[TensorRT EP] support TensorRT 10.7-GA (microsoft#23011)

### Description
<!-- Describe your changes. -->
Update CIs to TRT10.7

### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->

Author: yf711

cgmanifests/generated/cgmanifest.json

@@ -196,7 +196,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "bc0d2e35909b8456abe32f3b30a49bb0c125e8b7",
+          "commitHash": "9c69a24bc2e20c8a511a4e6b06fd49639ec5300a",
           "repositoryUrl": "https://github.com/onnx/onnx-tensorrt.git"
         },
         "comments": "onnx_tensorrt"

cmake/deps.txt

@@ -36,8 +36,8 @@ microsoft_wil;https://github.com/microsoft/wil/archive/refs/tags/v1.0.230629.1.z
 mimalloc;https://github.com/microsoft/mimalloc/archive/refs/tags/v2.1.1.zip;d5ee7d34223d0567892db5179849939c8769dc41
 mp11;https://github.com/boostorg/mp11/archive/refs/tags/boost-1.82.0.zip;9bc9e01dffb64d9e0773b2e44d2f22c51aace063
 onnx;https://github.com/onnx/onnx/archive/refs/tags/v1.16.1.zip;2eb9198bb352757d5ff13977cbe0634898e0837c
-# Use the latest commit of 10.6-GA-ORT-DDS
-onnx_tensorrt;https://github.com/onnx/onnx-tensorrt/archive/bc0d2e35909b8456abe32f3b30a49bb0c125e8b7.zip;f233ae871ad82c023da62e5dd620639f00bc2d15
+# Use the latest commit of 10.7-GA
+onnx_tensorrt;https://github.com/onnx/onnx-tensorrt/archive/9c69a24bc2e20c8a511a4e6b06fd49639ec5300a.zip;ff1fe9af78eb129b4a4cdcb7450b7390b4436dd3
 protobuf;https://github.com/protocolbuffers/protobuf/archive/refs/tags/v21.12.zip;7cf2733949036c7d52fda017badcab093fe73bfa
 protoc_win64;https://github.com/protocolbuffers/protobuf/releases/download/v21.12/protoc-21.12-win64.zip;b4521f7ada5b260380f94c4bd7f1b7684c76969a
 protoc_win32;https://github.com/protocolbuffers/protobuf/releases/download/v21.12/protoc-21.12-win32.zip;3688010318192c46ce73213cdfb6b3e5656da874

onnxruntime/python/tools/tensorrt/perf/build/build_image.py

@@ -15,10 +15,10 @@
 from typing import List, Optional
 
 TRT_DOCKER_FILES = {
-    "8.6.cuda_11_8_cudnn_8": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_8_tensorrt8_6",
-    "8.6.cuda_12_3_cudnn_9": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_3_tensorrt8_6",
-    "10.5.cuda_11_8_cudnn_8": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10",
-    "10.5.cuda_12_5_cudnn_9": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10",
+    "8.6_cuda11.8_cudnn8": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_8_tensorrt8_6",
+    "8.6_cuda12.3_cudnn9": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_3_tensorrt8_6",
+    "10.7_cuda11.8_cudnn8": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10",
+    "10.7_cuda12.5_cudnn9": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10",
     "BIN": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_tensorrt_bin",
 }
 

tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml

@@ -8,12 +8,12 @@ parameters:
 - name: TrtVersion
   displayName: TensorRT Version
   type: string
-  default: 10.5.cuda_12_5_cudnn_9
+  default: 10.7_cuda12.5_cudnn9
   values:
-  - 8.6.cuda_11_8_cudnn_8
-  - 8.6.cuda_12_3_cudnn_9
-  - 10.5.cuda_11_8_cudnn_8
-  - 10.5.cuda_12_5_cudnn_9
+  - 8.6_cuda11.8_cudnn8
+  - 8.6_cuda12.3_cudnn9
+  - 10.7_cuda11.8_cudnn8
+  - 10.7_cuda12.5_cudnn9
   - BIN
 
 - name: UseTensorrtOssParser
@@ -198,4 +198,4 @@ jobs:
       parameters :
         condition : 'succeeded'
 
-    - template: templates/clean-agent-build-directory-step.yml
+    - template: templates/clean-agent-build-directory-step.yml

tools/ci_build/github/azure-pipelines/py-cuda-alt-package-test-pipeline.yml

@@ -19,6 +19,6 @@ stages:
           python_wheel_suffix: '_gpu'
           timeout: 480
           docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241120.3
-          trt_version: '10.6.0.26-1.cuda11.8'
+          trt_version: '10.7.0.23-1.cuda11.8'
           cuda_version: '11.8'
 

tools/ci_build/github/azure-pipelines/templates/common-variables.yml

@@ -1,5 +1,5 @@
 variables:
-  common_trt_version: '10.6.0.26'
+  common_trt_version: '10.7.0.23'
   # As for Debian installation, replace '-1.' by '-1+' when assigning trt version below
   linux_trt_version_cuda11: ${{ variables.common_trt_version }}-1.cuda11.8
   linux_trt_version_cuda12: ${{ variables.common_trt_version }}-1.cuda12.6

tools/ci_build/github/azure-pipelines/templates/download-deps.yml

@@ -11,7 +11,7 @@ steps:
       packageType: upack
       feed: '/7424c8e4-5c62-490e-95c4-79446f31017c'
       definition: '517c4f6f-5437-4392-a70d-4f15ec5be2f0'
-      version: 1.0.201
+      version: 1.0.202
       downloadPath: $(Build.BinariesDirectory)/deps
 
 # The private ADO project
@@ -22,7 +22,7 @@ steps:
       packageType: upack
       feed: '/4c7631f5-24c0-4307-8822-1aa8f180c325'
       definition: 'fd9dd5ad-b73e-4678-890e-edcf680dbc1a'
-      version: 1.0.201
+      version: 1.0.202
       downloadPath: $(Build.BinariesDirectory)/deps
 
 # You can add more ADO accounts at here.

tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml

@@ -13,10 +13,10 @@ parameters:
       - 12.2
   - name: TrtVersion
     type: string
-    default: '10.6.0.26'
+    default: '10.7.0.23'
     values:
       - 8.6.1.6
-      - 10.6.0.26
+      - 10.7.0.23
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, true) }}:
@@ -42,7 +42,7 @@ steps:
         - powershell: |
             Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.0"
           displayName: Set trtCudaVersion
-    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '10.6.0.26')) }}:
+    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '10.7.0.23')) }}:
         - powershell: |
             Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.6"
           displayName: Set trtCudaVersion

tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml

@@ -15,10 +15,10 @@ parameters:
     default: '11.8'
   - name: win_trt_folder_cuda11
     type: string
-    default: 'TensorRT-10.6.0.26.Windows10.x86_64.cuda-11.8'
+    default: 'TensorRT-10.7.0.23.Windows10.x86_64.cuda-11.8'
   - name: win_trt_folder_cuda12
     type: string
-    default: 'TensorRT-10.6.0.26.Windows10.x86_64.cuda-12.6'
+    default: 'TensorRT-10.7.0.23.Windows10.x86_64.cuda-12.6'
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:

tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0

@@ -6,7 +6,7 @@
 
 # Build base image with required system packages
 ARG BASEIMAGE=nvidia/cuda:12.5.1-cudnn-devel-ubi8
-ARG TRT_VERSION=10.6.0.26-1.cuda12.6
+ARG TRT_VERSION=10.7.0.23-1.cuda12.6
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION
 ENV PATH=/opt/python/cp310-cp310/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/src/tensorrt/bin:${PATH}

tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0_torch

@@ -6,7 +6,7 @@
 
 # Build base image with required system packages
 ARG BASEIMAGE=nvidia/cuda:11.8.0-cudnn8-devel-ubi8
-ARG TRT_VERSION=10.6.0.26-1.cuda11.8
+ARG TRT_VERSION=10.7.0.23-1.cuda11.8
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION
 ENV PATH=/opt/python/cp310-cp310/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/src/tensorrt/bin:${PATH}

tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu

@@ -6,7 +6,7 @@
 
 # Build base image with required system packages
 ARG BASEIMAGE=nvidia/cuda:11.8.0-cudnn8-devel-ubuntu20.04
-ARG TRT_VERSION=10.6.0.26-1+cuda11.8
+ARG TRT_VERSION=10.7.0.23-1+cuda11.8
 ARG LD_LIBRARY_PATH_ARG=/usr/local/lib64:/usr/local/cuda/lib64
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION

tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2204_gpu_ffmpeg

@@ -6,7 +6,7 @@
 
 # Build base image with required system packages
 ARG BASEIMAGE=nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04
-ARG TRT_VERSION=10.6.0.26-1+cuda11.8
+ARG TRT_VERSION=10.7.0.23-1+cuda11.8
 ARG LD_LIBRARY_PATH_ARG=/usr/local/lib64:/usr/local/cuda/lib64
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION

tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2204_gpu_opencv

@@ -6,7 +6,7 @@
 
 # Build base image with required system packages
 ARG BASEIMAGE=nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04
-ARG TRT_VERSION=10.6.0.26-1+cuda11.8
+ARG TRT_VERSION=10.7.0.23-1+cuda11.8
 ARG LD_LIBRARY_PATH_ARG=/usr/local/lib64:/usr/local/cuda/lib64
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION

tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10

@@ -31,7 +31,7 @@ RUN pip install --upgrade pip
 RUN pip install psutil setuptools>=68.2.2
 
 # Install TensorRT
-RUN TRT_VERSION="10.6.0.26-1+cuda11.8" &&\
+RUN TRT_VERSION="10.7.0.23-1+cuda11.8" &&\
     apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/7fa2af80.pub &&\
     apt-get update &&\
     apt-get install -y \

tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10

@@ -31,7 +31,7 @@ RUN pip install --upgrade pip
 RUN pip install setuptools>=68.2.2 psutil
 
 # Install TensorRT
-RUN TRT_VERSION="10.6.0.26-1+cuda12.6" &&\
+RUN TRT_VERSION="10.7.0.23-1+cuda12.6" &&\
     apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/7fa2af80.pub &&\
     apt-get update &&\
     apt-get install -y \

tools/ci_build/github/linux/docker/inference/x86_64/python/cuda/Dockerfile

@@ -5,7 +5,7 @@
 ARG BASEIMAGE=nvidia/cuda:11.8.0-cudnn8-devel-ubi8
 
 FROM $BASEIMAGE
-ARG TRT_VERSION=10.6.0.26-1.cuda11.8
+ARG TRT_VERSION=10.7.0.23-1.cuda11.8
 
 #Install TensorRT only if TRT_VERSION is not empty
 RUN if [ -n "${TRT_VERSION}" ]; then  \

tools/ci_build/github/windows/setup_env_gpu.bat

@@ -6,10 +6,10 @@ if exist PATH=%AGENT_TEMPDIRECTORY%\v12.2\ (
 ) else (
     set PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2\bin;C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2\extras\CUPTI\lib64;%PATH%
 )
-set PATH=%AGENT_TEMPDIRECTORY%\TensorRT-10.6.0.26.Windows10.x86_64.cuda-12.6\lib;%PATH%
+set PATH=%AGENT_TEMPDIRECTORY%\TensorRT-10.7.0.23.Windows10.x86_64.cuda-12.6\lib;%PATH%
 
 @REM The default version is still cuda v12.2, because set cuda v11.8 after it
-set PATH=%PATH%;%AGENT_TEMPDIRECTORY%\TensorRT-10.6.0.26.Windows10.x86_64.cuda-11.8\lib
+set PATH=%PATH%;%AGENT_TEMPDIRECTORY%\TensorRT-10.7.0.23.Windows10.x86_64.cuda-11.8\lib
 if exist PATH=%AGENT_TEMPDIRECTORY%\v11.8\ (
     set PATH=%PATH%;%AGENT_TEMPDIRECTORY%\v11.8\bin;%AGENT_TEMPDIRECTORY%\v11.8\extras\CUPTI\lib64
 ) else (

tools/ci_build/github/windows/setup_env_trt.bat

@@ -6,6 +6,6 @@ if exist PATH=%AGENT_TEMPDIRECTORY%\v12.2\ (
 ) else (
     set PATH=%PATH%;C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2\bin;C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2\extras\CUPTI\lib64
 )
-set PATH=%AGENT_TEMPDIRECTORY%\TensorRT-10.6.0.26.Windows10.x86_64.cuda-12.6\lib;%PATH%
+set PATH=%AGENT_TEMPDIRECTORY%\TensorRT-10.7.0.23.Windows10.x86_64.cuda-12.6\lib;%PATH%
 set GRADLE_OPTS=-Dorg.gradle.daemon=false
 set CUDA_MODULE_LOADING=LAZY