TensorRT Release 10.14 (#4633)

Signed-off-by: Asfiya Baig <asfiyab@nvidia.com>

Author: asfiyab-nvidia

CHANGELOG.md

@@ -1,11 +1,27 @@
 # TensorRT OSS Release Changelog
 
+## 10.14 GA - 2025-11-7
+- Sample changes
+  - Replace all pycuda usages with cuda-python APIs
+  - Removed the efficientnet samples
+  - Deprecated tensorflow_object_detection and efficientdet samples
+  - Samples will no longer be released with the packages. The TensorRT GitHub repository will be the single source.
+
+
+- Parsers:
+  - Added support for the `Attention` operator
+  - Improved refit for `ConstantOfShape` nodes
+
+- Demos
+  - demoDiffusion:
+    - Added support for the Cosmos-Predict2 text2image and video2world pipelines
+
+
 ## 10.13.3 GA - 2025-9-8
 - Added support for TensorRT API Capture and Replay feature, see the [developer guide](https://docs.nvidia.com/deeplearning/tensorrt/latest/inference-library/advanced.html) for more information.
 - Demo changes
   - Added support for Flux Kontext pipeline.
 
-
 ## 10.13.2 GA - 2025-8-18
 - Added support for CUDA 13.0, dropped support for CUDA 11.X
 - Dropped support for Ubuntu 20.04
@@ -24,7 +40,6 @@
   - Added `loadModelProto`, `loadInitializer` and `refitModelProto` APIs for IParserRefitter. These APIs are meant to be used to load user initializers when refitting ONNX models.
   - Deprecated `IParser::parseWithWeightDescriptors`.
 
-
 ## 10.12.0 GA - 2025-6-10
 - Plugin changes
   - Migrated `IPluginV2`-descendent version 1 of `cropAndResizeDynamic`, to version 2, which implements `IPluginV3`.

CMakeLists.txt

@@ -67,6 +67,9 @@ endif()
 set(CMAKE_SKIP_BUILD_RPATH True)
 
 # CUDA targets
+set(DEFAULT_CUDA_VERSION 13.0.0)
+set_ifndef(CUDA_VERSION ${DEFAULT_CUDA_VERSION})
+message(STATUS "CUDA version set to ${CUDA_VERSION}")
 
 if (DEFINED GPU_ARCHS)
   message(STATUS "GPU_ARCHS defined as ${GPU_ARCHS}. Generating CUDA code for SM ${GPU_ARCHS}")
@@ -81,6 +84,9 @@ else()
       list(APPEND CMAKE_CUDA_ARCHITECTURES 100 120)
   endif()
 
+  if(CUDA_VERSION VERSION_GREATER_EQUAL 13.0)
+        list(APPEND CMAKE_CUDA_ARCHITECTURES 110)
+  endif()
   message(STATUS "GPU_ARCHS is not defined. Generating CUDA code for default SMs: ${CMAKE_CUDA_ARCHITECTURES}")
 endif()
 set(BERT_GENCODES)
@@ -105,6 +111,10 @@ project(TensorRT
         DESCRIPTION "TensorRT is a C++ library that facilitates high-performance inference on NVIDIA GPUs and deep learning accelerators."
         HOMEPAGE_URL "https://github.com/NVIDIA/TensorRT")
 
+if (WIN32)
+    enable_language(C)
+endif()
+
 if(CMAKE_INSTALL_PREFIX_INITIALIZED_TO_DEFAULT)
   set(CMAKE_INSTALL_PREFIX ${TRT_LIB_DIR}/../ CACHE PATH "TensorRT installation" FORCE)
 endif(CMAKE_INSTALL_PREFIX_INITIALIZED_TO_DEFAULT)
@@ -113,7 +123,7 @@ option(BUILD_PLUGINS "Build TensorRT plugin" ON)
 option(BUILD_PARSERS "Build TensorRT parsers" ON)
 option(BUILD_SAMPLES "Build TensorRT samples" ON)
 
-# C++14
+# C++17
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_CXX_EXTENSIONS OFF)
@@ -141,14 +151,10 @@ endif()
 
 ############################################################################################
 # Dependencies
-
-set(DEFAULT_CUDA_VERSION 12.2.0)
 set(DEFAULT_CUDNN_VERSION 8.9)
-set(DEFAULT_PROTOBUF_VERSION 3.20.1)
+set(DEFAULT_PROTOBUF_VERSION 3.20.3)
 
 # Dependency Version Resolution
-set_ifndef(CUDA_VERSION ${DEFAULT_CUDA_VERSION})
-message(STATUS "CUDA version set to ${CUDA_VERSION}")
 set_ifndef(CUDNN_VERSION ${DEFAULT_CUDNN_VERSION})
 message(STATUS "cuDNN version set to ${CUDNN_VERSION}")
 set_ifndef(PROTOBUF_VERSION ${DEFAULT_PROTOBUF_VERSION})
@@ -221,16 +227,18 @@ endif()
 ############################################################################################
 # TensorRT
 
+set(HINT_PATHS "${TRT_OUT_DIR}" "${TRT_LIB_DIR}")
+
 if(BUILD_PLUGINS)
     add_subdirectory(plugin)
 else()
-    find_library_create_target(nvinfer_plugin ${nvinfer_plugin_lib_name} SHARED "${TRT_OUT_DIR}" "${TRT_LIB_DIR}")
+    find_library_create_target(${nvinfer_plugin_lib_name} ${nvinfer_plugin_lib_name} SHARED "${HINT_PATHS}")
 endif()
 
 if(BUILD_PARSERS)
     add_subdirectory(parsers)
 else()
-    find_library_create_target(nvonnxparser ${nvonnxparser_lib_name} SHARED "${TRT_OUT_DIR}" "${TRT_LIB_DIR}")
+    find_library_create_target(${nvonnxparser_lib_name} ${nvonnxparser_lib_name} SHARED "${HINT_PATHS}")
 endif()
 
 if(BUILD_SAMPLES)

README.md

@@ -32,7 +32,7 @@ To build the TensorRT-OSS components, you will first need the following software
 
 **TensorRT GA build**
 
-- TensorRT v10.13.3.9
+- TensorRT v10.14.1.48
   - Available from direct download links listed below
 
 **System Packages**
@@ -86,24 +86,24 @@ To build the TensorRT-OSS components, you will first need the following software
 
    Else download and extract the TensorRT GA build from [NVIDIA Developer Zone](https://developer.nvidia.com) with the direct links below:
 
-   - [TensorRT 10.13.3.9 for CUDA 13.0, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.13.3/tars/TensorRT-10.13.3.9.Linux.x86_64-gnu.cuda-13.0.tar.gz)
-   - [TensorRT 10.13.3.9 for CUDA 12.9, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.13.3/tars/TensorRT-10.13.3.9.Linux.x86_64-gnu.cuda-12.9.tar.gz)
-   - [TensorRT 10.13.3.9 for CUDA 13.0, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.13.3/zip/TensorRT-10.13.3.9.Windows.win10.cuda-13.0.zip)
-   - [TensorRT 10.13.3.9 for CUDA 12.9, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.13.3/zip/TensorRT-10.13.3.9.Windows.win10.cuda-12.9.zip)
+   - [TensorRT 10.14.1.48 for CUDA 13.0, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.14.1/tars/TensorRT-10.14.1.48.Linux.x86_64-gnu.cuda-13.0.tar.gz)
+   - [TensorRT 10.14.1.48 for CUDA 12.9, Linux x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.14.1/tars/TensorRT-10.14.1.48.Linux.x86_64-gnu.cuda-12.9.tar.gz)
+   - [TensorRT 10.14.1.48 for CUDA 13.0, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.14.1/zip/TensorRT-10.14.1.48.Windows.win10.cuda-13.0.zip)
+   - [TensorRT 10.14.1.48 for CUDA 12.9, Windows x86_64](https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.14.1/zip/TensorRT-10.14.1.48.Windows.win10.cuda-12.9.zip)
 
    **Example: Ubuntu 22.04 on x86-64 with cuda-13.0**
 
    ```bash
    cd ~/Downloads
-   tar -xvzf TensorRT-10.13.3.9.Linux.x86_64-gnu.cuda-13.0.tar.gz
-   export TRT_LIBPATH=`pwd`/TensorRT-10.13.3.9
+   tar -xvzf TensorRT-10.14.1.48.Linux.x86_64-gnu.cuda-13.0.tar.gz
+   export TRT_LIBPATH=`pwd`/TensorRT-10.14.1.48
    ```
 
    **Example: Windows on x86-64 with cuda-12.9**
 
    ```powershell
-   Expand-Archive -Path TensorRT-10.13.3.9.Windows.win10.cuda-12.9.zip
-   $env:TRT_LIBPATH="$pwd\TensorRT-10.13.3.9\lib"
+   Expand-Archive -Path TensorRT-10.14.1.48.Windows.win10.cuda-12.9.zip
+   $env:TRT_LIBPATH="$pwd\TensorRT-10.14.1.48\lib"
    ```
 
 ## Setting Up The Build Environment
@@ -112,16 +112,16 @@ For Linux platforms, we recommend that you generate a docker container for build
 
 1. #### Generate the TensorRT-OSS build container.
 
-   **Example: Ubuntu 22.04 on x86-64 with cuda-13.0 (default)**
+   **Example: Ubuntu 24.04 on x86-64 with cuda-13.0 (default)**
 
    ```bash
-   ./docker/build.sh --file docker/ubuntu-22.04.Dockerfile --tag tensorrt-ubuntu22.04-cuda13.0
+   ./docker/build.sh --file docker/ubuntu-24.04.Dockerfile --tag tensorrt-ubuntu24.04-cuda13.0
    ```
 
-   **Example: Rockylinux8 on x86-64 with cuda-12.9**
+   **Example: Rockylinux8 on x86-64 with cuda-13.0**
 
    ```bash
-   ./docker/build.sh --file docker/rockylinux8.Dockerfile --tag tensorrt-rockylinux8-cuda12.9
+   ./docker/build.sh --file docker/rockylinux8.Dockerfile --tag tensorrt-rockylinux8-cuda13.0
    ```
 
    **Example: Ubuntu 24.04 cross-compile for Jetson (aarch64) with cuda-13.0 (JetPack SDK)**
@@ -137,9 +137,9 @@ For Linux platforms, we recommend that you generate a docker container for build
    ```
 
 2. #### Launch the TensorRT-OSS build container.
-   **Example: Ubuntu 22.04 build container**
+   **Example: Ubuntu 24.04 build container**
    ```bash
-   ./docker/launch.sh --tag tensorrt-ubuntu22.04-cuda13.0 --gpus all
+   ./docker/launch.sh --tag tensorrt-ubuntu24.04-cuda13.0 --gpus all
    ```
    > NOTE:
    > <br> 1. Use the `--tag` corresponding to build container generated in Step 1.
@@ -175,7 +175,7 @@ For Linux platforms, we recommend that you generate a docker container for build
   ```bash
   cd $TRT_OSSPATH
   mkdir -p build && cd build
-  cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DTRT_OUT_DIR=`pwd`/out -DTRT_PLATFORM_ID=aarch64 -DGPU_ARCHS=110
+  cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DTRT_OUT_DIR=`pwd`/out -DTRT_PLATFORM_ID=aarch64
   CC=/usr/bin/gcc make -j$(nproc)
   ```
 
@@ -186,7 +186,16 @@ For Linux platforms, we recommend that you generate a docker container for build
   ```bash
   cd $TRT_OSSPATH
   mkdir -p build && cd build
-  cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DCMAKE_TOOLCHAIN_FILE=$TRT_OSSPATH/cmake/toolchains/cmake_aarch64_cross.toolchain -DGPU_ARCHS=110
+  cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DCMAKE_TOOLCHAIN_FILE=$TRT_OSSPATH/cmake/toolchains/cmake_aarch64_cross.toolchain
+  make -j$(nproc)
+  ```
+
+  **Example: Ubuntu 24.04 Cross-Compile for DriveOS (aarch64) with cuda-13.0**
+
+  ```bash
+  cd $TRT_OSSPATH
+  mkdir -p build && cd build
+  cmake .. -DTRT_LIB_DIR=$TRT_LIBPATH -DCMAKE_TOOLCHAIN_FILE=$TRT_OSSPATH/cmake/toolchains/cmake_aarch64_dos_cross.toolchain
   make -j$(nproc)
   ```
 
@@ -196,7 +205,7 @@ For Linux platforms, we recommend that you generate a docker container for build
   cd $TRT_OSSPATH
   mkdir -p build
   cd -p build
-  cmake .. -DTRT_LIB_DIR="$env:TRT_LIBPATH" -DCUDNN_ROOT_DIR="$env:CUDNN_PATH" -DTRT_OUT_DIR="$pwd\\out"
+  cmake .. -DTRT_LIB_DIR="$env:TRT_LIBPATH" -DTRT_OUT_DIR="$pwd\\out"
   msbuild TensorRT.sln /property:Configuration=Release -m:$env:NUMBER_OF_PROCESSORS
   ```
 

VERSION

@@ -1 +1 @@
-10.13.3.9
+10.14.1.48