Feature Parity

Migration Progress Burndown
https://metrics.pytorch.org/d/gJaYvuInk/circleci-migration-progress?orgId=1

Feature Parity to match CircleCI

• cancel redundant workflows Add GHA to cancel redundant GHA workflows except on master #54689 @janeyx99
• check and rename all *_py3_* without a python subversion Give Python sub-version in GHA CUDA workflow name #57770 @samestep
• tools/print_test_stats.py Use tools/print_test_stats.py in GHA #57647 @samestep
• docker base image building / pushing (reference PR for CircleCI: Revert "Revert D22360735: .circleci: Build docker images as part of C… #40950) [reland] Build and push Docker images in GitHub Actions #58299 @samestep
• sharding (tracking issue): Add dynamic sharding for Github Actions workflows #59401
• periodic jobs: Move windows periodic jobs to GHA #61003, Move non-libtorch scheduled linux CI to GHA #61732, Migrate linux libtorch to GHA #61774 @janeyx99
• SSH access to FB users @seemethere
• Jump to error (i.e. either parse test result's xml or grep error pattern) from lengthy logs
• GitHub label dispatch management (and consolidate the usage of ci/* and ciflow/* labels
• Ephemeral runners for isolated runner infrastructure (important to lower the infra flakiness)
• Mac/iOS support
• Partnership login and control over GitHub runners (e.g. conducting self-serving releases)

Feature Advantage in GHA

• Squid proxy for network stability inside AWS runners
• Faster and no-cost ECR docker registry data transfer
• CIFlow only works for GHA for now
• AMIs to bake many dependencies into the runners instead of setup for every CI

Jobs may need to stay in CircleCI/Jenkins before new updates

• binary jobs
• nightly jobs
• release jobs
• LTS jobs
• Jenkins AMD/ROCM jobs
• SSH debugging support for non-FB users (due to the access limitation of GHA runners, and currently it's gated by VPN)

Jobs to migrate

Format

- GHA job name
  - CircleCI job name
  - PR <@author> (if any)

• pytorch-ios-12.0.0-arm64
  • pytorch_ios_12_0_0_arm64_build
• pytorch-ios-12.0.0-arm64-custom
  • pytorch_ios_12_0_0_arm64_custom_build
• pytorch-ios-12.0.0-arm64-lite-interpreter
  • pytorch_ios_12_0_0_arm64_lite_interpreter_build
• pytorch-ios-12.0.0-arm64-metal
  • pytorch_ios_12_0_0_arm64_metal_build
• pytorch-ios-12.0.0-x86_64
  • pytorch_ios_12_0_0_x86_64_build
• pytorch-ios-12.0.0-x86_64-lite-interpreter
  • pytorch_ios_12_0_0_x86_64_lite_interpreter_build
• pytorch-bazel
  • docker-pytorch-linux-xenial-py3.6-gcc7 • pytorch_bazel_build • pytorch_bazel_test
  • Add bazel actions workflow #61039 @rsemenov
• pytorch-linux-bionic-py3.6-clang9-noarch
  • docker-pytorch-linux-bionic-py3.6-clang9 • pytorch_linux_bionic_py3_6_clang9_noarch_build • pytorch_linux_bionic_py3_6_clang9_noarch_test
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-linux-bionic-py3.8-gcc9-coverage
  • docker-pytorch-linux-bionic-py3.8-gcc9 • pytorch_linux_bionic_py3_8_gcc9_coverage_build • pytorch_linux_bionic_py3_8_gcc9_coverage_test1 • pytorch_linux_bionic_py3_8_gcc9_coverage_test2
  • Migrate pytorch-linux-bionic-py3.8-gcc9-coverage to GHA #61050 @imaginary-person
• pytorch-linux-bionic-rocm3.9-py3.6
  • docker-pytorch-linux-bionic-rocm3.9-py3.6 • pytorch_linux_bionic_rocm3_9_py3_6_build
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-linux-backward-compatibility-check
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_linux_backward_compatibility_check_test
• pytorch-linux-xenial-py3.6-clang5-android-ndk-r19c-x86_32
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_android_ndk_r19c_x86_32_build
• pytorch-linux-xenial-py3.6-clang5-android-ndk-r19c-x86_64
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_android_ndk_r19c_x86_64_build
• pytorch-linux-xenial-py3.6-clang5-android-ndk-r19c-arm-v7a
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_android_ndk_r19c_arm_v7a_build
• pytorch-linux-xenial-py3.6-clang5-android-ndk-r19c-arm-v8a
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_android_ndk_r19c_arm_v8a_build
• pytorch-linux-xenial-py3.6-clang5-asan
  • docker-pytorch-linux-xenial-py3-clang5-asan • pytorch_linux_xenial_py3_clang5_asan_build • pytorch_linux_xenial_py3_clang5_asan_test1 • pytorch_linux_xenial_py3_clang5_asan_test2
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-linux-xenial-py3.6-clang5-mobile
  • docker-pytorch-linux-xenial-py3-clang5-asan • pytorch_linux_xenial_py3_clang5_mobile_build
• pytorch-linux-xenial-py3.6-clang5-mobile-code-analysis
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_mobile_code_analysis
• pytorch-linux-xenial-py3.6-clang5-mobile-custom-dynamic
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_mobile_custom_build_dynamic
• pytorch-linux-xenial-py3.6-clang5-mobile-custom-static
  • docker-pytorch-linux-xenial-py3-clang5-android-ndk-r19c • pytorch_linux_xenial_py3_clang5_mobile_custom_build_static
• pytorch-linux-xenial-py3.6-clang7-onnx
  • docker-pytorch-linux-xenial-py3-clang7-onnx • pytorch_linux_xenial_py3_clang7_onnx_build • pytorch_linux_xenial_py3_clang7_onnx_ort_test1 • pytorch_linux_xenial_py3_clang7_onnx_ort_test2
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-cpp-doc
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_cpp_doc_build
• pytorch-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_python_doc_build
  • Migrate pytorch_python_doc_build to github action #57371 @zhouzhuojie
• pytorch-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_python_doc_build • pytorch_python_doc_push
• pytorch-doc-test
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_doc_test
• pytorch-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_linux_xenial_py3_6_gcc5_4_test
  • [reland] .github: Add initial linux CI workflow #56280 @seemethere
• pytorch-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_linux_xenial_py3_6_gcc5_4_build • pytorch_linux_xenial_py3_6_gcc5_4_jit_legacy_test
  • ci: Disable jit legacy on circleci, enable on gha #65106 @seemethere
• pytorch-linux-xenial-py3.6-gcc7
  • docker-pytorch-linux-xenial-py3.6-gcc7 • pytorch_linux_xenial_py3_6_gcc7_build • pytorch_linux_xenial_py3_6_gcc7_test
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-parallelnative-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_parallelnative_linux_xenial_py3_6_gcc5_4_build • pytorch_parallelnative_linux_xenial_py3_6_gcc5_4_test
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-paralleltbb-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_paralleltbb_linux_xenial_py3_6_gcc5_4_build • pytorch_paralleltbb_linux_xenial_py3_6_gcc5_4_test
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-pure_torch-linux-xenial-py3.6-gcc5.4
  • docker-pytorch-linux-xenial-py3.6-gcc5.4 • pytorch_pure_torch_linux_xenial_py3_6_gcc5_4_build
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-vulkan-linux-bionic-py3.6-clang9
  • docker-pytorch-linux-bionic-py3.6-clang9 • pytorch_vulkan_linux_bionic_py3_6_clang9_build • pytorch_vulkan_linux_bionic_py3_6_clang9_test
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-xla-linux-bionic-py3.6-clang9
  • docker-pytorch-linux-bionic-py3.6-clang9 • pytorch_xla_linux_bionic_py3_6_clang9_build • pytorch_xla_linux_bionic_py3_6_clang9_test
  • [skip ci] Add more Linux workflows #60889 @driazati
• pytorch-linux-bionic-cuda10.2-cudnn7-py3.9-gcc7
  • docker-pytorch-linux-bionic-cuda10.2-cudnn7-py3.9-gcc7 • pytorch_linux_bionic_cuda10_2_cudnn7_py3_9_gcc7_build • pytorch_linux_bionic_cuda10_2_cudnn7_py3_9_gcc7_test1 • pytorch_linux_bionic_cuda10_2_cudnn7_py3_9_gcc7_test2
  • Enable bionic-cuda10.2-cudnn7-py3.9-gcc7 in GHA #60204 @samestep
• pytorch-libtorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_libtorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build
  • Migrate linux libtorch to GHA #61774 @janeyx99
• pytorch-libtorch-linux-xenial-cuda11.1-cudnn8-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda11.1-cudnn8-py3-gcc7 • pytorch_libtorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_build
  • Migrate linux libtorch to GHA #61774 @janeyx99
• pytorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_test1 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_test2
  • .github: Add initial Linux CI for CUDA #56494 @seemethere
• pytorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build • pytorch_linux_xenial_cuda10_2_cudnn7_py3_jit_legacy_test
  • Add other Linux GPU auxiliary test jobs #61055 @samestep
• pytorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build • pytorch_linux_xenial_cuda10_2_cudnn7_py3_multigpu_test
  • Enable multigpu_test in GHA #60221 @samestep
• pytorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build • pytorch_linux_xenial_cuda10_2_cudnn7_py3_nogpu_NO_AVX_test
  • Add other Linux GPU auxiliary test jobs #61055 @samestep
• pytorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build • pytorch_linux_xenial_cuda10_2_cudnn7_py3_nogpu_NO_AVX2_test
  • Add other Linux GPU auxiliary test jobs #61055 @samestep
• pytorch-linux-xenial-cuda10.2-cudnn7-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7 • pytorch_linux_xenial_cuda10_2_cudnn7_py3_gcc7_build • pytorch_linux_xenial_cuda10_2_cudnn7_py3_slow_test
  • Add other Linux GPU auxiliary test jobs #61055 @samestep
• pytorch-linux-xenial-cuda11.1-cudnn8-py3.6-gcc7
  • docker-pytorch-linux-xenial-cuda11.1-cudnn8-py3-gcc7 • pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_build • pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_test1 • pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_test2
  • Enable xenial-cuda11.1-cudnn8-py3.6-gcc7 in GHA #60196 @samestep
• pytorch-macos-10.13-py3.7
  • pytorch_macos_10_13_py3_build • pytorch_macos_10_13_py3_test • pytorch_macos_10_13_py3_lite_interpreter_build_test
• pytorch-macos-10.15-py3.8
  • pytorch_macos_10_15_py3_build
  • [do not merge] MacOS GHA #57474 @driazati
• pytorch-windows-vs2019-py36-cpu
  • pytorch_windows_vs2019_py36_cpu_build • pytorch_windows_vs2019_py36_cpu_test1 • pytorch_windows_vs2019_py36_cpu_test2
• pytorch-windows-vs2019-py36-cuda10.1
  • pytorch_windows_vs2019_py36_cuda10.1_build • pytorch_windows_vs2019_py36_cuda10.1_test1 • pytorch_windows_vs2019_py36_cuda10.1_test2
• pytorch-windows-vs2019-py36-cuda10.1-on-cpu
  • pytorch_windows_vs2019_py36_cuda10.1_build • pytorch_windows_vs2019_py36_cuda10.1_on_cpu_test1
• pytorch-windows-vs2019-py36-cuda11.1
  • pytorch_windows_vs2019_py36_cuda11.1_build • pytorch_windows_vs2019_py36_cuda11.1_test1 • pytorch_windows_vs2019_py36_cuda11.1_test2

cc @ezyang @seemethere @malfet @walterddr @pytorch/pytorch-dev-infra