Updating cudnn from 8 to 9 on exsiting cuda 12 docker image (microsof…

…t#20925) ### Description Adding support of cudnn 9 ### Motivation and Context Keep exsiting cuda 12.2 with nvidia dirver 535
titaiwangms · Jun 11, 2024 · 05032e5 · 05032e5
1 parent 043ef5c
commit 05032e5
Show file tree

Hide file tree

Showing 10 changed files with 12 additions and 9 deletions.
diff --git a/setup.py b/setup.py
@@ -198,6 +198,7 @@ def run(self):
                     "libcudart.so.11.0",
                     "libcudart.so.12",
                     "libcudnn.so.8",
+                    "libcudnn.so.9",
                     "libcufft.so.10",
                     "libcufft.so.11",
                     "libcurand.so.10",

diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-ci-pipeline.yml
@@ -50,7 +50,7 @@ variables:
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240531.1
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240531.1
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20240610.1
 
   - name: Repository
     ${{ if eq(parameters.CudaVersion, '11.8') }}:

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
@@ -40,7 +40,7 @@ variables:
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240531.1
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240531.1
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20240610.1
   - name: linux_trt_version
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: 10.0.1.6-1.cuda11.8

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
@@ -60,7 +60,7 @@ stages:
           Context: tools/ci_build/github/linux/docker/
           ${{ if eq(parameters.CudaVersion, '12.2') }}:
             DockerBuildArgs: "
-            --build-arg BASEIMAGE=nvidia/cuda:12.2.2-cudnn8-devel-ubuntu20.04
+            --build-arg BASEIMAGE=nvidia/cuda:12.2.2-devel-ubuntu20.04
             --build-arg TRT_VERSION=10.0.1.6-1+cuda12.4
             --build-arg BUILD_UID=$( id -u )
             "

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
@@ -141,7 +141,7 @@ stages:
       ${{ if eq(parameters.CudaVersion, '11.8') }}:
         value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240531.1
       ${{ if eq(parameters.CudaVersion, '12.2') }}:
-        value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240531.1
+        value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20240610.1
     timeoutInMinutes: 60
 
     steps:

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
@@ -46,7 +46,7 @@ jobs:
         ${{ if eq(parameters.CudaVersion, '11.8') }}:
           value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
         ${{ if eq(parameters.CudaVersion, '12.2') }}:
-          value: nvidia/cuda:12.2.2-cudnn8-devel-ubi8
+          value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20240610.1
       - name: linux_trt_version
         ${{ if eq(parameters.CudaVersion, '11.8') }}:
           value: 10.0.1.6-1.cuda11.8

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
@@ -81,5 +81,5 @@ stages:
             docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240531.1
             trt_version: 10.0.1.6-1.cuda11.8
           ${{ if eq(parameters.cuda_version, '12.2') }}:
-            docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240531.1
+            docker_base_image: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20240610.1
             trt_version: 10.0.1.6-1.cuda12.4
diff --git a/tools/ci_build/github/linux/docker/Dockerfile.manylinux2_28_training_cuda12_2 b/tools/ci_build/github/linux/docker/Dockerfile.manylinux2_28_training_cuda12_2
@@ -1,4 +1,4 @@
-FROM onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240531.1
+FROM onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12:20240610.1
 
 ARG PYTHON_VERSION=3.9
 ARG TORCH_VERSION=2.1.0

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
@@ -46,7 +46,9 @@ RUN apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/
     libnvonnxparsers-dev=${TRT_VERSION} \
     libnvonnxparsers10=${TRT_VERSION} \
     tensorrt-dev=${TRT_VERSION} \
-    libnvinfer-bin=${TRT_VERSION}
+    libnvinfer-bin=${TRT_VERSION} &&\
+    if [ $(echo $CUDA_VERSION | cut -d"." -f1) -ge 12 ]; then apt-get install -y cudnn9-cuda-12 ; fi
+#   ^^^^^^^^^^^If cuda version is 12 or higher, install cudnn 9 for cuda 12
 
 ADD scripts /tmp/scripts
 RUN cd /tmp/scripts && /tmp/scripts/install_dotnet.sh && rm -rf /tmp/scripts

diff --git a/tools/ci_build/github/linux/docker/inference/x86_64/default/cuda12/Dockerfile b/tools/ci_build/github/linux/docker/inference/x86_64/default/cuda12/Dockerfile
@@ -2,7 +2,7 @@
 # Licensed under the MIT License.
 
 # This file is used by Zip-Nuget Packaging NoContribOps Pipeline,Zip-Nuget-Java Packaging Pipeline
-FROM onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12_dotnet:20240531.1
+FROM onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_x64_ubi8_gcc12_dotnet:20240610.1
 ARG TRT_VERSION
 
 #Install TensorRT only if TRT_VERSION is not empty