From 0d33602ef0ff6971a92d634c92f6963b96f98077 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 10 Oct 2024 17:23:57 -0700
Subject: [PATCH 01/44] test

---
 .../github/azure-pipelines/bigmodels-ci-pipeline.yml     | 8 ++++----
 .../azure-pipelines/templates/common-variables.yml       | 9 +++++++--
 2 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
index 6fd02f6b59867..edeaa28fe4dbf 100644
--- a/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
@@ -42,11 +42,11 @@ parameters:
 variables:
   - template: templates/common-variables.yml
   - name: docker_base_image
-    value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
+    value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
   - name: linux_trt_version
-    value: 10.3.0.26-1.cuda11.8
+    value: ${{ variables.linux_trt_version_cuda12 }}
   - name: Repository
-    value: 'onnxruntimecuda11manylinuxbuild'
+    value: 'onnxruntimecuda12manylinuxbuild'
 
 stages:
 - stage: Build_Onnxruntime_Cuda
@@ -123,7 +123,7 @@ stages:
                 --build_shared_lib \
                 --parallel \
                 --build_wheel \
-                --enable_onnx_tests --use_cuda --cuda_version=${{variables.common_cuda_version}} --cuda_home=/usr/local/cuda-${{variables.common_cuda_version}} --cudnn_home=/usr/local/cuda-${{variables.common_cuda_version}} \
+                --enable_onnx_tests --use_cuda --cuda_version=12.2 --cuda_home=/usr/local/cuda-12 --cudnn_home=/usr/local/cuda-12 \
                 --enable_cuda_profiling --enable_cuda_nhwc_ops \
                 --enable_pybind --build_java \
                 --use_cache \
diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index e7f703fa592a3..007ae16a62be1 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -1,3 +1,8 @@
 variables:
-  common_cuda_version: '11.8'
-  common_cuda_baseimg: 'nvidia/cuda:11.8.0-cudnn8-devel-ubi8'
+  common_trt_version: '10.4.0.26'
+  linux_trt_version_cuda11: ${{ variables.common_trt_version }}-1.cuda11.8
+  linux_trt_version_cuda12: ${{ variables.common_trt_version }}-1.cuda12.6
+  common_linux_trt_version: ${{ variables.linux_trt_version_cuda11 }}
+  win_trt_folder_cuda11: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-11.8
+  win_trt_folder_cuda12: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-12.6
+  common_win_trt_folder: ${{ variables.win_trt_folder_cuda11 }}

From 12bb26ee60a726c5f4c28d936b8b656dda986022 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Fri, 11 Oct 2024 10:34:18 -0700
Subject: [PATCH 02/44] change

---
 .../github/azure-pipelines/bigmodels-ci-pipeline.yml     | 9 ++++-----
 .../azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml   | 1 +
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
index edeaa28fe4dbf..8477195697ab8 100644
--- a/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
@@ -40,13 +40,12 @@ parameters:
   default: 0
 
 variables:
-  - template: templates/common-variables.yml
   - name: docker_base_image
-    value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
+    value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
   - name: linux_trt_version
-    value: ${{ variables.linux_trt_version_cuda12 }}
+    value: 10.3.0.26-1.cuda11.8
   - name: Repository
-    value: 'onnxruntimecuda12manylinuxbuild'
+    value: 'onnxruntimecuda11manylinuxbuild'
 
 stages:
 - stage: Build_Onnxruntime_Cuda
@@ -123,7 +122,7 @@ stages:
                 --build_shared_lib \
                 --parallel \
                 --build_wheel \
-                --enable_onnx_tests --use_cuda --cuda_version=12.2 --cuda_home=/usr/local/cuda-12 --cudnn_home=/usr/local/cuda-12 \
+                --enable_onnx_tests --use_cuda --cuda_version=11.8 --cuda_home=/usr/local/cuda-11.8 --cudnn_home=/usr/local/cuda-11.8 \
                 --enable_cuda_profiling --enable_cuda_nhwc_ops \
                 --enable_pybind --build_java \
                 --use_cache \
diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
index 6717e8b4faaa2..fee017a9a2149 100644
--- a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
@@ -37,6 +37,7 @@ parameters:
       - 12.2
 
 variables:
+  - template: templates/common-variables.yml
   - name: docker_base_image
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1

From 6732c0002c601ec15935875ec90bef55da87880b Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 13:37:55 -0700
Subject: [PATCH 03/44] update linux trt ci

---
 .../github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
index fee017a9a2149..06d0aa8a91da1 100644
--- a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
@@ -45,9 +45,9 @@ variables:
       value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
   - name: linux_trt_version
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: 10.4.0.26-1.cuda11.8
+      value: linux_trt_version_cuda11
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: 10.4.0.26-1.cuda12.6
+      value: linux_trt_version_cuda12
 
 jobs:
 - job: Linux_Build

From 8f9f8fdb177744ed5a2828e32cdbda5de5d89417 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 13:40:20 -0700
Subject: [PATCH 04/44] update win trt ci

---
 .../github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml  | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
index 7c04d6aa2e739..676916ba2e063 100644
--- a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
@@ -37,11 +37,12 @@ parameters:
     - 12.2
 
 variables:
+  - template: templates/common-variables.yml
   - name: win_trt_folder
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8
+      value: win_trt_folder_cuda11
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6
+      value: win_trt_folder_cuda12
 
 jobs:
 - job: 'build'

From f80d9f75641672475e8cbc7c091a659aadbfe244 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 13:55:29 -0700
Subject: [PATCH 05/44] simplify py-linux-gpu

---
 .../stages/py-cuda-packaging-stage.yml           |  2 --
 .../azure-pipelines/templates/py-linux-gpu.yml   | 16 +++++++++-------
 .../templates/py-packaging-stage.yml             |  1 -
 3 files changed, 9 insertions(+), 10 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
index f9f90b43f0cf6..eb240a9b999f3 100644
--- a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
@@ -79,7 +79,5 @@ stages:
           cuda_version: ${{ parameters.cuda_version }}
           ${{ if eq(parameters.cuda_version, '11.8') }}:
             docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
-            trt_version: 10.4.0.26-1.cuda11.8
           ${{ if eq(parameters.cuda_version, '12.2') }}:
             docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
-            trt_version: 10.4.0.26-1.cuda12.6
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
index d19472bcbab5a..0df30bfd0f971 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
@@ -20,12 +20,6 @@ parameters:
 - name: docker_base_image
   type: string
 
-- name: trt_version
-  type: string
-  default: '10.4.0.26-1.cuda11.8'
-  values:
-    - 10.4.0.26-1.cuda11.8
-    - 10.4.0.26-1.cuda12.6
 - name: cuda_version
   type: string
   default: '11.8'
@@ -33,6 +27,14 @@ parameters:
    - 11.8
    - 12.2
 
+variables:
+  - template: templates/common-variables.yml
+  - name: trt_version
+    ${{ if eq(parameters.cuda_version, '11.8') }}:
+      value: linux_trt_version_cuda11
+    ${{ if eq(parameters.cuda_version, '12.2') }}:
+      value: linux_trt_version_cuda12
+
 stages:
 - stage: Linux_py_GPU_Wheels_${{ parameters.arch }}
   dependsOn: []
@@ -62,7 +64,7 @@ stages:
         parameters:
           Dockerfile: tools/ci_build/github/linux/docker/inference/${{ parameters.arch }}/python/cuda/Dockerfile
           Context: tools/ci_build/github/linux/docker/inference/${{ parameters.arch }}/python/cuda
-          DockerBuildArgs: "--build-arg BASEIMAGE=${{ parameters.docker_base_image }} --build-arg TRT_VERSION=${{ parameters.trt_version }} --build-arg BUILD_UID=$( id -u )"
+          DockerBuildArgs: "--build-arg BASEIMAGE=${{ parameters.docker_base_image }} --build-arg TRT_VERSION=${{ variables.trt_version }} --build-arg BUILD_UID=$( id -u )"
           Repository: onnxruntimecuda${{ replace(parameters.cuda_version, '.', '') }}xtrt86build${{ parameters.arch }}
 
 
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
index edaae227ee78f..598f72af1bf90 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
@@ -446,7 +446,6 @@ stages:
           docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
           extra_build_arg: ${{ parameters.build_py_parameters }}
           cmake_build_type: ${{ parameters.cmake_build_type }}
-          trt_version: '10.4.0.26-1.cuda11.8'
           cuda_version: '11.8'
 
   - ${{ if eq(parameters.enable_windows_arm64_qnn, true) }}:

From f24af9cddc269c9a014debe28eef6a218a5416c5 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 13:57:50 -0700
Subject: [PATCH 06/44] fix

---
 .../github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml | 4 ++--
 .../github/azure-pipelines/templates/py-linux-gpu.yml         | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
index 06d0aa8a91da1..d8c0465ca280e 100644
--- a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
@@ -45,9 +45,9 @@ variables:
       value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
   - name: linux_trt_version
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: linux_trt_version_cuda11
+      value: ${{ variables.linux_trt_version_cuda11 }}
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: linux_trt_version_cuda12
+      value: ${{ variables.linux_trt_version_cuda12 }}
 
 jobs:
 - job: Linux_Build
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
index 0df30bfd0f971..16e3a74df126e 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
@@ -31,9 +31,9 @@ variables:
   - template: templates/common-variables.yml
   - name: trt_version
     ${{ if eq(parameters.cuda_version, '11.8') }}:
-      value: linux_trt_version_cuda11
+      value: ${{ variables.linux_trt_version_cuda11 }}
     ${{ if eq(parameters.cuda_version, '12.2') }}:
-      value: linux_trt_version_cuda12
+      value: ${{ variables.linux_trt_version_cuda12 }}
 
 stages:
 - stage: Linux_py_GPU_Wheels_${{ parameters.arch }}

From 211a98a42c9de52fdda026a7ef744e63bf0ccfd3 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 14:00:46 -0700
Subject: [PATCH 07/44] fix

---
 .../github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml   | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
index 676916ba2e063..f4ab9ee5b4a5c 100644
--- a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
@@ -40,9 +40,9 @@ variables:
   - template: templates/common-variables.yml
   - name: win_trt_folder
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: win_trt_folder_cuda11
+      value: ${{ variables.win_trt_folder_cuda11 }}
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: win_trt_folder_cuda12
+      value: ${{ variables.win_trt_folder_cuda12 }}
 
 jobs:
 - job: 'build'

From d161f8206a63e216d730d272fba2837e76c295ee Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 14:05:21 -0700
Subject: [PATCH 08/44] test

---
 .../azure-pipelines/templates/py-linux-gpu.yml     | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
index 16e3a74df126e..e83a8db5000e9 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
@@ -27,14 +27,6 @@ parameters:
    - 11.8
    - 12.2
 
-variables:
-  - template: templates/common-variables.yml
-  - name: trt_version
-    ${{ if eq(parameters.cuda_version, '11.8') }}:
-      value: ${{ variables.linux_trt_version_cuda11 }}
-    ${{ if eq(parameters.cuda_version, '12.2') }}:
-      value: ${{ variables.linux_trt_version_cuda12 }}
-
 stages:
 - stage: Linux_py_GPU_Wheels_${{ parameters.arch }}
   dependsOn: []
@@ -53,6 +45,12 @@ stages:
           value: -x ${{ parameters.extra_build_arg }}
         ${{ if eq(parameters.extra_build_arg, '') }}:
           value: ''
+      - template: templates/common-variables.yml
+      - name: trt_version
+        ${{ if eq(parameters.cuda_version, '11.8') }}:
+          value: ${{ variables.linux_trt_version_cuda11 }}
+        ${{ if eq(parameters.cuda_version, '12.2') }}:
+          value: ${{ variables.linux_trt_version_cuda12 }}
     steps:
       - checkout: self
         clean: true

From d36e43618c33344e1667eb1e6d3f7228eb6da711 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 14:07:27 -0700
Subject: [PATCH 09/44] fix

---
 .../ci_build/github/azure-pipelines/templates/py-linux-gpu.yml  | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
index e83a8db5000e9..65326f3507a34 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-linux-gpu.yml
@@ -45,7 +45,7 @@ stages:
           value: -x ${{ parameters.extra_build_arg }}
         ${{ if eq(parameters.extra_build_arg, '') }}:
           value: ''
-      - template: templates/common-variables.yml
+      - template: common-variables.yml
       - name: trt_version
         ${{ if eq(parameters.cuda_version, '11.8') }}:
           value: ${{ variables.linux_trt_version_cuda11 }}

From 0230216bd05aa10014a12dd2d7ec083061451897 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 14:40:25 -0700
Subject: [PATCH 10/44] test

---
 .../github/azure-pipelines/templates/jobs/set-winenv.yml        | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index 6a2b7f4566b61..a9a24bef8f6c7 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
@@ -15,6 +15,7 @@ parameters:
     default: '11.8'
 
 steps:
+  - template: ../common-variables.yml
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:
       - powershell: |
           azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v${{ parameters.PrimaryCUDAVersion }}" $(Agent.TempDirectory)
@@ -34,5 +35,6 @@ steps:
     displayName: 'setup env'
     inputs:
       filename: '$(Build.SourcesDirectory)\tools\ci_build\github\windows\${{ parameters.EnvSetupScript }}'
+      arguments: ''
       modifyEnvironment: true
       workingFolder: '$(Build.BinariesDirectory)'

From 32f41fdf7fb9a25054680f23b23c62d0a0eab8fc Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 14:45:02 -0700
Subject: [PATCH 11/44] test set-winenv

---
 .../templates/jobs/set-winenv.yml               | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index a9a24bef8f6c7..3047b3fbccb35 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
@@ -14,8 +14,15 @@ parameters:
     type: string
     default: '11.8'
 
+variables:
+  - template: templates/common-variables.yml
+  - name: win_trt_folder
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: ${{ variables.win_trt_folder_cuda11 }}
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: ${{ variables.win_trt_folder_cuda12 }}
+
 steps:
-  - template: ../common-variables.yml
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:
       - powershell: |
           azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v${{ parameters.PrimaryCUDAVersion }}" $(Agent.TempDirectory)
@@ -25,11 +32,11 @@ steps:
         displayName: 'Download Secondary CUDA SDK v${{ parameters.SecondaryCUDAVersion }}'
   - ${{ if eq(parameters.DownloadTRT, 'true') }}:
       - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" $(Agent.TempDirectory)
-        displayName: 'Download TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8'
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ variables.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ variables.win_trt_folder_cuda11 }}'
       - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6" $(Agent.TempDirectory)
-        displayName: 'Download TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6'
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ variables.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ variables.win_trt_folder_cuda11 }}'
 
   - task: BatchScript@1
     displayName: 'setup env'

From 3fc325eba2d9bbf18cedb83577548a68f455e92e Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 14:57:43 -0700
Subject: [PATCH 12/44] test set-winenv

---
 .../templates/jobs/set-winenv.yml             | 22 +++++++++----------
 1 file changed, 10 insertions(+), 12 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index 3047b3fbccb35..aaa475d3daf0e 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
@@ -13,14 +13,12 @@ parameters:
   - name: SecondaryCUDAVersion
     type: string
     default: '11.8'
-
-variables:
-  - template: templates/common-variables.yml
-  - name: win_trt_folder
-    ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: ${{ variables.win_trt_folder_cuda11 }}
-    ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: ${{ variables.win_trt_folder_cuda12 }}
+  - name: win_trt_folder_cuda11
+    type: string
+    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8'
+  - name: win_trt_folder_cuda12
+    type: string
+    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6'
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:
@@ -32,11 +30,11 @@ steps:
         displayName: 'Download Secondary CUDA SDK v${{ parameters.SecondaryCUDAVersion }}'
   - ${{ if eq(parameters.DownloadTRT, 'true') }}:
       - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ variables.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
-        displayName: 'Download ${{ variables.win_trt_folder_cuda11 }}'
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ parameters.win_trt_folder_cuda11 }}'
       - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ variables.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
-        displayName: 'Download ${{ variables.win_trt_folder_cuda11 }}'
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder_cuda12 }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ variables.win_trt_folder_cuda12 }}'
 
   - task: BatchScript@1
     displayName: 'setup env'

From efb057f6ee67612eadffc692c7311ef059ed392d Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 15:12:13 -0700
Subject: [PATCH 13/44] update c-api-pkg & cuda-pkg

---
 .../azure-pipelines/c-api-noopenmp-packaging-pipelines.yml   | 3 ++-
 .../github/azure-pipelines/cuda-packaging-pipeline.yml       | 5 +++--
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/c-api-noopenmp-packaging-pipelines.yml b/tools/ci_build/github/azure-pipelines/c-api-noopenmp-packaging-pipelines.yml
index e2d977bd60986..4fb53ef2d056f 100644
--- a/tools/ci_build/github/azure-pipelines/c-api-noopenmp-packaging-pipelines.yml
+++ b/tools/ci_build/github/azure-pipelines/c-api-noopenmp-packaging-pipelines.yml
@@ -77,13 +77,14 @@ resources:
     ref: 5eda9aded5462201e6310105728d33016e637ea7
 
 variables:
+- template: templates/common-variables.yml
 - name: ReleaseVersionSuffix
   value: ''
 - name: win_trt_version
   value: 11.8
 
 - name: win_trt_home
-  value: $(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8
+  value: $(Agent.TempDirectory)\${{ variables.win_trt_folder_cuda11 }}
 - name: win_cuda_home
   value: $(Agent.TempDirectory)\v11.8
 
diff --git a/tools/ci_build/github/azure-pipelines/cuda-packaging-pipeline.yml b/tools/ci_build/github/azure-pipelines/cuda-packaging-pipeline.yml
index 7118e85e9ea4b..9b103715e734d 100644
--- a/tools/ci_build/github/azure-pipelines/cuda-packaging-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/cuda-packaging-pipeline.yml
@@ -59,13 +59,14 @@ parameters:
       - 12.2
 
 variables:
+  - template: templates/common-variables.yml
   - name: ReleaseVersionSuffix
     value: ''
   - name: win_trt_home
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: $(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8
+      value: $(Agent.TempDirectory)\${{ variables.win_trt_folder_cuda11 }}
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: $(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6
+      value: $(Agent.TempDirectory)\${{ variables.win_trt_folder_cuda12 }}
   - name: win_cuda_home
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: $(Agent.TempDirectory)\v11.8

From 382cdae7ad75681cc5badd2f814fc7c12fbeb78e Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 15:15:29 -0700
Subject: [PATCH 14/44] cuda12 as common version

---
 .../github/azure-pipelines/templates/common-variables.yml     | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index 007ae16a62be1..3e891524b2c6c 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -2,7 +2,7 @@ variables:
   common_trt_version: '10.4.0.26'
   linux_trt_version_cuda11: ${{ variables.common_trt_version }}-1.cuda11.8
   linux_trt_version_cuda12: ${{ variables.common_trt_version }}-1.cuda12.6
-  common_linux_trt_version: ${{ variables.linux_trt_version_cuda11 }}
+  common_linux_trt_version: ${{ variables.linux_trt_version_cuda12 }}
   win_trt_folder_cuda11: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-11.8
   win_trt_folder_cuda12: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-12.6
-  common_win_trt_folder: ${{ variables.win_trt_folder_cuda11 }}
+  common_win_trt_folder: ${{ variables.win_trt_folder_cuda12 }}

From 57d36048d1c45ff71d48a16acdb187096d8bf156 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Mon, 14 Oct 2024 15:36:30 -0700
Subject: [PATCH 15/44] post merge: set cuda12 as default

---
 .../azure-pipelines/post-merge-jobs.yml       | 21 +++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/post-merge-jobs.yml b/tools/ci_build/github/azure-pipelines/post-merge-jobs.yml
index 833e97b437c33..e981cb3e83ce6 100644
--- a/tools/ci_build/github/azure-pipelines/post-merge-jobs.yml
+++ b/tools/ci_build/github/azure-pipelines/post-merge-jobs.yml
@@ -1,3 +1,20 @@
+parameters:
+- name: CudaVersion
+  displayName: CUDA version
+  type: string
+  default: '12.2'
+  values:
+    - 11.8
+    - 12.2
+
+variables:
+  - template: templates/common-variables.yml
+  - name: win_trt_folder
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: ${{ variables.win_trt_folder_cuda11 }}
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: ${{ variables.win_trt_folder_cuda12 }}
+
 stages:
 - ${{ if or(startsWith(variables['System.CollectionUri'], 'https://dev.azure.com/aiinfra/'),startsWith(variables['System.CollectionUri'], 'https://aiinfra.visualstudio.com/')) }}:
   - template: templates/web-ci.yml
@@ -206,7 +223,7 @@ stages:
         BuildConfig: 'RelWithDebInfo'
         EnvSetupScript: setup_env_cuda.bat
         buildArch: x64
-        additionalBuildFlags: --enable_pybind --build_java --build_nodejs --use_cuda --cuda_home="$(Agent.TempDirectory)\v11.8" --enable_cuda_profiling --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
+        additionalBuildFlags: --enable_pybind --build_java --build_nodejs --use_cuda --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}" --enable_cuda_profiling --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
         msbuildPlatform: x64
         isX86: false
         job_name_suffix: x64_RelWithDebInfo
@@ -226,7 +243,7 @@ stages:
         BuildConfig: 'RelWithDebInfo'
         EnvSetupScript: setup_env_trt.bat
         buildArch: x64
-        additionalBuildFlags: --enable_pybind --build_java --build_nodejs --use_cuda --cuda_home="$(Agent.TempDirectory)\v11.8" --enable_cuda_profiling --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
+        additionalBuildFlags: --enable_pybind --build_java --build_nodejs --use_cuda --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}" --enable_cuda_profiling --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\${{ variables.win_trt_folder }}" --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=86
         msbuildPlatform: x64
         isX86: false
         job_name_suffix: x64_RelWithDebInfo

From 658d893b2a68d8225fe06171dfa35b80397ae1f6 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 10:53:44 -0700
Subject: [PATCH 16/44] test trt10.5

---
 .../github/azure-pipelines/templates/common-variables.yml       | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index 3e891524b2c6c..c2e8fdb4be9ec 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -1,5 +1,5 @@
 variables:
-  common_trt_version: '10.4.0.26'
+  common_trt_version: '10.5.0.18'
   linux_trt_version_cuda11: ${{ variables.common_trt_version }}-1.cuda11.8
   linux_trt_version_cuda12: ${{ variables.common_trt_version }}-1.cuda12.6
   common_linux_trt_version: ${{ variables.linux_trt_version_cuda12 }}

From 1765761309aabd9af19579b099a482a9e4ba5400 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 10:59:45 -0700
Subject: [PATCH 17/44] test trt10.5 on ep_perf

---
 onnxruntime/python/tools/tensorrt/perf/build/build_image.py   | 2 +-
 .../linux-gpu-tensorrt-daily-perf-pipeline.yml                | 2 +-
 .../github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10   | 4 ++--
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/onnxruntime/python/tools/tensorrt/perf/build/build_image.py b/onnxruntime/python/tools/tensorrt/perf/build/build_image.py
index 763d160fa56b5..f207f072018ba 100644
--- a/onnxruntime/python/tools/tensorrt/perf/build/build_image.py
+++ b/onnxruntime/python/tools/tensorrt/perf/build/build_image.py
@@ -18,7 +18,7 @@
     "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",
     "10.4.cuda_11_8_cudnn_8": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10",
-    "10.4.cuda_12_5_cudnn_9": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10",
+    "10.5.cuda_12_5_cudnn_9": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10",
     "BIN": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_tensorrt_bin",
 }
 
diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml
index fb2c86dbf68e3..865307ec95e94 100644
--- a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml
@@ -13,7 +13,7 @@ parameters:
   - 8.6.cuda_11_8_cudnn_8
   - 8.6.cuda_12_3_cudnn_9
   - 10.4.cuda_11_8_cudnn_8
-  - 10.4.cuda_12_5_cudnn_9
+  - 10.5.cuda_12_5_cudnn_9
   - BIN
 
 - name: UseTensorrtOssParser
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10 b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
index 523318f09aba6..e75ede5aeb8f3 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
+++ b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
@@ -5,7 +5,7 @@
 # Dockerfile to run ONNXRuntime with TensorRT integration
 
 # Build base image with required system packages
-FROM nvidia/cuda:12.5.1-cudnn-devel-ubuntu20.04 AS base
+FROM nvidia/cuda:12.6.1-cudnn-devel-ubuntu20.04 AS base
 
 # The local directory into which to build and install CMAKE
 ARG ONNXRUNTIME_LOCAL_CODE_DIR=/code
@@ -31,7 +31,7 @@ RUN pip install --upgrade pip
 RUN pip install setuptools>=68.2.2 psutil
 
 # Install TensorRT
-RUN version="10.4.0.26-1+cuda12.6" &&\
+RUN version="10.5.0.18-1+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 \

From ec627eafeef079178388a785d418eb9fe7eb168a Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 14:16:15 -0700
Subject: [PATCH 18/44] revert ep_perf cuda

---
 .../github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10     | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10 b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
index e75ede5aeb8f3..d53aad676c447 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
+++ b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
@@ -5,7 +5,7 @@
 # Dockerfile to run ONNXRuntime with TensorRT integration
 
 # Build base image with required system packages
-FROM nvidia/cuda:12.6.1-cudnn-devel-ubuntu20.04 AS base
+FROM nvidia/cuda:12.5.1-cudnn-devel-ubuntu20.04 AS base
 
 # The local directory into which to build and install CMAKE
 ARG ONNXRUNTIME_LOCAL_CODE_DIR=/code

From 8193f814d0b3c87d482dcebe738112e1e92e8d0b Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 14:35:27 -0700
Subject: [PATCH 19/44] test nuget-linux-cuda-packaging-stage for nuget
 cuda11/12 pkg ci

---
 .../nuget-linux-cuda-packaging-stage.yml      | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml
index fc6da88917f62..af0fe4b4184a6 100644
--- a/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml
@@ -68,6 +68,7 @@ stages:
     timeoutInMinutes: 180
     pool: 'onnxruntime-Ubuntu2204-AMD-CPU'
     variables:
+    - template: ../templates/common-variables.yml
     - name: CUDA_VERSION_MAJOR
       ${{ if eq(parameters.CudaVersion, '11.8') }}:
         value: '11'
@@ -75,12 +76,11 @@ stages:
         value: '12'
     - name: CUDA_VERSION
       value: ${{ parameters.CudaVersion }}
-
     - name: linux_trt_version
-      ${{ if eq(parameters.CudaVersion, '11.8') }}:
-        value: 10.4.0.26-1.cuda11.8
-      ${{ if eq(parameters.CudaVersion, '12.2') }}:
-        value: 10.4.0.26-1.cuda12.6
+      ${{ if eq(parameters.cuda_version, '11.8') }}:
+        value: ${{ variables.linux_trt_version_cuda11 }}
+      ${{ if eq(parameters.cuda_version, '12.2') }}:
+        value: ${{ variables.linux_trt_version_cuda12 }}
     steps:
     - checkout: self
       clean: true
@@ -140,16 +140,17 @@ stages:
       clean: all
     pool: 'Onnxruntime-Linux-GPU'
     variables:
+    - template: ../templates/common-variables.yml
     - name: CUDA_VERSION_MAJOR
       ${{ if eq(parameters.CudaVersion, '11.8') }}:
         value: '11'
       ${{ if eq(parameters.CudaVersion, '12.2') }}:
         value: '12'
     - name: linux_trt_version
-      ${{ if eq(parameters.CudaVersion, '11.8') }}:
-        value: 10.4.0.26-1.cuda11.8
-      ${{ if eq(parameters.CudaVersion, '12.2') }}:
-        value: 10.4.0.26-1.cuda12.6
+      ${{ if eq(parameters.cuda_version, '11.8') }}:
+        value: ${{ variables.linux_trt_version_cuda11 }}
+      ${{ if eq(parameters.cuda_version, '12.2') }}:
+        value: ${{ variables.linux_trt_version_cuda12 }}
     steps:
     - checkout: self                           # due to checkout multiple repos, the root directory is $(Build.SourcesDirectory)/onnxruntime
       submodules: false

From 20079c813801a08f0bd4df2ad5de1f3db0640164 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 14:42:34 -0700
Subject: [PATCH 20/44] update nuget/templates/test_linux.yml for nuget
 cuda11/12 pkg ci

---
 .../github/azure-pipelines/nuget/templates/test_linux.yml      | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml b/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml
index b1e5816fb748e..d302ecca72d0d 100644
--- a/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml
+++ b/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml
@@ -23,6 +23,7 @@ stages:
     pool: ${{ parameters.AgentPool }}
 
     variables:
+    - template: ../../templates/common-variables.yml
     - name: OnnxRuntimeBuildDirectory
       value: '$(Build.BinariesDirectory)'
 
@@ -61,7 +62,7 @@ stages:
           ${{ if eq(parameters.CudaVersion, '12.2') }}:
             DockerBuildArgs: "
             --build-arg BASEIMAGE=nvidia/cuda:12.2.2-devel-ubuntu20.04
-            --build-arg TRT_VERSION=10.4.0.26-1+cuda12.6
+            --build-arg TRT_VERSION=${{ variables.linux_trt_version_cuda12 }}
             --build-arg BUILD_UID=$( id -u )
             "
           ${{ else }}:

From 8e1903f3cad14c013992004cbea9fb0e1a5d96a4 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 14:50:31 -0700
Subject: [PATCH 21/44] fix

---
 .../stages/nuget-linux-cuda-packaging-stage.yml           | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml
index af0fe4b4184a6..d331c76bc264e 100644
--- a/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/nuget-linux-cuda-packaging-stage.yml
@@ -77,9 +77,9 @@ stages:
     - name: CUDA_VERSION
       value: ${{ parameters.CudaVersion }}
     - name: linux_trt_version
-      ${{ if eq(parameters.cuda_version, '11.8') }}:
+      ${{ if eq(parameters.CudaVersion, '11.8') }}:
         value: ${{ variables.linux_trt_version_cuda11 }}
-      ${{ if eq(parameters.cuda_version, '12.2') }}:
+      ${{ if eq(parameters.CudaVersion, '12.2') }}:
         value: ${{ variables.linux_trt_version_cuda12 }}
     steps:
     - checkout: self
@@ -147,9 +147,9 @@ stages:
       ${{ if eq(parameters.CudaVersion, '12.2') }}:
         value: '12'
     - name: linux_trt_version
-      ${{ if eq(parameters.cuda_version, '11.8') }}:
+      ${{ if eq(parameters.CudaVersion, '11.8') }}:
         value: ${{ variables.linux_trt_version_cuda11 }}
-      ${{ if eq(parameters.cuda_version, '12.2') }}:
+      ${{ if eq(parameters.CudaVersion, '12.2') }}:
         value: ${{ variables.linux_trt_version_cuda12 }}
     steps:
     - checkout: self                           # due to checkout multiple repos, the root directory is $(Build.SourcesDirectory)/onnxruntime

From 20d44f992262cf8424c8511e84d952bc498a8cdb Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 15:05:12 -0700
Subject: [PATCH 22/44] Py-cuda11-test CI: introducing variables to
 py-package-test-pipeline.yml

---
 .../github/azure-pipelines/py-package-test-pipeline.yml      | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml b/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
index fc66cd9f145f7..69e8a176fa806 100644
--- a/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
@@ -5,6 +5,9 @@ resources:
     trigger: true
     branch: main # branch to pick the artifact, Used only for manual triggered pipeline runs for testing the pipeline itself
 
+variables:
+  - template: templates/common-variables.yml
+
 stages:
 - stage: Linux_Test_CPU_x86_64_stage
   jobs:
@@ -55,7 +58,7 @@ stages:
       python_wheel_suffix: '_gpu'
       timeout: 480
       docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
-      trt_version: '10.4.0.26-1.cuda11.8'
+      trt_version: ${{ variables.linux_trt_version_cuda11 }}
       cuda_version: '11.8'
 
 

From 02a53d502a72279d59735c3fd9fa13df8cf8de9f Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 15:06:45 -0700
Subject: [PATCH 23/44] revert trt version

---
 .../github/azure-pipelines/templates/common-variables.yml       | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index c2e8fdb4be9ec..3e891524b2c6c 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -1,5 +1,5 @@
 variables:
-  common_trt_version: '10.5.0.18'
+  common_trt_version: '10.4.0.26'
   linux_trt_version_cuda11: ${{ variables.common_trt_version }}-1.cuda11.8
   linux_trt_version_cuda12: ${{ variables.common_trt_version }}-1.cuda12.6
   common_linux_trt_version: ${{ variables.linux_trt_version_cuda12 }}

From 2b32be73111182b8788c276350653f1621a8d63c Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 15:24:50 -0700
Subject: [PATCH 24/44] Py-Cuda11/12 Test CIs: remove trt_version usage from
 parameters

---
 .../py-cuda-package-test-pipeline.yml           |  1 -
 .../py-package-test-pipeline.yml                |  4 ----
 .../templates/py-packaging-linux-test-cuda.yml  | 17 +++++++++--------
 3 files changed, 9 insertions(+), 13 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/py-cuda-package-test-pipeline.yml b/tools/ci_build/github/azure-pipelines/py-cuda-package-test-pipeline.yml
index 062b88a326dcf..5cbe8ff35ccf5 100644
--- a/tools/ci_build/github/azure-pipelines/py-cuda-package-test-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/py-cuda-package-test-pipeline.yml
@@ -19,6 +19,5 @@ stages:
           python_wheel_suffix: '_gpu'
           timeout: 480
           docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
-          trt_version: '10.4.0.26-1.cuda12.6'
           cuda_version: '12.2'
 
diff --git a/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml b/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
index 69e8a176fa806..a907f798cfd30 100644
--- a/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
@@ -5,9 +5,6 @@ resources:
     trigger: true
     branch: main # branch to pick the artifact, Used only for manual triggered pipeline runs for testing the pipeline itself
 
-variables:
-  - template: templates/common-variables.yml
-
 stages:
 - stage: Linux_Test_CPU_x86_64_stage
   jobs:
@@ -58,7 +55,6 @@ stages:
       python_wheel_suffix: '_gpu'
       timeout: 480
       docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
-      trt_version: ${{ variables.linux_trt_version_cuda11 }}
       cuda_version: '11.8'
 
 
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-linux-test-cuda.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-linux-test-cuda.yml
index 4ca462bf962f5..08826fe7817dd 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-linux-test-cuda.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-linux-test-cuda.yml
@@ -16,12 +16,6 @@ parameters:
 - name: docker_base_image
   type: string
 
-- name: trt_version
-  type: string
-  default: '10.4.0.26-1.cuda11.8'
-  values:
-    - 10.4.0.26-1.cuda11.8
-    - 10.4.0.26-1.cuda12.6
 - name: cuda_version
   type: string
   default: '11.8'
@@ -47,7 +41,14 @@ jobs:
 - job: Linux_Test_GPU${{ parameters.extra_job_id }}_${{ parameters.arch }}
   timeoutInMinutes: ${{ parameters.timeout }}
   variables:
-    skipComponentGovernanceDetection: true
+  - template: common-variables.yml
+  - name: skipComponentGovernanceDetection
+    value: true
+  - name: trt_version
+    ${{ if eq(parameters.cuda_version, '11.8') }}:
+      value: ${{ variables.linux_trt_version_cuda11 }}
+    ${{ if eq(parameters.cuda_version, '12.2') }}:
+      value: ${{ variables.linux_trt_version_cuda12 }}
   workspace:
     clean: all
   pool: ${{ parameters.machine_pool }}
@@ -92,7 +93,7 @@ jobs:
       parameters:
         Dockerfile: tools/ci_build/github/linux/docker/inference/${{ parameters.arch }}/python/cuda/Dockerfile
         Context: tools/ci_build/github/linux/docker/inference/${{ parameters.arch }}/python/cuda
-        DockerBuildArgs: "--build-arg BASEIMAGE=${{ parameters.docker_base_image }} --build-arg TRT_VERSION=${{ parameters.trt_version }} --build-arg BUILD_UID=$( id -u )"
+        DockerBuildArgs: "--build-arg BASEIMAGE=${{ parameters.docker_base_image }} --build-arg TRT_VERSION=${{ variables.trt_version }} --build-arg BUILD_UID=$( id -u )"
         Repository: onnxruntimecuda${{ replace(parameters.cuda_version, '.', '') }}xtrt86build${{ parameters.arch }}
 
   - task: Bash@3

From f152222fd996e2809e40861c4e72c9be94cb9808 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Tue, 15 Oct 2024 17:45:35 -0700
Subject: [PATCH 25/44] py-linux-cuda-package-test-job

---
 .../stages/jobs/py-linux-cuda-package-test-job.yml         | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml b/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml
index 545996a4ffc79..5da38fa6bf76e 100644
--- a/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml
@@ -42,16 +42,17 @@ jobs:
       dependsOn: [ ]
     timeoutInMinutes: ${{ parameters.timeout }}
     variables:
+      - template: ../../templates/common-variables.yml
       - name: docker_base_image
         ${{ if eq(parameters.CudaVersion, '11.8') }}:
-          value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
+          value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
         ${{ if eq(parameters.CudaVersion, '12.2') }}:
           value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241008.1
       - name: linux_trt_version
         ${{ if eq(parameters.CudaVersion, '11.8') }}:
-          value: 10.4.0.26-1.cuda11.8
+          value: ${{ variables.linux_trt_version_cuda11 }}
         ${{ if eq(parameters.CudaVersion, '12.2') }}:
-          value: 10.4.0.26-1.cuda12.6
+          value: ${{ variables.linux_trt_version_cuda12 }}
     pool: ${{ parameters.machine_pool }}
     steps:
       - checkout: self

From 18003c64b664244f2a30bfe43bcdffaa17cc9703 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 10:01:14 -0700
Subject: [PATCH 26/44] test updating py-win-gpu template

---
 .../azure-pipelines/templates/py-win-gpu.yml     | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index e89227d51de32..fc53ae758eddc 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -63,9 +63,19 @@ stages:
       pool:
         name: onnxruntime-Win-CPU-2022
       variables:
-        GRADLE_OPTS: '-Dorg.gradle.daemon=false'
-        VSGenerator: 'Visual Studio 17 2022'
-        CUDA_MODULE_LOADING: 'LAZY'
+      # The build machine pool doesn't have dotnet, so it can't run CG.
+      - template: common-variables.yml
+      - name: GRADLE_OPTS
+        value: '-Dorg.gradle.daemon=false'
+      - name: VSGenerator
+        value: 'Visual Studio 17 2022'
+      - name: CUDA_MODULE_LOADING
+        value: 'LAZY'
+      - name: win_trt_folder
+        ${{ if eq(parameters.CudaVersion, '11.8') }}:
+          value: ${{ variables.win_trt_folder_cuda11 }}
+        ${{ if eq(parameters.CudaVersion, '12.2') }}:
+          value: ${{ variables.win_trt_folder_cuda12 }}
       steps:
           - task: mspremier.PostBuildCleanup.PostBuildCleanup-task.PostBuildCleanup@3
             displayName: 'Clean Agent Directories'

From e47cf6ae35447d1af155221f1e16012850581b96 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 13:57:08 -0700
Subject: [PATCH 27/44] simplify py-cuda-packaging-stage

---
 .../stages/py-cuda-packaging-stage.yml              | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
index eb240a9b999f3..82cdc9735bc38 100644
--- a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
@@ -54,6 +54,14 @@ parameters:
     - '3.11'
     - '3.12'
 
+variables:
+  - template: ../templates/common-variables.yml
+  - name: win_trt_folder
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: ${{ variables.win_trt_folder_cuda11 }}
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: ${{ variables.win_trt_folder_cuda12 }}
+
 stages:
   - ${{ if eq(parameters.enable_windows_gpu, true) }}:
     - ${{ each python_version in parameters.PythonVersions }}:
@@ -64,10 +72,7 @@ stages:
           CudaVersion: ${{ parameters.cuda_version }}
           SpecificArtifact: ${{ parameters.SpecificArtifact }}
           BuildId: ${{ parameters.BuildId }}
-          ${{ if eq(parameters.cuda_version, '11.8') }}:
-            EP_BUILD_FLAGS: --enable_lto --use_tensorrt --tensorrt_home=$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8 --cuda_home=$(Agent.TempDirectory)\v11.8  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
-          ${{ if eq(parameters.cuda_version, '12.2') }}:
-            EP_BUILD_FLAGS: --enable_lto --use_tensorrt --tensorrt_home=$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6 --cuda_home=$(Agent.TempDirectory)\v12.2  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+          EP_BUILD_FLAGS: --enable_lto --use_tensorrt --tensorrt_home=$(Agent.TempDirectory)\${{ variables.win_trt_folder }} --cuda_home=$(Agent.TempDirectory)\v${{ parameters.cuda_version }}  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
 
   - ${{ if eq(parameters.enable_linux_gpu, true) }}:
       - template: ../templates/py-linux-gpu.yml

From d8737d42ce02d525701a47acf02c036515274b10 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 14:13:54 -0700
Subject: [PATCH 28/44] adjust py-cuda-packaging-stage

---
 .../stages/py-cuda-packaging-stage.yml            | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
index 82cdc9735bc38..5ff36b4849c18 100644
--- a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
@@ -54,15 +54,14 @@ parameters:
     - '3.11'
     - '3.12'
 
-variables:
-  - template: ../templates/common-variables.yml
-  - name: win_trt_folder
-    ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: ${{ variables.win_trt_folder_cuda11 }}
-    ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: ${{ variables.win_trt_folder_cuda12 }}
-
 stages:
+  - variables:
+    - template: ../templates/common-variables.yml
+    - name: win_trt_folder
+      ${{ if eq(parameters.cuda_version, '11.8') }}:
+        value: ${{ variables.win_trt_folder_cuda11 }}
+      ${{ if eq(parameters.cuda_version, '12.2') }}:
+        value: ${{ variables.win_trt_folder_cuda12 }}
   - ${{ if eq(parameters.enable_windows_gpu, true) }}:
     - ${{ each python_version in parameters.PythonVersions }}:
       - template: ../templates/py-win-gpu.yml

From f493f8c0a32d92361412be25c2280e234c5e6f99 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 14:53:39 -0700
Subject: [PATCH 29/44] Free trt_build_flag out of EP_BUILD_FLAG; read
 variables by py-win-gpu

---
 .../stages/py-cuda-packaging-stage.yml        | 10 ++-------
 .../templates/py-packaging-stage.yml          |  3 +++
 .../azure-pipelines/templates/py-win-gpu.yml  | 21 ++++++++++++-------
 3 files changed, 19 insertions(+), 15 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
index 5ff36b4849c18..c96c582d19b6a 100644
--- a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
@@ -55,13 +55,6 @@ parameters:
     - '3.12'
 
 stages:
-  - variables:
-    - template: ../templates/common-variables.yml
-    - name: win_trt_folder
-      ${{ if eq(parameters.cuda_version, '11.8') }}:
-        value: ${{ variables.win_trt_folder_cuda11 }}
-      ${{ if eq(parameters.cuda_version, '12.2') }}:
-        value: ${{ variables.win_trt_folder_cuda12 }}
   - ${{ if eq(parameters.enable_windows_gpu, true) }}:
     - ${{ each python_version in parameters.PythonVersions }}:
       - template: ../templates/py-win-gpu.yml
@@ -71,7 +64,8 @@ stages:
           CudaVersion: ${{ parameters.cuda_version }}
           SpecificArtifact: ${{ parameters.SpecificArtifact }}
           BuildId: ${{ parameters.BuildId }}
-          EP_BUILD_FLAGS: --enable_lto --use_tensorrt --tensorrt_home=$(Agent.TempDirectory)\${{ variables.win_trt_folder }} --cuda_home=$(Agent.TempDirectory)\v${{ parameters.cuda_version }}  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+          EP_BUILD_FLAGS: --enable_lto --cuda_home=$(Agent.TempDirectory)\v${{ parameters.cuda_version }}  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+          use_tensorrt: True
 
   - ${{ if eq(parameters.enable_linux_gpu, true) }}:
       - template: ../templates/py-linux-gpu.yml
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
index 598f72af1bf90..4a6dcc02a0a02 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
@@ -287,6 +287,7 @@ stages:
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
         cmake_build_type: ${{ parameters.cmake_build_type }}
+        use_tensorrt: True
 
     - template: py-win-gpu.yml
       parameters:
@@ -297,6 +298,7 @@ stages:
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
         cmake_build_type: ${{ parameters.cmake_build_type }}
+        use_tensorrt: True
 
     - template: py-win-gpu.yml
       parameters:
@@ -307,6 +309,7 @@ stages:
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
         cmake_build_type: ${{ parameters.cmake_build_type }}
+        use_tensorrt: True
 
     - template: py-win-gpu.yml
       parameters:
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index fc53ae758eddc..dcb32c0472ab8 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -52,6 +52,10 @@ parameters:
   type: boolean
   default: false
 
+- name: use_tensorrt
+  type: boolean
+  default: false
+
 stages:
   - stage: Win_py_${{ parameters.EP_NAME }}_Wheels_${{ replace(parameters.PYTHON_VERSION,'.','_') }}_Build
     dependsOn: []
@@ -76,6 +80,11 @@ stages:
           value: ${{ variables.win_trt_folder_cuda11 }}
         ${{ if eq(parameters.CudaVersion, '12.2') }}:
           value: ${{ variables.win_trt_folder_cuda12 }}
+      - name: trt_build_flag
+        ${{ if eq(parameters.use_tensorrt, true) }}:
+          value: '--use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\${{ variables.win_trt_folder }}"'
+        ${{ if eq(parameters.use_tensorrt, false) }}:
+          value: ''
       steps:
           - task: mspremier.PostBuildCleanup.PostBuildCleanup-task.PostBuildCleanup@3
             displayName: 'Clean Agent Directories'
@@ -105,19 +114,17 @@ stages:
             - template: jobs/set-winenv.yml
               parameters:
                 EnvSetupScript: ${{ parameters.ENV_SETUP_SCRIPT }}
-                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), contains(parameters.EP_BUILD_FLAGS, 'use_tensorrt')) }}:
+                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true) }}:
                   DownloadCUDA: true
-                ${{ if contains(parameters.EP_BUILD_FLAGS, 'use_tensorrt') }}:
-                  DownloadTRT: true
+                DownloadTRT: ${{ parameters.use_tensorrt }}
 
           - ${{ if eq(parameters.ENV_SETUP_SCRIPT, '') }}:
             - template: jobs/download_win_gpu_library.yml
               parameters:
                 CudaVersion: ${{ parameters.CudaVersion }}
-                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), contains(parameters.EP_BUILD_FLAGS, 'use_tensorrt')) }}:
+                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true) }}:
                   DownloadCUDA: true
-                ${{ if contains(parameters.EP_BUILD_FLAGS, 'use_tensorrt') }}:
-                  DownloadTRT: true
+                DownloadTRT: ${{ parameters.use_tensorrt }}
 
           - task: PythonScript@0
             displayName: 'Update deps.txt'
@@ -147,7 +154,7 @@ stages:
                 --enable_pybind
                 --enable_onnx_tests
                 --parallel --use_binskim_compliant_compile_flags --update
-                $(TelemetryOption) ${{ parameters.BUILD_PY_PARAMETERS }} ${{ parameters.EP_BUILD_FLAGS }}
+                $(TelemetryOption) ${{ parameters.BUILD_PY_PARAMETERS }} ${{ parameters.EP_BUILD_FLAGS }} ${{ variables.trt_build_flag }}
               workingDirectory: '$(Build.BinariesDirectory)'
 
           # building with build.py so the parallelization parameters are added to the msbuild command

From 1c7dca3cc34b5b4c10f93de01e48180f5e58cb5f Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 14:55:49 -0700
Subject: [PATCH 30/44] fix

---
 .../ci_build/github/azure-pipelines/templates/py-win-gpu.yml  | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index dcb32c0472ab8..2324235cd74a0 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -114,7 +114,7 @@ stages:
             - template: jobs/set-winenv.yml
               parameters:
                 EnvSetupScript: ${{ parameters.ENV_SETUP_SCRIPT }}
-                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true) }}:
+                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true)) }}:
                   DownloadCUDA: true
                 DownloadTRT: ${{ parameters.use_tensorrt }}
 
@@ -122,7 +122,7 @@ stages:
             - template: jobs/download_win_gpu_library.yml
               parameters:
                 CudaVersion: ${{ parameters.CudaVersion }}
-                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true) }}:
+                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true)) }}:
                   DownloadCUDA: true
                 DownloadTRT: ${{ parameters.use_tensorrt }}
 

From 00275e96288012411cdeee47623831d4b6c493ff Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 15:15:25 -0700
Subject: [PATCH 31/44] explicitly assign CudaVersion for py-cuda11 pkg stages

---
 .../azure-pipelines/templates/py-packaging-stage.yml     | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
index 4a6dcc02a0a02..561dc49742774 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
@@ -282,7 +282,8 @@ stages:
       parameters:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.10'
-        EP_BUILD_FLAGS: --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        CudaVersion: 11.8
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -293,7 +294,8 @@ stages:
       parameters:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.11'
-        EP_BUILD_FLAGS: --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        CudaVersion: 11.8
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -304,7 +306,8 @@ stages:
       parameters:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.12'
-        EP_BUILD_FLAGS: --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        CudaVersion: 11.8
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}

From 1d0c40a86ab3327cda418f6d8eacc83cc3c49f40 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 18:42:18 -0700
Subject: [PATCH 32/44] clean unused param of download_win_gpu_library,
 introduce "win_trt_folder"

---
 .../nuget/templates/test_win.yml              | 10 ++++-
 .../stages/java-cuda-packaging-stage.yml      |  8 +++-
 .../stages/nuget-win-cuda-packaging-stage.yml | 12 +++++-
 .../jobs/download_win_gpu_library.yml         | 41 ++++---------------
 .../azure-pipelines/templates/py-win-gpu.yml  |  6 +--
 .../azure-pipelines/templates/win-ci.yml      | 16 ++++----
 6 files changed, 47 insertions(+), 46 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml b/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml
index ddcea447adc94..b29000f1730a0 100644
--- a/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml
+++ b/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml
@@ -26,16 +26,22 @@ stages:
     pool: ${{ parameters.AgentPool }}
 
     variables:
+    - template: ../../templates/common-variables.yml
     - name: OnnxRuntimeBuildDirectory
       value: '$(Build.BinariesDirectory)'
     - name: SKIPNONPACKAGETESTS
       value: 'ON'
     - name: runCodesignValidationInjection
       value: false
-    - name: CUDA_MODULE_LOADINGL
+    - name: CUDA_MODULE_LOADING
       value: 'LAZY'
     - name: GRADLE_OPTS
       value: '-Dorg.gradle.daemon=false'
+    - name: win_trt_folder
+      ${{ if eq(parameters.CudaVersion, '11.8') }}:
+        value: ${{ variables.win_trt_folder_cuda11 }}
+      ${{ if eq(parameters.CudaVersion, '12.2') }}:
+        value: ${{ variables.win_trt_folder_cuda12 }}
 
     steps:
     - task: UsePythonVersion@0
@@ -51,9 +57,9 @@ stages:
     - ${{ if ne( parameters.CudaVersion, '') }}:
       - template: ../../templates/jobs/download_win_gpu_library.yml
         parameters:
-          DownloadCUDA: true
           DownloadTRT: true
           CudaVersion: ${{ parameters.CudaVersion }}
+          win_trt_folder: ${{ variables.win_trt_folder }}
 
     - task: BatchScript@1
       displayName: 'Setup Visual Studio env vars'
diff --git a/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml
index 7bc61268805f2..f4cdc521353e6 100644
--- a/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml
@@ -88,8 +88,14 @@ stages:
     pool: 'onnxruntime-Win2022-GPU-A10'
     timeoutInMinutes: 60
     variables:
+    - template: ../templates/common-variables.yml
     - name: runCodesignValidationInjection
       value: false
+    - name: win_trt_folder
+      ${{ if eq(parameters.CudaVersion, '11.8') }}:
+        value: ${{ variables.win_trt_folder_cuda11 }}
+      ${{ if eq(parameters.CudaVersion, '12.2') }}:
+        value: ${{ variables.win_trt_folder_cuda12 }}
 
     steps:
     - template: ../templates/set-version-number-variables-step.yml
@@ -97,8 +103,8 @@ stages:
     - template: ../templates/jobs/download_win_gpu_library.yml
       parameters:
         CudaVersion: ${{ parameters.CudaVersion }}
-        DownloadCUDA: true
         DownloadTRT: true
+        win_trt_folder: ${{ variables.win_trt_folder }}
 
     - template: ../templates/flex-downloadPipelineArtifact.yml
       parameters:
diff --git a/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml
index 445066f08995a..2704479d125db 100644
--- a/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml
@@ -18,6 +18,9 @@ parameters:
 - name: CudaVersion
   type: string
   default: '11.8'
+  values:
+    - 11.8
+    - 12.2
 
 - name: win_cuda_home
   type: string
@@ -89,8 +92,15 @@ stages:
         clean: all
       pool: 'onnxruntime-Win2022-GPU-A10'
       variables:
+      - template: ../templates/common-variables.yml
+      - name: win_trt_folder
+        ${{ if eq(parameters.CudaVersion, '11.8') }}:
+          value: ${{ variables.win_trt_folder_cuda11 }}
+        ${{ if eq(parameters.CudaVersion, '12.2') }}:
+          value: ${{ variables.win_trt_folder_cuda12 }}
         CUDA_MODULE_LOADINGL: 'LAZY'
         GRADLE_OPTS: '-Dorg.gradle.daemon=false'
+      
       steps:
         - checkout: self                           # due to checkout multiple repos, the root directory is $(Build.SourcesDirectory)/onnxruntime
         - checkout: onnxruntime-inference-examples # due to checkout multiple repos, the root directory is $(Build.SourcesDirectory)/onnxruntime-inference-examples
@@ -102,9 +112,9 @@ stages:
         - script: dir $(Build.SourcesDirectory)
         - template: ../templates/jobs/download_win_gpu_library.yml
           parameters:
-            DownloadCUDA: true
             DownloadTRT: true
             CudaVersion: ${{ parameters.CudaVersion }}
+            win_trt_folder: ${{ variables.win_trt_folder }}
 
         - template: ../templates/set-version-number-variables-step.yml
           parameters:
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
index e196ecb312f96..0db6c1f889b9f 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
@@ -1,7 +1,4 @@
 parameters:
-  - name: DownloadCUDA
-    type: boolean
-    default: false
   - name: DownloadTRT
     type: boolean
     default: false
@@ -11,12 +8,10 @@ parameters:
     values:
       - 11.8
       - 12.2
-  - name: TrtVersion
+  - name: win_trt_folder
     type: string
-    default: '10.4.0.26'
-    values:
-      - 8.6.1.6
-      - 10.4.0.26
+    default: ''
+
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, true) }}:
@@ -32,35 +27,17 @@ steps:
         script: |
           echo %PATH%
       displayName: 'Print PATH after download CUDA SDK'
-
-  - ${{ if eq(parameters.DownloadTRT, true) }}:
-    - ${{ if eq(parameters.CudaVersion, '11.8') }}:
-        - powershell: |
-            Write-Host "##vso[task.setvariable variable=trtCudaVersion;]11.8"
-          displayName: Set trtCudaVersion
-    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '8.6.1.6')) }}:
-        - powershell: |
-            Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.0"
-          displayName: Set trtCudaVersion
-    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '10.4.0.26')) }}:
-        - powershell: |
-            Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.6"
-          displayName: Set trtCudaVersion
-
-    - script: |
-        echo $(trtCudaVersion) && echo TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion)
-      displayName: Get trtCudaVersion and Directory Name
-
+  - ${{ if and(eq(parameters.DownloadTRT, true), ne( parameters.win_trt_folder, '')) }}:
     - powershell: |
-        azcopy.exe cp --recursive https://lotusscus.blob.core.windows.net/models/local/TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion) $(Agent.TempDirectory)
-      displayName: 'Download TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion)'
+        azcopy.exe cp --recursive https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder }} $(Agent.TempDirectory)
+      displayName: 'Download ${{ parameters.win_trt_folder }}'
 
     - powershell: |
-        Write-Host "##vso[task.prependpath]$(Agent.TempDirectory)\TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion)\lib"
-      displayName: 'Append TensorRT-${{ parameters.TrtVersion }} Directory to PATH'
+        Write-Host "##vso[task.prependpath]$(Agent.TempDirectory)\${{ parameters.win_trt_folder }}\lib"
+      displayName: 'Append ${{ parameters.win_trt_folder }} Directory to PATH'
 
     - task: CmdLine@2
       inputs:
         script: |
           echo %PATH%
-      displayName: 'Print PATH after download TensorRT'
+      displayName: 'Print PATH after download TensorRT'
\ No newline at end of file
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index 2324235cd74a0..79e85b87bd0cd 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -122,9 +122,9 @@ stages:
             - template: jobs/download_win_gpu_library.yml
               parameters:
                 CudaVersion: ${{ parameters.CudaVersion }}
-                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true)) }}:
-                  DownloadCUDA: true
-                DownloadTRT: ${{ parameters.use_tensorrt }}
+                ${{ if eq(parameters.use_tensorrt, true) }}:
+                  DownloadTRT: true
+                  win_trt_folder: ${{ variables.win_trt_folder }}
 
           - task: PythonScript@0
             displayName: 'Update deps.txt'
diff --git a/tools/ci_build/github/azure-pipelines/templates/win-ci.yml b/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
index 52547fd9a796b..56d7ba1762868 100644
--- a/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
@@ -85,6 +85,7 @@ stages:
 - stage: Windows_Packaging_${{ parameters.stage_name_suffix }}
   dependsOn: []
   variables:
+  - template: common-variables.yml
     GRADLE_OPTS: '-Dorg.gradle.daemon=false'
     VSGenerator: 'Visual Studio 17 2022'
     CUDA_MODULE_LOADING: 'LAZY'
@@ -96,6 +97,11 @@ stages:
       timeoutParameter: '--test_all_timeout 72000'
     ${{ else }}:
       timeoutParameter: ''
+  - name: win_trt_folder
+    ${{ if eq(parameters.CudaVersion, '11.8') }}:
+      value: ${{ variables.win_trt_folder_cuda11 }}
+    ${{ if eq(parameters.CudaVersion, '12.2') }}:
+      value: ${{ variables.win_trt_folder_cuda12 }}
   jobs:
   - job: Windows_Packaging_${{ parameters.stage_name_suffix }}
     workspace:
@@ -142,14 +148,12 @@ stages:
         - template: jobs/download_win_gpu_library.yml
           parameters:
             CudaVersion: ${{ parameters.CudaVersion }}
-            ${{ if contains(parameters.buildparameter, 'use_cuda') }}:
-              DownloadCUDA: true
             ${{ if contains(parameters.buildparameter, 'use_tensorrt') }}:
-              DownloadCUDA: true
               DownloadTRT: true
+              win_trt_folder: ${{ variables.win_trt_folder }}
       - powershell: |
           Write-Host "##vso[task.prependpath]C:\Program Files (x86)\dotnet"
-        displayName: 'Append dotnet x86  Directory to PATH'
+        displayName: 'Append dotnet x86 Directory to PATH'
         condition: and(succeeded(), eq('${{ parameters.buildArch}}', 'x86'))
 
       - template: download-deps.yml
@@ -373,11 +377,9 @@ stages:
           - template: jobs/download_win_gpu_library.yml
             parameters:
               CudaVersion: ${{ parameters.CudaVersion }}
-              ${{ if contains(parameters.buildparameter, 'use_cuda') }}:
-                DownloadCUDA: true
               ${{ if contains(parameters.buildparameter, 'use_tensorrt') }}:
-                DownloadCUDA: true
                 DownloadTRT: true
+                win_trt_folder: ${{ variables.win_trt_folder }}
 
         - powershell: |
             Write-Host "##vso[task.prependpath]C:\Program Files (x86)\dotnet"

From 20c40a96965e6dfdd2b0e7cdf765dec6027299ad Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 09:11:36 -0700
Subject: [PATCH 33/44] pass win_trt_folder path via parameters

---
 .../templates/jobs/set-winenv.yml             | 21 ++++++-------------
 .../templates/jobs/win-ci-vs-2022-job.yml     |  4 ++++
 .../azure-pipelines/templates/py-win-gpu.yml  |  3 ++-
 .../win-gpu-tensorrt-ci-pipeline.yml          | 14 ++++++-------
 4 files changed, 19 insertions(+), 23 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index aaa475d3daf0e..dcf9c195cdd10 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
@@ -4,21 +4,15 @@ parameters:
   - name: DownloadCUDA
     type: boolean
     default: false
-  - name: DownloadTRT
-    type: boolean
-    default: false
   - name: PrimaryCUDAVersion
     type: string
     default: '12.2'
   - name: SecondaryCUDAVersion
     type: string
     default: '11.8'
-  - name: win_trt_folder_cuda11
+  - name: win_trt_folder
     type: string
-    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8'
-  - name: win_trt_folder_cuda12
-    type: string
-    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6'
+    default: ''
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:
@@ -28,13 +22,10 @@ steps:
       - powershell: |
           azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v${{ parameters.SecondaryCUDAVersion }}" $(Agent.TempDirectory)
         displayName: 'Download Secondary CUDA SDK v${{ parameters.SecondaryCUDAVersion }}'
-  - ${{ if eq(parameters.DownloadTRT, 'true') }}:
-      - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
-        displayName: 'Download ${{ parameters.win_trt_folder_cuda11 }}'
+  - ${{ if ne(parameters.win_trt_folder, '') }}:
       - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder_cuda12 }}" $(Agent.TempDirectory)
-        displayName: 'Download ${{ variables.win_trt_folder_cuda12 }}'
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ parameters.win_trt_folder }}'
 
   - task: BatchScript@1
     displayName: 'setup env'
@@ -42,4 +33,4 @@ steps:
       filename: '$(Build.SourcesDirectory)\tools\ci_build\github\windows\${{ parameters.EnvSetupScript }}'
       arguments: ''
       modifyEnvironment: true
-      workingFolder: '$(Build.BinariesDirectory)'
+      workingFolder: '$(Build.BinariesDirectory)'
\ No newline at end of file
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml
index 34de1201fa049..d9864afde55af 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml
@@ -55,10 +55,14 @@ parameters:
   type: number
   default: 0
 
+- name: Cuda_Version
+  type: string
+  default: ''
 
 jobs:
 - job: build_${{ parameters.job_name_suffix }}
   variables:
+
     OrtPackageId: 'Microsoft.ML.OnnxRuntime'
     OnnxRuntimeBuildDirectory: '$(Build.BinariesDirectory)'
     DOTNET_SKIP_FIRST_TIME_EXPERIENCE: true
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index 79e85b87bd0cd..215a08e988319 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -116,7 +116,8 @@ stages:
                 EnvSetupScript: ${{ parameters.ENV_SETUP_SCRIPT }}
                 ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true)) }}:
                   DownloadCUDA: true
-                DownloadTRT: ${{ parameters.use_tensorrt }}
+                ${{ if eq(parameters.use_tensorrt, true) }}:
+                  win_trt_folder: ${{ variables.win_trt_folder }}
 
           - ${{ if eq(parameters.ENV_SETUP_SCRIPT, '') }}:
             - template: jobs/download_win_gpu_library.yml
diff --git a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
index f4ab9ee5b4a5c..187a92e76a5f5 100644
--- a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
@@ -34,20 +34,18 @@ parameters:
   default: '12.2'
   values:
     - 11.8
-    - 12.2
+    - 12.2  
 
-variables:
+jobs:
+- job: 'build'
+  pool: 'onnxruntime-Win2022-GPU-A10'
+  variables:
   - template: templates/common-variables.yml
   - name: win_trt_folder
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: ${{ variables.win_trt_folder_cuda11 }}
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
       value: ${{ variables.win_trt_folder_cuda12 }}
-
-jobs:
-- job: 'build'
-  pool: 'onnxruntime-Win2022-GPU-A10'
-  variables:
     MsbuildArguments: '-detailedsummary -maxcpucount -consoleloggerparameters:PerformanceSummary'
     EnvSetupScript: setup_env_trt.bat
     skipComponentGovernanceDetection: true
@@ -61,6 +59,8 @@ jobs:
       EnvSetupScript: $(EnvSetupScript)
       DownloadCUDA: true
       DownloadTRT: true
+      win_trt_folder_cuda11: ${{ variables.win_trt_folder_cuda11 }}
+      win_trt_folder_cuda12: ${{ variables.win_trt_folder_cuda12 }}
       BuildArch: 'x64'
       BuildConfig: RelWithDebInfo
       MachinePool: 'onnxruntime-Win2022-GPU-A10'

From e9a393eb67170ea98518bbe0f83780729fc4732a Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 09:15:30 -0700
Subject: [PATCH 34/44] Revert "pass win_trt_folder path via parameters"

This reverts commit 20c40a96965e6dfdd2b0e7cdf765dec6027299ad.
---
 .../templates/jobs/set-winenv.yml             | 21 +++++++++++++------
 .../templates/jobs/win-ci-vs-2022-job.yml     |  4 ----
 .../azure-pipelines/templates/py-win-gpu.yml  |  3 +--
 .../win-gpu-tensorrt-ci-pipeline.yml          | 14 ++++++-------
 4 files changed, 23 insertions(+), 19 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index dcf9c195cdd10..aaa475d3daf0e 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
@@ -4,15 +4,21 @@ parameters:
   - name: DownloadCUDA
     type: boolean
     default: false
+  - name: DownloadTRT
+    type: boolean
+    default: false
   - name: PrimaryCUDAVersion
     type: string
     default: '12.2'
   - name: SecondaryCUDAVersion
     type: string
     default: '11.8'
-  - name: win_trt_folder
+  - name: win_trt_folder_cuda11
     type: string
-    default: ''
+    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8'
+  - name: win_trt_folder_cuda12
+    type: string
+    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6'
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:
@@ -22,10 +28,13 @@ steps:
       - powershell: |
           azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v${{ parameters.SecondaryCUDAVersion }}" $(Agent.TempDirectory)
         displayName: 'Download Secondary CUDA SDK v${{ parameters.SecondaryCUDAVersion }}'
-  - ${{ if ne(parameters.win_trt_folder, '') }}:
+  - ${{ if eq(parameters.DownloadTRT, 'true') }}:
+      - powershell: |
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder_cuda11 }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ parameters.win_trt_folder_cuda11 }}'
       - powershell: |
-          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder }}" $(Agent.TempDirectory)
-        displayName: 'Download ${{ parameters.win_trt_folder }}'
+          azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder_cuda12 }}" $(Agent.TempDirectory)
+        displayName: 'Download ${{ variables.win_trt_folder_cuda12 }}'
 
   - task: BatchScript@1
     displayName: 'setup env'
@@ -33,4 +42,4 @@ steps:
       filename: '$(Build.SourcesDirectory)\tools\ci_build\github\windows\${{ parameters.EnvSetupScript }}'
       arguments: ''
       modifyEnvironment: true
-      workingFolder: '$(Build.BinariesDirectory)'
\ No newline at end of file
+      workingFolder: '$(Build.BinariesDirectory)'
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml
index d9864afde55af..34de1201fa049 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/win-ci-vs-2022-job.yml
@@ -55,14 +55,10 @@ parameters:
   type: number
   default: 0
 
-- name: Cuda_Version
-  type: string
-  default: ''
 
 jobs:
 - job: build_${{ parameters.job_name_suffix }}
   variables:
-
     OrtPackageId: 'Microsoft.ML.OnnxRuntime'
     OnnxRuntimeBuildDirectory: '$(Build.BinariesDirectory)'
     DOTNET_SKIP_FIRST_TIME_EXPERIENCE: true
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index 215a08e988319..79e85b87bd0cd 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -116,8 +116,7 @@ stages:
                 EnvSetupScript: ${{ parameters.ENV_SETUP_SCRIPT }}
                 ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true)) }}:
                   DownloadCUDA: true
-                ${{ if eq(parameters.use_tensorrt, true) }}:
-                  win_trt_folder: ${{ variables.win_trt_folder }}
+                DownloadTRT: ${{ parameters.use_tensorrt }}
 
           - ${{ if eq(parameters.ENV_SETUP_SCRIPT, '') }}:
             - template: jobs/download_win_gpu_library.yml
diff --git a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
index 187a92e76a5f5..f4ab9ee5b4a5c 100644
--- a/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/win-gpu-tensorrt-ci-pipeline.yml
@@ -34,18 +34,20 @@ parameters:
   default: '12.2'
   values:
     - 11.8
-    - 12.2  
+    - 12.2
 
-jobs:
-- job: 'build'
-  pool: 'onnxruntime-Win2022-GPU-A10'
-  variables:
+variables:
   - template: templates/common-variables.yml
   - name: win_trt_folder
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: ${{ variables.win_trt_folder_cuda11 }}
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
       value: ${{ variables.win_trt_folder_cuda12 }}
+
+jobs:
+- job: 'build'
+  pool: 'onnxruntime-Win2022-GPU-A10'
+  variables:
     MsbuildArguments: '-detailedsummary -maxcpucount -consoleloggerparameters:PerformanceSummary'
     EnvSetupScript: setup_env_trt.bat
     skipComponentGovernanceDetection: true
@@ -59,8 +61,6 @@ jobs:
       EnvSetupScript: $(EnvSetupScript)
       DownloadCUDA: true
       DownloadTRT: true
-      win_trt_folder_cuda11: ${{ variables.win_trt_folder_cuda11 }}
-      win_trt_folder_cuda12: ${{ variables.win_trt_folder_cuda12 }}
       BuildArch: 'x64'
       BuildConfig: RelWithDebInfo
       MachinePool: 'onnxruntime-Win2022-GPU-A10'

From fb38fdca63897182a2117b752baa3023fb995b91 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 09:15:35 -0700
Subject: [PATCH 35/44] Revert "clean unused param of
 download_win_gpu_library,"

This reverts commit 1d0c40a86ab3327cda418f6d8eacc83cc3c49f40.
---
 .../nuget/templates/test_win.yml              | 10 +----
 .../stages/java-cuda-packaging-stage.yml      |  8 +---
 .../stages/nuget-win-cuda-packaging-stage.yml | 12 +-----
 .../jobs/download_win_gpu_library.yml         | 41 +++++++++++++++----
 .../azure-pipelines/templates/py-win-gpu.yml  |  6 +--
 .../azure-pipelines/templates/win-ci.yml      | 16 ++++----
 6 files changed, 46 insertions(+), 47 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml b/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml
index b29000f1730a0..ddcea447adc94 100644
--- a/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml
+++ b/tools/ci_build/github/azure-pipelines/nuget/templates/test_win.yml
@@ -26,22 +26,16 @@ stages:
     pool: ${{ parameters.AgentPool }}
 
     variables:
-    - template: ../../templates/common-variables.yml
     - name: OnnxRuntimeBuildDirectory
       value: '$(Build.BinariesDirectory)'
     - name: SKIPNONPACKAGETESTS
       value: 'ON'
     - name: runCodesignValidationInjection
       value: false
-    - name: CUDA_MODULE_LOADING
+    - name: CUDA_MODULE_LOADINGL
       value: 'LAZY'
     - name: GRADLE_OPTS
       value: '-Dorg.gradle.daemon=false'
-    - name: win_trt_folder
-      ${{ if eq(parameters.CudaVersion, '11.8') }}:
-        value: ${{ variables.win_trt_folder_cuda11 }}
-      ${{ if eq(parameters.CudaVersion, '12.2') }}:
-        value: ${{ variables.win_trt_folder_cuda12 }}
 
     steps:
     - task: UsePythonVersion@0
@@ -57,9 +51,9 @@ stages:
     - ${{ if ne( parameters.CudaVersion, '') }}:
       - template: ../../templates/jobs/download_win_gpu_library.yml
         parameters:
+          DownloadCUDA: true
           DownloadTRT: true
           CudaVersion: ${{ parameters.CudaVersion }}
-          win_trt_folder: ${{ variables.win_trt_folder }}
 
     - task: BatchScript@1
       displayName: 'Setup Visual Studio env vars'
diff --git a/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml
index f4cdc521353e6..7bc61268805f2 100644
--- a/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/java-cuda-packaging-stage.yml
@@ -88,14 +88,8 @@ stages:
     pool: 'onnxruntime-Win2022-GPU-A10'
     timeoutInMinutes: 60
     variables:
-    - template: ../templates/common-variables.yml
     - name: runCodesignValidationInjection
       value: false
-    - name: win_trt_folder
-      ${{ if eq(parameters.CudaVersion, '11.8') }}:
-        value: ${{ variables.win_trt_folder_cuda11 }}
-      ${{ if eq(parameters.CudaVersion, '12.2') }}:
-        value: ${{ variables.win_trt_folder_cuda12 }}
 
     steps:
     - template: ../templates/set-version-number-variables-step.yml
@@ -103,8 +97,8 @@ stages:
     - template: ../templates/jobs/download_win_gpu_library.yml
       parameters:
         CudaVersion: ${{ parameters.CudaVersion }}
+        DownloadCUDA: true
         DownloadTRT: true
-        win_trt_folder: ${{ variables.win_trt_folder }}
 
     - template: ../templates/flex-downloadPipelineArtifact.yml
       parameters:
diff --git a/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml
index 2704479d125db..445066f08995a 100644
--- a/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/nuget-win-cuda-packaging-stage.yml
@@ -18,9 +18,6 @@ parameters:
 - name: CudaVersion
   type: string
   default: '11.8'
-  values:
-    - 11.8
-    - 12.2
 
 - name: win_cuda_home
   type: string
@@ -92,15 +89,8 @@ stages:
         clean: all
       pool: 'onnxruntime-Win2022-GPU-A10'
       variables:
-      - template: ../templates/common-variables.yml
-      - name: win_trt_folder
-        ${{ if eq(parameters.CudaVersion, '11.8') }}:
-          value: ${{ variables.win_trt_folder_cuda11 }}
-        ${{ if eq(parameters.CudaVersion, '12.2') }}:
-          value: ${{ variables.win_trt_folder_cuda12 }}
         CUDA_MODULE_LOADINGL: 'LAZY'
         GRADLE_OPTS: '-Dorg.gradle.daemon=false'
-      
       steps:
         - checkout: self                           # due to checkout multiple repos, the root directory is $(Build.SourcesDirectory)/onnxruntime
         - checkout: onnxruntime-inference-examples # due to checkout multiple repos, the root directory is $(Build.SourcesDirectory)/onnxruntime-inference-examples
@@ -112,9 +102,9 @@ stages:
         - script: dir $(Build.SourcesDirectory)
         - template: ../templates/jobs/download_win_gpu_library.yml
           parameters:
+            DownloadCUDA: true
             DownloadTRT: true
             CudaVersion: ${{ parameters.CudaVersion }}
-            win_trt_folder: ${{ variables.win_trt_folder }}
 
         - template: ../templates/set-version-number-variables-step.yml
           parameters:
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
index 0db6c1f889b9f..e196ecb312f96 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
@@ -1,4 +1,7 @@
 parameters:
+  - name: DownloadCUDA
+    type: boolean
+    default: false
   - name: DownloadTRT
     type: boolean
     default: false
@@ -8,10 +11,12 @@ parameters:
     values:
       - 11.8
       - 12.2
-  - name: win_trt_folder
+  - name: TrtVersion
     type: string
-    default: ''
-
+    default: '10.4.0.26'
+    values:
+      - 8.6.1.6
+      - 10.4.0.26
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, true) }}:
@@ -27,17 +32,35 @@ steps:
         script: |
           echo %PATH%
       displayName: 'Print PATH after download CUDA SDK'
-  - ${{ if and(eq(parameters.DownloadTRT, true), ne( parameters.win_trt_folder, '')) }}:
+
+  - ${{ if eq(parameters.DownloadTRT, true) }}:
+    - ${{ if eq(parameters.CudaVersion, '11.8') }}:
+        - powershell: |
+            Write-Host "##vso[task.setvariable variable=trtCudaVersion;]11.8"
+          displayName: Set trtCudaVersion
+    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '8.6.1.6')) }}:
+        - powershell: |
+            Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.0"
+          displayName: Set trtCudaVersion
+    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '10.4.0.26')) }}:
+        - powershell: |
+            Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.6"
+          displayName: Set trtCudaVersion
+
+    - script: |
+        echo $(trtCudaVersion) && echo TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion)
+      displayName: Get trtCudaVersion and Directory Name
+
     - powershell: |
-        azcopy.exe cp --recursive https://lotusscus.blob.core.windows.net/models/local/${{ parameters.win_trt_folder }} $(Agent.TempDirectory)
-      displayName: 'Download ${{ parameters.win_trt_folder }}'
+        azcopy.exe cp --recursive https://lotusscus.blob.core.windows.net/models/local/TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion) $(Agent.TempDirectory)
+      displayName: 'Download TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion)'
 
     - powershell: |
-        Write-Host "##vso[task.prependpath]$(Agent.TempDirectory)\${{ parameters.win_trt_folder }}\lib"
-      displayName: 'Append ${{ parameters.win_trt_folder }} Directory to PATH'
+        Write-Host "##vso[task.prependpath]$(Agent.TempDirectory)\TensorRT-${{ parameters.TrtVersion }}.Windows10.x86_64.cuda-$(trtCudaVersion)\lib"
+      displayName: 'Append TensorRT-${{ parameters.TrtVersion }} Directory to PATH'
 
     - task: CmdLine@2
       inputs:
         script: |
           echo %PATH%
-      displayName: 'Print PATH after download TensorRT'
\ No newline at end of file
+      displayName: 'Print PATH after download TensorRT'
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
index 79e85b87bd0cd..2324235cd74a0 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-win-gpu.yml
@@ -122,9 +122,9 @@ stages:
             - template: jobs/download_win_gpu_library.yml
               parameters:
                 CudaVersion: ${{ parameters.CudaVersion }}
-                ${{ if eq(parameters.use_tensorrt, true) }}:
-                  DownloadTRT: true
-                  win_trt_folder: ${{ variables.win_trt_folder }}
+                ${{ if or(contains(parameters.EP_BUILD_FLAGS, 'use_cuda'), eq(parameters.use_tensorrt, true)) }}:
+                  DownloadCUDA: true
+                DownloadTRT: ${{ parameters.use_tensorrt }}
 
           - task: PythonScript@0
             displayName: 'Update deps.txt'
diff --git a/tools/ci_build/github/azure-pipelines/templates/win-ci.yml b/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
index 56d7ba1762868..52547fd9a796b 100644
--- a/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/win-ci.yml
@@ -85,7 +85,6 @@ stages:
 - stage: Windows_Packaging_${{ parameters.stage_name_suffix }}
   dependsOn: []
   variables:
-  - template: common-variables.yml
     GRADLE_OPTS: '-Dorg.gradle.daemon=false'
     VSGenerator: 'Visual Studio 17 2022'
     CUDA_MODULE_LOADING: 'LAZY'
@@ -97,11 +96,6 @@ stages:
       timeoutParameter: '--test_all_timeout 72000'
     ${{ else }}:
       timeoutParameter: ''
-  - name: win_trt_folder
-    ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: ${{ variables.win_trt_folder_cuda11 }}
-    ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: ${{ variables.win_trt_folder_cuda12 }}
   jobs:
   - job: Windows_Packaging_${{ parameters.stage_name_suffix }}
     workspace:
@@ -148,12 +142,14 @@ stages:
         - template: jobs/download_win_gpu_library.yml
           parameters:
             CudaVersion: ${{ parameters.CudaVersion }}
+            ${{ if contains(parameters.buildparameter, 'use_cuda') }}:
+              DownloadCUDA: true
             ${{ if contains(parameters.buildparameter, 'use_tensorrt') }}:
+              DownloadCUDA: true
               DownloadTRT: true
-              win_trt_folder: ${{ variables.win_trt_folder }}
       - powershell: |
           Write-Host "##vso[task.prependpath]C:\Program Files (x86)\dotnet"
-        displayName: 'Append dotnet x86 Directory to PATH'
+        displayName: 'Append dotnet x86  Directory to PATH'
         condition: and(succeeded(), eq('${{ parameters.buildArch}}', 'x86'))
 
       - template: download-deps.yml
@@ -377,9 +373,11 @@ stages:
           - template: jobs/download_win_gpu_library.yml
             parameters:
               CudaVersion: ${{ parameters.CudaVersion }}
+              ${{ if contains(parameters.buildparameter, 'use_cuda') }}:
+                DownloadCUDA: true
               ${{ if contains(parameters.buildparameter, 'use_tensorrt') }}:
+                DownloadCUDA: true
                 DownloadTRT: true
-                win_trt_folder: ${{ variables.win_trt_folder }}
 
         - powershell: |
             Write-Host "##vso[task.prependpath]C:\Program Files (x86)\dotnet"

From 6ae457a45743c55173240377e69f2f0f00d88f17 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 10:07:17 -0700
Subject: [PATCH 36/44] update py-packaging-selectable-stage (not being used)

---
 .../templates/py-packaging-selectable-stage.yml                | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-selectable-stage.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-selectable-stage.yml
index 8a6434e757a3c..80a848f218980 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-selectable-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-selectable-stage.yml
@@ -379,9 +379,10 @@ stages:
       pool: 'onnxruntime-Win2022-GPU-A10'
       timeoutInMinutes:  300
       variables:
+      - template: common-variables.yml
         CUDA_VERSION: '11.8'
         buildArch: x64
-        EpBuildFlags: --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" --cuda_version=$(CUDA_VERSION) --cuda_home="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v$(CUDA_VERSION)" --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=37;50;52;60;61;70;75;80"
+        EpBuildFlags: --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\${{ variables.win_trt_folder_cuda11 }}" --cuda_version=$(CUDA_VERSION) --cuda_home="C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v$(CUDA_VERSION)" --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=37;50;52;60;61;70;75;80"
         EnvSetupScript: setup_env_gpu.bat
         EP_NAME: gpu
         VSGenerator: 'Visual Studio 17 2022'

From 71e060f035cbb57a1a05190bc22475d4a1744785 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 10:08:35 -0700
Subject: [PATCH 37/44] apply trt10.5

---
 .../github/azure-pipelines/templates/common-variables.yml   | 2 +-
 .../templates/jobs/download_win_gpu_library.yml             | 6 +++---
 .../github/azure-pipelines/templates/jobs/set-winenv.yml    | 4 ++--
 .../linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0  | 2 +-
 .../docker/Dockerfile.package_ubi8_cuda_tensorrt10_0_torch  | 2 +-
 .../github/linux/docker/Dockerfile.package_ubuntu_2004_gpu  | 2 +-
 .../linux/docker/Dockerfile.package_ubuntu_2004_gpu_ffmpeg  | 2 +-
 .../github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10 | 2 +-
 .../linux/docker/inference/x86_64/python/cuda/Dockerfile    | 2 +-
 tools/ci_build/github/windows/setup_env_gpu.bat             | 4 ++--
 tools/ci_build/github/windows/setup_env_trt.bat             | 2 +-
 11 files changed, 15 insertions(+), 15 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index 3e891524b2c6c..c2e8fdb4be9ec 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -1,5 +1,5 @@
 variables:
-  common_trt_version: '10.4.0.26'
+  common_trt_version: '10.5.0.18'
   linux_trt_version_cuda11: ${{ variables.common_trt_version }}-1.cuda11.8
   linux_trt_version_cuda12: ${{ variables.common_trt_version }}-1.cuda12.6
   common_linux_trt_version: ${{ variables.linux_trt_version_cuda12 }}
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
index e196ecb312f96..8b3101a34f760 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/download_win_gpu_library.yml
+++ b/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: '10.4.0.26'
+    default: '10.5.0.18'
     values:
       - 8.6.1.6
-      - 10.4.0.26
+      - 10.5.0.18
 
 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, '10.4.0.26')) }}:
+    - ${{ if and(eq(parameters.CudaVersion, '12.2'), eq(parameters.TrtVersion, '10.5.0.18')) }}:
         - powershell: |
             Write-Host "##vso[task.setvariable variable=trtCudaVersion;]12.6"
           displayName: Set trtCudaVersion
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index aaa475d3daf0e..99d3ccef187da 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/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-10.4.0.26.Windows10.x86_64.cuda-11.8'
+    default: 'TensorRT-10.5.0.18.Windows10.x86_64.cuda-11.8'
   - name: win_trt_folder_cuda12
     type: string
-    default: 'TensorRT-10.4.0.26.Windows10.x86_64.cuda-12.6'
+    default: 'TensorRT-10.5.0.18.Windows10.x86_64.cuda-12.6'
 
 steps:
   - ${{ if eq(parameters.DownloadCUDA, 'true') }}:
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0 b/tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0
index c1a445e29fc89..4ddfa65939906 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0
+++ b/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=10.4.0.26-1.cuda12.6
+ARG TRT_VERSION=10.5.0.18-1.cuda12.6
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION
 ENV PATH /opt/python/cp38-cp38/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/src/tensorrt/bin:${PATH}
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0_torch b/tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0_torch
index a228ebed165eb..44eafa08e5ef5 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.package_ubi8_cuda_tensorrt10_0_torch
+++ b/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=10.4.0.26-1.cuda11.8
+ARG TRT_VERSION=10.5.0.18-1.cuda11.8
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION
 ENV PATH /opt/python/cp38-cp38/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/src/tensorrt/bin:${PATH}
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu b/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu
index 6a4244b7aad0d..f5855c65eb9bd 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu
+++ b/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=10.4.0.26-1+cuda11.8
+ARG TRT_VERSION=10.5.0.18-1+cuda11.8
 ARG LD_LIBRARY_PATH_ARG=/usr/local/lib64:/usr/local/cuda/lib64
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu_ffmpeg b/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu_ffmpeg
index 418c551ab38b4..47562467df893 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu_ffmpeg
+++ b/tools/ci_build/github/linux/docker/Dockerfile.package_ubuntu_2004_gpu_ffmpeg
@@ -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=10.4.0.26-1+cuda11.8
+ARG TRT_VERSION=10.5.0.18-1+cuda11.8
 ARG LD_LIBRARY_PATH_ARG=/usr/local/lib64:/usr/local/cuda/lib64
 FROM $BASEIMAGE AS base
 ARG TRT_VERSION
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10 b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10
index a7d8f220ea9b3..da9cc789f674b 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10
+++ b/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 version="10.4.0.26-1+cuda11.8" &&\
+RUN version="10.5.0.18-1+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 \
diff --git a/tools/ci_build/github/linux/docker/inference/x86_64/python/cuda/Dockerfile b/tools/ci_build/github/linux/docker/inference/x86_64/python/cuda/Dockerfile
index 85b1469a038fd..af7c2ba6d654e 100644
--- a/tools/ci_build/github/linux/docker/inference/x86_64/python/cuda/Dockerfile
+++ b/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=10.4.0.26-1.cuda11.8
+ARG TRT_VERSION=10.5.0.18-1.cuda11.8
 
 #Install TensorRT only if TRT_VERSION is not empty
 RUN if [ -n "${TRT_VERSION}" ]; then  \
diff --git a/tools/ci_build/github/windows/setup_env_gpu.bat b/tools/ci_build/github/windows/setup_env_gpu.bat
index 6a660ecaa40d2..7285e66a149d1 100644
--- a/tools/ci_build/github/windows/setup_env_gpu.bat
+++ b/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-10.4.0.26.Windows10.x86_64.cuda-12.6\lib;%PATH%
+set PATH=%AGENT_TEMPDIRECTORY%\TensorRT-10.5.0.18.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-10.4.0.26.Windows10.x86_64.cuda-11.8\lib
+set PATH=%PATH%;%AGENT_TEMPDIRECTORY%\TensorRT-10.5.0.18.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 (
diff --git a/tools/ci_build/github/windows/setup_env_trt.bat b/tools/ci_build/github/windows/setup_env_trt.bat
index 4f2272e306570..3e4ecf66f457a 100644
--- a/tools/ci_build/github/windows/setup_env_trt.bat
+++ b/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-10.4.0.26.Windows10.x86_64.cuda-12.6\lib;%PATH%
+set PATH=%AGENT_TEMPDIRECTORY%\TensorRT-10.5.0.18.Windows10.x86_64.cuda-12.6\lib;%PATH%
 set GRADLE_OPTS=-Dorg.gradle.daemon=false
 set CUDA_MODULE_LOADING=LAZY

From 2db4db3604321715a34710c3634a8a4dc7a0c8f8 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 10:40:28 -0700
Subject: [PATCH 38/44] fix ep_perf with 10.5

---
 onnxruntime/python/tools/tensorrt/perf/build/build_image.py   | 2 +-
 .../linux-gpu-tensorrt-daily-perf-pipeline.yml                | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/onnxruntime/python/tools/tensorrt/perf/build/build_image.py b/onnxruntime/python/tools/tensorrt/perf/build/build_image.py
index f207f072018ba..3ebc33c02592d 100644
--- a/onnxruntime/python/tools/tensorrt/perf/build/build_image.py
+++ b/onnxruntime/python/tools/tensorrt/perf/build/build_image.py
@@ -17,7 +17,7 @@
 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",
-    "10.4.cuda_11_8_cudnn_8": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10",
+    "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",
     "BIN": "tools/ci_build/github/linux/docker/Dockerfile.ubuntu_tensorrt_bin",
 }
diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml
index 865307ec95e94..83cf26614a285 100644
--- a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-daily-perf-pipeline.yml
@@ -8,11 +8,11 @@ parameters:
 - name: TrtVersion
   displayName: TensorRT Version
   type: string
-  default: 10.4.cuda_12_5_cudnn_9
+  default: 10.5.cuda_12_5_cudnn_9
   values:
   - 8.6.cuda_11_8_cudnn_8
   - 8.6.cuda_12_3_cudnn_9
-  - 10.4.cuda_11_8_cudnn_8
+  - 10.5.cuda_11_8_cudnn_8
   - 10.5.cuda_12_5_cudnn_9
   - BIN
 

From 498853f7d99a4550e37df7db1fcf9ad3c0de4c0f Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Thu, 17 Oct 2024 11:23:26 -0700
Subject: [PATCH 39/44] fix merge conflicts

---
 .../stages/jobs/py-linux-cuda-package-test-job.yml     |  2 +-
 .../azure-pipelines/stages/py-cuda-packaging-stage.yml |  2 +-
 .../azure-pipelines/templates/jobs/set-winenv.yml      |  1 -
 .../azure-pipelines/templates/py-packaging-stage.yml   | 10 ++++++----
 4 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml b/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml
index d8a2a34c84f78..c3e79c4b45613 100644
--- a/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/jobs/py-linux-cuda-package-test-job.yml
@@ -45,7 +45,7 @@ jobs:
       - template: ../../templates/common-variables.yml
       - name: docker_base_image
         ${{ if eq(parameters.CudaVersion, '11.8') }}:
-          value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241008.1
+          value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20241015.1
         ${{ if eq(parameters.CudaVersion, '12.2') }}:
           value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20241015.1
       - name: linux_trt_version
diff --git a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
index a352e54cf602a..a2d9f6a39b990 100644
--- a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
@@ -63,7 +63,7 @@ stages:
           CudaVersion: ${{ parameters.cuda_version }}
           SpecificArtifact: ${{ parameters.SpecificArtifact }}
           BuildId: ${{ parameters.BuildId }}
-          EP_BUILD_FLAGS: --enable_lto --cuda_home=$(Agent.TempDirectory)\v${{ parameters.cuda_version }}  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+          EP_BUILD_FLAGS: --enable_lto --cuda_home=$(Agent.TempDirectory)\v$(CudaVersion) --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
           use_tensorrt: True
 
   - ${{ if eq(parameters.enable_linux_gpu, true) }}:
diff --git a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
index 99d3ccef187da..71b929b995349 100644
--- a/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/jobs/set-winenv.yml
@@ -40,6 +40,5 @@ steps:
     displayName: 'setup env'
     inputs:
       filename: '$(Build.SourcesDirectory)\tools\ci_build\github\windows\${{ parameters.EnvSetupScript }}'
-      arguments: ''
       modifyEnvironment: true
       workingFolder: '$(Build.BinariesDirectory)'
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
index 0e143ecbf9744..c0ede38566a80 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
@@ -287,7 +287,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.10'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -299,7 +299,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.11'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -311,7 +311,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.12'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v${{ parameters.CudaVersion }}"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -322,11 +322,13 @@ stages:
       parameters:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.13'
-        EP_BUILD_FLAGS: --use_tensorrt --tensorrt_home="$(Agent.TempDirectory)\TensorRT-10.4.0.26.Windows10.x86_64.cuda-11.8" --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        CudaVersion: 11.8
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
         cmake_build_type: ${{ parameters.cmake_build_type }}
+        use_tensorrt: True
 
     - template: py-win-gpu.yml
       parameters:

From bb0d023389e28cd49e824e3b758f86312c7be55a Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Fri, 18 Oct 2024 13:27:28 -0700
Subject: [PATCH 40/44] Fix debian version format

---
 .../nuget/templates/test_linux.yml            |  4 +--
 .../templates/common-variables.yml            |  1 +
 .../Dockerfile.ubuntu_cuda11_tensorrt10       | 34 +++++++++----------
 .../Dockerfile.ubuntu_cuda12_tensorrt10       | 34 +++++++++----------
 4 files changed, 37 insertions(+), 36 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml b/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml
index d302ecca72d0d..f9ecfb7cf7938 100644
--- a/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml
+++ b/tools/ci_build/github/azure-pipelines/nuget/templates/test_linux.yml
@@ -53,7 +53,7 @@ stages:
       inputs:
         script: |
           ln -sf /data/models $(Build.BinariesDirectory)
-
+    # As for Debian installation, replace '-1.' by '-1+' when assigning trt version below
     - ${{if contains(parameters.StageSuffix , 'GPU') }}:
       - template: ../../templates/get-docker-image-steps.yml
         parameters:
@@ -62,7 +62,7 @@ stages:
           ${{ if eq(parameters.CudaVersion, '12.2') }}:
             DockerBuildArgs: "
             --build-arg BASEIMAGE=nvidia/cuda:12.2.2-devel-ubuntu20.04
-            --build-arg TRT_VERSION=${{ variables.linux_trt_version_cuda12 }}
+            --build-arg TRT_VERSION=${{ replace(variables.linux_trt_version_cuda12, '-1.', '-1+') }}
             --build-arg BUILD_UID=$( id -u )
             "
           ${{ else }}:
diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index c2e8fdb4be9ec..8113bd8e3bfca 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -1,5 +1,6 @@
 variables:
   common_trt_version: '10.5.0.18'
+  # 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
   common_linux_trt_version: ${{ variables.linux_trt_version_cuda12 }}
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10 b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10
index da9cc789f674b..84dee11536033 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10
+++ b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda11_tensorrt10
@@ -31,26 +31,26 @@ RUN pip install --upgrade pip
 RUN pip install psutil setuptools>=68.2.2
 
 # Install TensorRT
-RUN version="10.5.0.18-1+cuda11.8" &&\
+RUN TRT_VERSION="10.5.0.18-1+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 \
-    libnvinfer-dev=${version} \
-    libnvinfer-dispatch-dev=${version} \
-    libnvinfer-dispatch10=${version} \
-    libnvinfer-headers-dev=${version} \
-    libnvinfer-headers-plugin-dev=${version} \
-    libnvinfer-lean-dev=${version} \
-    libnvinfer-lean10=${version} \
-    libnvinfer-plugin-dev=${version} \
-    libnvinfer-plugin10=${version} \
-    libnvinfer-vc-plugin-dev=${version} \
-    libnvinfer-vc-plugin10=${version} \
-    libnvinfer10=${version} \
-    libnvonnxparsers-dev=${version} \
-    libnvonnxparsers10=${version} \
-    tensorrt-dev=${version} \
-    libnvinfer-bin=${version}
+    libnvinfer-dev=${TRT_VERSION} \
+    libnvinfer-dispatch-dev=${TRT_VERSION} \
+    libnvinfer-dispatch10=${TRT_VERSION} \
+    libnvinfer-headers-dev=${TRT_VERSION} \
+    libnvinfer-headers-plugin-dev=${TRT_VERSION} \
+    libnvinfer-lean-dev=${TRT_VERSION} \
+    libnvinfer-lean10=${TRT_VERSION} \
+    libnvinfer-plugin-dev=${TRT_VERSION} \
+    libnvinfer-plugin10=${TRT_VERSION} \
+    libnvinfer-vc-plugin-dev=${TRT_VERSION} \
+    libnvinfer-vc-plugin10=${TRT_VERSION} \
+    libnvinfer10=${TRT_VERSION} \
+    libnvonnxparsers-dev=${TRT_VERSION} \
+    libnvonnxparsers10=${TRT_VERSION} \
+    tensorrt-dev=${TRT_VERSION} \
+    libnvinfer-bin=${TRT_VERSION}
 
 # Compile trtexec if not installed
 RUN if [ ! -d /usr/src/tensorrt/bin ] || [ ! -f /usr/src/tensorrt/bin/trtexec ]; then \
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10 b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
index d53aad676c447..372e4f1763379 100644
--- a/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
+++ b/tools/ci_build/github/linux/docker/Dockerfile.ubuntu_cuda12_tensorrt10
@@ -31,26 +31,26 @@ RUN pip install --upgrade pip
 RUN pip install setuptools>=68.2.2 psutil
 
 # Install TensorRT
-RUN version="10.5.0.18-1+cuda12.6" &&\
+RUN TRT_VERSION="10.5.0.18-1+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 \
-    libnvinfer-dev=${version} \
-    libnvinfer-dispatch-dev=${version} \
-    libnvinfer-dispatch10=${version} \
-    libnvinfer-headers-dev=${version} \
-    libnvinfer-headers-plugin-dev=${version} \
-    libnvinfer-lean-dev=${version} \
-    libnvinfer-lean10=${version} \
-    libnvinfer-plugin-dev=${version} \
-    libnvinfer-plugin10=${version} \
-    libnvinfer-vc-plugin-dev=${version} \
-    libnvinfer-vc-plugin10=${version} \
-    libnvinfer10=${version} \
-    libnvonnxparsers-dev=${version} \
-    libnvonnxparsers10=${version} \
-    tensorrt-dev=${version} \
-    libnvinfer-bin=${version}
+    libnvinfer-dev=${TRT_VERSION} \
+    libnvinfer-dispatch-dev=${TRT_VERSION} \
+    libnvinfer-dispatch10=${TRT_VERSION} \
+    libnvinfer-headers-dev=${TRT_VERSION} \
+    libnvinfer-headers-plugin-dev=${TRT_VERSION} \
+    libnvinfer-lean-dev=${TRT_VERSION} \
+    libnvinfer-lean10=${TRT_VERSION} \
+    libnvinfer-plugin-dev=${TRT_VERSION} \
+    libnvinfer-plugin10=${TRT_VERSION} \
+    libnvinfer-vc-plugin-dev=${TRT_VERSION} \
+    libnvinfer-vc-plugin10=${TRT_VERSION} \
+    libnvinfer10=${TRT_VERSION} \
+    libnvonnxparsers-dev=${TRT_VERSION} \
+    libnvonnxparsers10=${TRT_VERSION} \
+    tensorrt-dev=${TRT_VERSION} \
+    libnvinfer-bin=${TRT_VERSION}
 
 # Compile trtexec if not installed
 RUN if [ ! -d /usr/src/tensorrt/bin ] || [ ! -f /usr/src/tensorrt/bin/trtexec ]; then \

From 67786ccf8272c96bb9f336f7cde681bfce3f08eb Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Fri, 18 Oct 2024 13:27:46 -0700
Subject: [PATCH 41/44] fix cuda version usage

---
 .../azure-pipelines/stages/py-cuda-packaging-stage.yml    | 2 +-
 .../azure-pipelines/templates/py-packaging-stage.yml      | 8 ++++----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
index a2d9f6a39b990..5e7534bad99f4 100644
--- a/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/stages/py-cuda-packaging-stage.yml
@@ -63,7 +63,7 @@ stages:
           CudaVersion: ${{ parameters.cuda_version }}
           SpecificArtifact: ${{ parameters.SpecificArtifact }}
           BuildId: ${{ parameters.BuildId }}
-          EP_BUILD_FLAGS: --enable_lto --cuda_home=$(Agent.TempDirectory)\v$(CudaVersion) --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+          EP_BUILD_FLAGS: --enable_lto --cuda_home=$(Agent.TempDirectory)\v${{ parameters.cuda_version }} --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
           use_tensorrt: True
 
   - ${{ if eq(parameters.enable_linux_gpu, true) }}:
diff --git a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
index c0ede38566a80..76186cbd50492 100644
--- a/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/py-packaging-stage.yml
@@ -287,7 +287,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.10'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -299,7 +299,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.11'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -311,7 +311,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.12'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}
@@ -323,7 +323,7 @@ stages:
         MACHINE_POOL: 'onnxruntime-Win2022-GPU-A10'
         PYTHON_VERSION: '3.13'
         CudaVersion: 11.8
-        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v$(CudaVersion)"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
+        EP_BUILD_FLAGS: --cuda_home="$(Agent.TempDirectory)\v11.8"  --cmake_extra_defines "CMAKE_CUDA_ARCHITECTURES=52;60;61;70;75;80"
         ENV_SETUP_SCRIPT: setup_env_gpu.bat
         EP_NAME: gpu
         publish_symbols: ${{ parameters.publish_symbols }}

From 8391b24447fcca4c01599b3270255fbf76ac8a21 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Fri, 18 Oct 2024 13:43:46 -0700
Subject: [PATCH 42/44] remove redundant var

---
 .../github/azure-pipelines/templates/common-variables.yml     | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
index 8113bd8e3bfca..b57e45a1069fe 100644
--- a/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
+++ b/tools/ci_build/github/azure-pipelines/templates/common-variables.yml
@@ -3,7 +3,5 @@ variables:
   # 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
-  common_linux_trt_version: ${{ variables.linux_trt_version_cuda12 }}
   win_trt_folder_cuda11: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-11.8
-  win_trt_folder_cuda12: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-12.6
-  common_win_trt_folder: ${{ variables.win_trt_folder_cuda12 }}
+  win_trt_folder_cuda12: TensorRT-${{ variables.common_trt_version }}.Windows10.x86_64.cuda-12.6
\ No newline at end of file

From ada49925b83d268f206ad9371c339b0a01e1310b Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Wed, 16 Oct 2024 17:44:22 -0700
Subject: [PATCH 43/44] unlease nested control flow op pytest

---
 tools/ci_build/build.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/ci_build/build.py b/tools/ci_build/build.py
index 384569997b9b6..9ca6a255966f7 100644
--- a/tools/ci_build/build.py
+++ b/tools/ci_build/build.py
@@ -2123,10 +2123,10 @@ def run_onnxruntime_tests(args, source_dir, ctest_path, build_dir, configs):
             if not args.disable_ml_ops and not args.use_tensorrt:
                 run_subprocess([sys.executable, "onnxruntime_test_python_mlops.py"], cwd=cwd, dll_path=dll_path)
 
-            # if args.use_tensorrt:
-            #     run_subprocess(
-            #         [sys.executable, "onnxruntime_test_python_nested_control_flow_op.py"], cwd=cwd, dll_path=dll_path
-            #     )
+            if args.use_tensorrt:
+                run_subprocess(
+                    [sys.executable, "onnxruntime_test_python_nested_control_flow_op.py"], cwd=cwd, dll_path=dll_path
+                )
 
             try:
                 import onnx  # noqa: F401

From 755b9cd488356204108072dd4c9e5ba86ef2d090 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Fri, 25 Oct 2024 13:58:22 -0700
Subject: [PATCH 44/44] fix merge

---
 .../ci_build/github/azure-pipelines/py-package-test-pipeline.yml | 1 -
 1 file changed, 1 deletion(-)

diff --git a/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml b/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
index b913273c358ae..a0e49692220f9 100644
--- a/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
+++ b/tools/ci_build/github/azure-pipelines/py-package-test-pipeline.yml
@@ -50,7 +50,6 @@ stages:
   - Linux_Test_CPU_aarch64_stage
   - Packages_Somking_Test
   jobs:
-  jobs:
   - job: Final
     # Run this step only if all previous steps are succeeded and (this build was triggered by a resource trigger or it was triggered by another build).
     condition: and(succeeded(), eq(variables['Build.Reason'], 'ResourceTrigger'))