Merge branch 'master' into numpy

.azure/gpu-example-tests.yml

@@ -8,12 +8,10 @@ pr:
   drafts: 'true'
 
 jobs:
-- template: testing-template.yml
+- template: template-examples.yml
   parameters:
     domains:
     - "image"
-    - "icevision"
-    - "vissl"
     - "text"
     - "tabular"
     - "video"

.azure/gpu-special-tests.yml

@@ -19,69 +19,59 @@ jobs:
     timeoutInMinutes: "45"
     # how much time to give 'run always even if cancelled tasks' before stopping them
     cancelTimeoutInMinutes: "2"
-
-    pool: lit-rtx-3090
+    pool: "lit-rtx-3090"
     variables:
       DEVICES: $( python -c 'print("$(Agent.Name)".split("_")[-1])' )
-
     container:
-      image: "pytorchlightning/pytorch_lightning:base-cuda-py3.9-torch1.12-cuda11.6.1"
-      options: "--ipc=host --gpus=all"
-
+      # image: "pytorchlightning/pytorch_lightning:base-cuda-py3.9-torch1.11"
+      image: "pytorch/pytorch:1.11.0-cuda11.3-cudnn8-runtime"
+      options: "--ipc=host --gpus=all -v /usr/bin/docker:/tmp/docker:ro"
     workspace:
       clean: all
-
     steps:
 
-    - bash: echo "##vso[task.setvariable variable=CUDA_VISIBLE_DEVICES]$(DEVICES)"
-      displayName: 'set visible devices'
-
-    - bash: |
-        echo $CUDA_VISIBLE_DEVICES
-        lspci | egrep 'VGA|3D'
-        whereis nvidia
-        nvidia-smi
-        python --version
-        pip --version
-        pip list
-        df -kh /dev/shm
-      displayName: 'Image info & NVIDIA'
+      - bash: |
+          echo "##vso[task.setvariable variable=CUDA_VISIBLE_DEVICES]$(DEVICES)"
+          echo "##vso[task.setvariable variable=CONTAINER_ID]$(head -1 /proc/self/cgroup|cut -d/ -f3)"
+        displayName: 'Set environment variables'
 
-    - bash: |
-        python -c "import torch ; mgpu = torch.cuda.device_count() ; assert mgpu >= 2, f'GPU: {mgpu}'"
-      displayName: 'Sanity check'
+      - script: |
+          /tmp/docker exec -t -u 0 $CONTAINER_ID \
+          sh -c "apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -o Dpkg::Options::="--force-confold" -y install sudo"
+        displayName: 'Install Sudo in container (thanks Microsoft!)'
 
-    - bash: |
-        # python -m pip install "pip==20.1"
-        pip install '.[image]' learn2learn
-        pip install '.[test]' --upgrade-strategy only-if-needed
-        pip list
-      displayName: 'Install dependencies'
+      - bash: |
+          echo $CUDA_VISIBLE_DEVICES
+          lspci | egrep 'VGA|3D'
+          whereis nvidia
+          nvidia-smi
+          python --version
+          pip --version
+          pip list
+          df -kh /dev/shm
+        displayName: 'Image info & NVIDIA'
 
-    - bash: |
-        bash tests/special_tests.sh
-      displayName: 'Testing: special'
+      - bash: |
+          python -c "import torch ; mgpu = torch.cuda.device_count() ; assert mgpu >= 2, f'GPU: {mgpu}'"
+        displayName: 'Sanity check'
 
-    - bash: |
-        python -m coverage report
-        python -m coverage xml
-        python -m coverage html
-        python -m codecov --token=$(CODECOV_TOKEN) --commit=$(Build.SourceVersion) --flags=gpu,pytest --name="GPU-coverage" --env=linux,azure
-        ls -l
-      displayName: 'Statistics'
+      - script: |
+          sudo apt-get install -y build-essential gcc cmake software-properties-common
+          python -m pip install "pip==22.2.1"
+          pip --version
+          pip install '.[image,test]' -r requirements/testing_image.txt -U
+          pip list
+        env:
+          FREEZE_REQUIREMENTS: 1
+        displayName: 'Install dependencies'
 
-    - task: PublishTestResults@2
-      displayName: 'Publish test results'
-      inputs:
-        testResultsFiles: '$(Build.StagingDirectory)/test-results.xml'
-        testRunTitle: '$(Agent.OS) - $(Build.DefinitionName) - Python $(python.version)'
-      condition: succeededOrFailed()
+      - bash: |
+          bash tests/special_tests.sh
+        displayName: 'Testing: special'
 
-    - task: PublishCodeCoverageResults@1
-      displayName: 'Publish coverage report'
-      inputs:
-        codeCoverageTool: 'cobertura'
-        summaryFileLocation: 'coverage.xml'
-        reportDirectory: '$(Build.SourcesDirectory)/htmlcov'
-        testRunTitle: '$(Agent.OS) - $(Build.BuildNumber)[$(Agent.JobName)] - Python $(python.version)'
-      condition: succeededOrFailed()
+      - bash: |
+          python -m coverage report
+          python -m coverage xml
+          # python -m codecov --token=$(CODECOV_TOKEN) --commit=$(Build.SourceVersion) --flags=gpu,pytest --name="GPU-coverage" --env=linux,azure
+          ls -l
+        displayName: 'Statistics'

.azure/template-examples.yml

@@ -0,0 +1,68 @@
+jobs:
+  - ${{ each topic in parameters.domains }}:
+    - job:
+      displayName: "domain ${{topic}} with 2 GPU"
+      # how long to run the job before automatically cancelling
+      timeoutInMinutes: "45"
+      # how much time to give 'run always even if cancelled tasks' before stopping them
+      cancelTimeoutInMinutes: "2"
+
+      pool: "lit-rtx-3090"
+      variables:
+        DEVICES: $( python -c 'print("$(Agent.Name)".split("_")[-1])' )
+
+      # this need to have installed docker in the base image...
+      container:
+        # base ML image: mcr.microsoft.com/azureml/openmpi3.1.2-cuda10.2-cudnn8-ubuntu18.04
+        # image: "pytorchlightning/pytorch_lightning:base-cuda-py3.9-torch1.11"
+        image: "pytorch/pytorch:1.11.0-cuda11.3-cudnn8-runtime"
+        options: "-it --rm --gpus=all --shm-size=16g -v /usr/bin/docker:/tmp/docker:ro"
+
+      workspace:
+        clean: all
+      steps:
+
+      - bash: |
+          echo "##vso[task.setvariable variable=CUDA_VISIBLE_DEVICES]$(DEVICES)"
+          echo "##vso[task.setvariable variable=CONTAINER_ID]$(head -1 /proc/self/cgroup|cut -d/ -f3)"
+        displayName: 'Set environment variables'
+
+      - script: |
+          /tmp/docker exec -t -u 0 $CONTAINER_ID \
+          sh -c "apt-get update && DEBIAN_FRONTEND=noninteractive apt-get -o Dpkg::Options::="--force-confold" -y install sudo"
+        displayName: 'Install Sudo in container (thanks Microsoft!)'
+
+      - bash: |
+          echo $CUDA_VISIBLE_DEVICES
+          lspci | egrep 'VGA|3D'
+          whereis nvidia
+          nvidia-smi
+          pip --version
+          pip list
+          df -kh /dev/shm
+        displayName: 'Image info & NVIDIA'
+
+      - script: |
+          python -c "import torch ; mgpu = torch.cuda.device_count() ; assert mgpu >= 2, f'GPU: {mgpu}'"
+        displayName: 'Sanity check'
+
+      - script: |
+          sudo apt-get install -y build-essential gcc cmake software-properties-common
+          python -m pip install "pip==22.2.1"
+          pip --version
+          pip install '.[${{topic}},test]' -r "requirements/testing_${{topic}}.txt" -U --prefer-binary
+        env:
+          FREEZE_REQUIREMENTS: 1
+        displayName: 'Install dependencies'
+
+      - script: |
+          pip list
+          python -m coverage run --source flash -m pytest tests/examples -vV --durations=30
+        displayName: 'Testing'
+
+      - bash: |
+          python -m coverage report
+          python -m coverage xml
+          # python -m codecov --token=$(CODECOV_TOKEN) --commit=$(Build.SourceVersion) --flags=gpu,pytest --name="GPU-coverage" --env=linux,azure
+          ls -l
+        displayName: 'Statistics'