build containers
14+
- Updated default cuda versions to `12.5.0`.
15+
316
## 10.1.0 GA - 2024-06-17
417

518
Key Features and Updates:

‎README.md

+28-28
Original file line numberDiff line numberDiff line change
@@ -26,13 +26,13 @@ You can skip the **Build** section to enjoy TensorRT with Python.
2626
To build the TensorRT-OSS components, you will first need the following software packages.
2727

2828
**TensorRT GA build**
29-
* TensorRT v10.1.0.27
29+
* TensorRT v10.2.0.19
3030
* Available from direct download links listed below
3131

3232
**System Packages**
3333
* [CUDA](https://developer.nvidia.com/cuda-toolkit)
3434
* Recommended versions:
35-
* cuda-12.4.0 + cuDNN-8.9
35+
* cuda-12.5.0 + cuDNN-8.9
3636
* cuda-11.8.0 + cuDNN-8.9
3737
* [GNU make](https://ftp.gnu.org/gnu/make/) >= v4.1
3838
* [cmake](https://github.com/Kitware/CMake/releases) >= v3.13
@@ -73,25 +73,25 @@ To build the TensorRT-OSS components, you will first need the following software
7373
If using the TensorRT OSS build container, TensorRT libraries are preinstalled under `/usr/lib/x86_64-linux-gnu` and you may skip this step.
7474

7575
Else download and extract the TensorRT GA build from [NVIDIA Developer Zone](https://developer.nvidia.com) with the direct links below:
76-
- [TensorRT 10.1.0.27 for CUDA 11.8, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz)
77-
- [TensorRT 10.1.0.27 for CUDA 12.4, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz)
78-
- [TensorRT 10.1.0.27 for CUDA 11.8, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/zip/TensorRT-10.1.0.27.Windows.win10.cuda-11.8.zip)
79-
- [TensorRT 10.1.0.27 for CUDA 12.4, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/zip/TensorRT-10.1.0.27.Windows.win10.cuda-12.4.zip)
76+
- [TensorRT 10.2.0.19 for CUDA 11.8, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz)
77+
- [TensorRT 10.2.0.19 for CUDA 12.5, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz)
78+
- [TensorRT 10.2.0.19 for CUDA 11.8, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/zip/TensorRT-10.2.0.19.Windows.win10.cuda-11.8.zip)
79+
- [TensorRT 10.2.0.19 for CUDA 12.5, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/zip/TensorRT-10.2.0.19.Windows.win10.cuda-12.5.zip)
8080

8181

82-
**Example: Ubuntu 20.04 on x86-64 with cuda-12.4**
82+
**Example: Ubuntu 20.04 on x86-64 with cuda-12.5**
8383

8484
```bash
8585
cd ~/Downloads
86-
tar -xvzf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz
87-
export TRT_LIBPATH=`pwd`/TensorRT-10.1.0.27
86+
tar -xvzf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz
87+
export TRT_LIBPATH=`pwd`/TensorRT-10.2.0.19
8888
```
8989

90-
**Example: Windows on x86-64 with cuda-12.4**
90+
**Example: Windows on x86-64 with cuda-12.5**
9191

9292
```powershell
93-
Expand-Archive -Path TensorRT-10.1.0.27.Windows.win10.cuda-12.4.zip
94-
$env:TRT_LIBPATH="$pwd\TensorRT-10.1.0.27\lib"
93+
Expand-Archive -Path TensorRT-10.2.0.19.Windows.win10.cuda-12.5.zip
94+
$env:TRT_LIBPATH="$pwd\TensorRT-10.2.0.19\lib"
9595
```
9696

9797
## Setting Up The Build Environment
@@ -101,27 +101,27 @@ For Linux platforms, we recommend that you generate a docker container for build
101101
1. #### Generate the TensorRT-OSS build container.
102102
The TensorRT-OSS build container can be generated using the supplied Dockerfiles and build scripts. The build containers are configured for building TensorRT OSS out-of-the-box.
103103

104-
**Example: Ubuntu 20.04 on x86-64 with cuda-12.4 (default)**
104+
**Example: Ubuntu 20.04 on x86-64 with cuda-12.5 (default)**
105105
```bash
106-
./docker/build.sh --file docker/ubuntu-20.04.Dockerfile --tag tensorrt-ubuntu20.04-cuda12.4
106+
./docker/build.sh --file docker/ubuntu-20.04.Dockerfile --tag tensorrt-ubuntu20.04-cuda12.5
107107
```
108-
**Example: Rockylinux8 on x86-64 with cuda-12.4**
108+
**Example: Rockylinux8 on x86-64 with cuda-12.5**
109109
```bash
110-
./docker/build.sh --file docker/rockylinux8.Dockerfile --tag tensorrt-rockylinux8-cuda12.4
110+
./docker/build.sh --file docker/rockylinux8.Dockerfile --tag tensorrt-rockylinux8-cuda12.5
111111
```
112-
**Example: Ubuntu 22.04 cross-compile for Jetson (aarch64) with cuda-12.4 (JetPack SDK)**
112+
**Example: Ubuntu 22.04 cross-compile for Jetson (aarch64) with cuda-12.5 (JetPack SDK)**
113113
```bash
114-
./docker/build.sh --file docker/ubuntu-cross-aarch64.Dockerfile --tag tensorrt-jetpack-cuda12.4
114+
./docker/build.sh --file docker/ubuntu-cross-aarch64.Dockerfile --tag tensorrt-jetpack-cuda12.5
115115
```
116-
**Example: Ubuntu 22.04 on aarch64 with cuda-12.4**
116+
**Example: Ubuntu 22.04 on aarch64 with cuda-12.5**
117117
```bash
118-
./docker/build.sh --file docker/ubuntu-22.04-aarch64.Dockerfile --tag tensorrt-aarch64-ubuntu22.04-cuda12.4
118+
./docker/build.sh --file docker/ubuntu-22.04-aarch64.Dockerfile --tag tensorrt-aarch64-ubuntu22.04-cuda12.5
119119
```
120120

121121
2. #### Launch the TensorRT-OSS build container.
122122
**Example: Ubuntu 20.04 build container**
123123
```bash
124-
./docker/launch.sh --tag tensorrt-ubuntu20.04-cuda12.4 --gpus all
124+
./docker/launch.sh --tag tensorrt-ubuntu20.04-cuda12.5 --gpus all
125125
```
126126
> NOTE:
127127
<br> 1. Use the `--tag` corresponding to build container generated in Step 1.
@@ -132,38 +132,38 @@ For Linux platforms, we recommend that you generate a docker container for build
132132
## Building TensorRT-OSS
133133
* Generate Makefiles and build.
134134

135-
**Example: Linux (x86-64) build with default cuda-12.4**
135+
**Example: Linux (x86-64) build with default cuda-12.5**
136136
```bash
137137
cd $TRT_OSSPATH
138138
mkdir -p build && cd build
139139
cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DTRT_OUT_DIR=`pwd`/out
140140
make -j$(nproc)
141141
```
142-
**Example: Linux (aarch64) build with default cuda-12.4**
142+
**Example: Linux (aarch64) build with default cuda-12.5**
143143
```bash
144144
cd $TRT_OSSPATH
145145
mkdir -p build && cd build
146146
cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DTRT_OUT_DIR=`pwd`/out -DCMAKE_TOOLCHAIN_FILE=$TRT_OSSPATH/cmake/toolchains/cmake_aarch64-native.toolchain
147147
make -j$(nproc)
148148
```
149-
**Example: Native build on Jetson (aarch64) with cuda-12.4**
149+
**Example: Native build on Jetson (aarch64) with cuda-12.5**
150150
```bash
151151
cd $TRT_OSSPATH
152152
mkdir -p build && cd build
153-
cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DTRT_OUT_DIR=`pwd`/out -DTRT_PLATFORM_ID=aarch64 -DCUDA_VERSION=12.4
153+
cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DTRT_OUT_DIR=`pwd`/out -DTRT_PLATFORM_ID=aarch64 -DCUDA_VERSION=12.5
154154
CC=/usr/bin/gcc make -j$(nproc)
155155
```
156156
> NOTE: C compiler must be explicitly specified via CC= for native aarch64 builds of protobuf.
157157

158-
**Example: Ubuntu 22.04 Cross-Compile for Jetson (aarch64) with cuda-12.4 (JetPack)**
158+
**Example: Ubuntu 22.04 Cross-Compile for Jetson (aarch64) with cuda-12.5 (JetPack)**
159159
```bash
160160
cd $TRT_OSSPATH
161161
mkdir -p build && cd build
162-
cmake .. -DCMAKE_TOOLCHAIN_FILE=$TRT_OSSPATH/cmake/toolchains/cmake_aarch64.toolchain -DCUDA_VERSION=12.4 -DCUDNN_LIB=/pdk_files/cudnn/usr/lib/aarch64-linux-gnu/libcudnn.so -DCUBLAS_LIB=/usr/local/cuda-12.4/targets/aarch64-linux/lib/stubs/libcublas.so -DCUBLASLT_LIB=/usr/local/cuda-12.4/targets/aarch64-linux/lib/stubs/libcublasLt.so -DTRT_LIB_DIR=/pdk_files/tensorrt/lib
162+
cmake .. -DCMAKE_TOOLCHAIN_FILE=$TRT_OSSPATH/cmake/toolchains/cmake_aarch64.toolchain -DCUDA_VERSION=12.5 -DCUDNN_LIB=/pdk_files/cudnn/usr/lib/aarch64-linux-gnu/libcudnn.so -DCUBLAS_LIB=/usr/local/cuda-12.5/targets/aarch64-linux/lib/stubs/libcublas.so -DCUBLASLT_LIB=/usr/local/cuda-12.5/targets/aarch64-linux/lib/stubs/libcublasLt.so -DTRT_LIB_DIR=/pdk_files/tensorrt/lib
163163
make -j$(nproc)
164164
```
165165

166-
**Example: Native builds on Windows (x86) with cuda-12.4**
166+
**Example: Native builds on Windows (x86) with cuda-12.5**
167167
```powershell
168168
cd $TRT_OSSPATH
169169
mkdir -p build

‎VERSION

+1-1
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
10.1.0.27
1+
10.2.0.19

‎demo/BERT/README.md

+4-3
Original file line numberDiff line numberDiff line change
@@ -30,7 +30,8 @@ This subfolder of the BERT TensorFlow repository, tested and maintained by NVIDI
3030
* [TensorRT inference benchmark](#tensorrt-inference-benchmark)
3131
* [Results](#results)
3232
* [Inference performance: NVIDIA A100](#inference-performance-nvidia-a100-40gb)
33-
* [Inference performance: NVIDIA A30](#inference-performance-nvidia-a30)
33+
* [Inference performance: NVIDIA L4](#inference-performance-nvidia-l4)
34+
* [Inference performance: NVIDIA L40S](#inference-performance-nvidia-l40s)
3435

3536

3637
## Model overview
@@ -74,8 +75,8 @@ The following software version configuration has been tested:
7475
|Software|Version|
7576
|--------|-------|
7677
|Python|>=3.8|
77-
|TensorRT|10.1.0.27|
78-
|CUDA|12.4|
78+
|TensorRT|10.2.0.19|
79+
|CUDA|12.5|
7980

8081
## Setup
8182

‎demo/DeBERTa/README.md

+3-3
Original file line numberDiff line numberDiff line change
@@ -75,7 +75,7 @@ Note that the performance gap between BERT's self-attention and DeBERTa's disent
7575
## Environment Setup
7676
It is recommended to use docker for reproducing the following steps. Follow the setup steps in TensorRT OSS [README](https://github.com/NVIDIA/TensorRT#setting-up-the-build-environment) to build and launch the container and build OSS:
7777

78-
**Example: Ubuntu 20.04 on x86-64 with cuda-12.4 (default)**
78+
**Example: Ubuntu 20.04 on x86-64 with cuda-12.5 (default)**
7979
```bash
8080
# Download this TensorRT OSS repo
8181
git clone -b main https://github.com/nvidia/TensorRT TensorRT
@@ -84,10 +84,10 @@ git submodule update --init --recursive
8484

8585
## at root of TensorRT OSS
8686
# build container
87-
./docker/build.sh --file docker/ubuntu-20.04.Dockerfile --tag tensorrt-ubuntu20.04-cuda12.4
87+
./docker/build.sh --file docker/ubuntu-20.04.Dockerfile --tag tensorrt-ubuntu20.04-cuda12.5
8888

8989
# launch container
90-
./docker/launch.sh --tag tensorrt-ubuntu20.04-cuda12.4 --gpus all
90+
./docker/launch.sh --tag tensorrt-ubuntu20.04-cuda12.5 --gpus all
9191

9292
## now inside container
9393
# build OSS (only required for pre-8.4.3 TensorRT versions)

‎demo/Diffusion/README.md

+2-2
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@ This demo application ("demoDiffusion") showcases the acceleration of Stable Dif
77
### Clone the TensorRT OSS repository
88

99
```bash
10-
git clone [email protected]:NVIDIA/TensorRT.git -b release/10.1 --single-branch
10+
git clone [email protected]:NVIDIA/TensorRT.git -b release/10.2 --single-branch
1111
cd TensorRT
1212
```
1313

@@ -48,7 +48,7 @@ onnx 1.15.0
4848
onnx-graphsurgeon 0.5.2
4949
onnxruntime 1.16.3
5050
polygraphy 0.49.9
51-
tensorrt 10.1.0.27
51+
tensorrt 10.2.0.19
5252
tokenizers 0.13.3
5353
torch 2.2.0
5454
transformers 4.33.1

‎docker/rockylinux8.Dockerfile

+10-10
Original file line numberDiff line numberDiff line change
@@ -15,7 +15,7 @@
1515
# limitations under the License.
1616
#
1717

18-
ARG CUDA_VERSION=12.4.0
18+
ARG CUDA_VERSION=12.5.0
1919

2020
FROM nvidia/cuda:${CUDA_VERSION}-devel-rockylinux8
2121
LABEL maintainer="NVIDIA CORPORATION"
@@ -25,7 +25,7 @@ ENV NV_CUDNN_VERSION 8.9.6.50-1
2525
ENV NV_CUDNN_PACKAGE libcudnn8-${NV_CUDNN_VERSION}.cuda12.2
2626
ENV NV_CUDNN_PACKAGE_DEV libcudnn8-devel-${NV_CUDNN_VERSION}.cuda12.2
2727

28-
ENV TRT_VERSION 10.1.0.27
28+
ENV TRT_VERSION 10.2.0.19
2929
SHELL ["/bin/bash", "-c"]
3030

3131
RUN dnf install -y \
@@ -62,15 +62,15 @@ RUN dnf install -y python38 python38-devel &&\
6262

6363
# Install TensorRT
6464
RUN if [ "${CUDA_VERSION:0:2}" = "11" ]; then \
65-
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz \
66-
&& tar -xf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz \
67-
&& cp -a TensorRT-10.1.0.27/lib/*.so* /usr/lib64 \
68-
&& pip install TensorRT-10.1.0.27/python/tensorrt-10.1.0-cp38-none-linux_x86_64.whl ;\
65+
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz \
66+
&& tar -xf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz \
67+
&& cp -a TensorRT-10.2.0.19/lib/*.so* /usr/lib64 \
68+
&& pip install TensorRT-10.2.0.19/python/tensorrt-10.2.0-cp38-none-linux_x86_64.whl ;\
6969
elif [ "${CUDA_VERSION:0:2}" = "12" ]; then \
70-
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz \
71-
&& tar -xf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz \
72-
&& cp -a TensorRT-10.1.0.27/lib/*.so* /usr/lib64 \
73-
&& pip install TensorRT-10.1.0.27/python/tensorrt-10.1.0-cp38-none-linux_x86_64.whl ;\
70+
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz \
71+
&& tar -xf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz \
72+
&& cp -a TensorRT-10.2.0.19/lib/*.so* /usr/lib64 \
73+
&& pip install TensorRT-10.2.0.19/python/tensorrt-10.2.0-cp38-none-linux_x86_64.whl ;\
7474
else \
7575
echo "Invalid CUDA_VERSION"; \
7676
exit 1; \

‎docker/rockylinux9.Dockerfile

+10-10
Original file line numberDiff line numberDiff line change
@@ -15,7 +15,7 @@
1515
# limitations under the License.
1616
#
1717

18-
ARG CUDA_VERSION=12.4.0
18+
ARG CUDA_VERSION=12.5.0
1919

2020
FROM nvidia/cuda:${CUDA_VERSION}-devel-rockylinux9
2121
LABEL maintainer="NVIDIA CORPORATION"
@@ -25,7 +25,7 @@ ENV NV_CUDNN_VERSION 8.9.6.50-1
2525
ENV NV_CUDNN_PACKAGE libcudnn8-${NV_CUDNN_VERSION}.cuda12.2
2626
ENV NV_CUDNN_PACKAGE_DEV libcudnn8-devel-${NV_CUDNN_VERSION}.cuda12.2
2727

28-
ENV TRT_VERSION 10.1.0.27
28+
ENV TRT_VERSION 10.2.0.19
2929
SHELL ["/bin/bash", "-c"]
3030

3131
RUN dnf install -y \
@@ -67,15 +67,15 @@ RUN dnf -y install \
6767

6868
# Install TensorRT
6969
RUN if [ "${CUDA_VERSION:0:2}" = "11" ]; then \
70-
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz \
71-
&& tar -xf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz \
72-
&& cp -a TensorRT-10.1.0.27/lib/*.so* /usr/lib64 \
73-
&& pip install TensorRT-10.1.0.27/python/tensorrt-10.1.0-cp39-none-linux_x86_64.whl ;\
70+
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz \
71+
&& tar -xf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz \
72+
&& cp -a TensorRT-10.2.0.19/lib/*.so* /usr/lib64 \
73+
&& pip install TensorRT-10.2.0.19/python/tensorrt-10.2.0-cp39-none-linux_x86_64.whl ;\
7474
elif [ "${CUDA_VERSION:0:2}" = "12" ]; then \
75-
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz \
76-
&& tar -xf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz \
77-
&& cp -a TensorRT-10.1.0.27/lib/*.so* /usr/lib64 \
78-
&& pip install TensorRT-10.1.0.27/python/tensorrt-10.1.0-cp39-none-linux_x86_64.whl ;\
75+
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz \
76+
&& tar -xf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz \
77+
&& cp -a TensorRT-10.2.0.19/lib/*.so* /usr/lib64 \
78+
&& pip install TensorRT-10.2.0.19/python/tensorrt-10.2.0-cp39-none-linux_x86_64.whl ;\
7979
else \
8080
echo "Invalid CUDA_VERSION"; \
8181
exit 1; \

‎docker/ubuntu-20.04.Dockerfile

+10-10
Original file line numberDiff line numberDiff line change
@@ -15,7 +15,7 @@
1515
# limitations under the License.
1616
#
1717

18-
ARG CUDA_VERSION=12.4.0
18+
ARG CUDA_VERSION=12.5.0
1919

2020
FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu20.04
2121
LABEL maintainer="NVIDIA CORPORATION"
@@ -28,7 +28,7 @@ ENV CUDA_VERSION_MAJOR_MINOR=12.2
2828
ENV NV_CUDNN_PACKAGE "libcudnn8=$NV_CUDNN_VERSION-1+cuda${CUDA_VERSION_MAJOR_MINOR}"
2929
ENV NV_CUDNN_PACKAGE_DEV "libcudnn8-dev=$NV_CUDNN_VERSION-1+cuda${CUDA_VERSION_MAJOR_MINOR}"
3030

31-
ENV TRT_VERSION 10.1.0.27
31+
ENV TRT_VERSION 10.2.0.19
3232
SHELL ["/bin/bash", "-c"]
3333

3434
RUN apt-get update && apt-get install -y --no-install-recommends \
@@ -84,15 +84,15 @@ RUN apt-get install -y --no-install-recommends \
8484

8585
# Install TensorRT
8686
RUN if [ "${CUDA_VERSION:0:2}" = "11" ]; then \
87-
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz \
88-
&& tar -xf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-11.8.tar.gz \
89-
&& cp -a TensorRT-10.1.0.27/lib/*.so* /usr/lib/x86_64-linux-gnu \
90-
&& pip install TensorRT-10.1.0.27/python/tensorrt-10.1.0-cp38-none-linux_x86_64.whl ;\
87+
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz \
88+
&& tar -xf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-11.8.tar.gz \
89+
&& cp -a TensorRT-10.2.0.19/lib/*.so* /usr/lib/x86_64-linux-gnu \
90+
&& pip install TensorRT-10.2.0.19/python/tensorrt-10.2.0-cp38-none-linux_x86_64.whl ;\
9191
elif [ "${CUDA_VERSION:0:2}" = "12" ]; then \
92-
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.1.0/tars/TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz \
93-
&& tar -xf TensorRT-10.1.0.27.Linux.x86_64-gnu.cuda-12.4.tar.gz \
94-
&& cp -a TensorRT-10.1.0.27/lib/*.so* /usr/lib/x86_64-linux-gnu \
95-
&& pip install TensorRT-10.1.0.27/python/tensorrt-10.1.0-cp38-none-linux_x86_64.whl ;\
92+
wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.2.0/tars/TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz \
93+
&& tar -xf TensorRT-10.2.0.19.Linux.x86_64-gnu.cuda-12.5.tar.gz \
94+
&& cp -a TensorRT-10.2.0.19/lib/*.so* /usr/lib/x86_64-linux-gnu \
95+
&& pip install TensorRT-10.2.0.19/python/tensorrt-10.2.0-cp38-none-linux_x86_64.whl ;\
9696
else \
9797
echo "Invalid CUDA_VERSION"; \
9898
exit 1; \

‎docker/ubuntu-22.04-aarch64.Dockerfile

+3-3
Original file line numberDiff line numberDiff line change
@@ -15,12 +15,12 @@
1515
# limitations under the License.
1616
#
1717

18-
ARG CUDA_VERSION=12.4.0
18+
ARG CUDA_VERSION=12.5.0
1919

2020
# Multi-arch container support available in non-cudnn containers.
2121
FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
2222

23-
ENV TRT_VERSION 10.1.0.27
23+
ENV TRT_VERSION 10.2.0.19
2424
SHELL ["/bin/bash", "-c"]
2525

2626
# Setup user account
@@ -71,7 +71,7 @@ RUN apt-get install -y --no-install-recommends \
7171
# Install TensorRT. This will also pull in CUDNN
7272
RUN ver="${CUDA_VERSION%.*}" &&\
7373
if [ "${ver%.*}" = "12" ] ; then \
74-
ver="12.4"; \
74+
ver="12.5"; \
7575
fi &&\
7676
v="${TRT_VERSION}-1+cuda${ver}" &&\
7777
apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/sbsa/3bf863cc.pub &&\

0 commit comments

Comments
 (0)