[inductor] Use multiple outputs for flex-attention

[peterbell10]

Stack from ghstack (oldest at bottom):

• [inductor] Refactor fusion of inplace operations #128979
• [inductor] Kill mark_node_as_mutating #129346
• -> [inductor] Use multiple outputs for flex-attention #129344
• [inductor] Make UserDefinedTritonKernel a multi-output operation #129325
• [inductor] Separate Buffer and Operation into two concepts #128893

This fixes the DCE issue for attention output

cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/129344

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

❌ 44 Cancelled Jobs, 1 Unrelated Failure

As of commit a0488af with merge base c549629 ():

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• inductor / cuda12.1-py3.10-gcc9-sm86 / build (gh)
  ##[error]The operation was canceled.
• inductor / cuda12.1-py3.12-gcc9-sm86 / build (gh)
  ##[error]The operation was canceled.
• inductor / cuda12.4-py3.10-gcc9-sm86 / build (gh)
  ##[error]The operation was canceled.
• inductor / linux-jammy-cpu-py3.12-gcc11-inductor-halide / build (gh)
  ##[error]The operation was canceled.
• inductor / linux-jammy-cpu-py3.8-gcc11-inductor / build (gh)
  ##[error]The operation was canceled.
• inductor / rocm6.1-py3.8-inductor / build (gh)
  ##[error]The operation was canceled.
• inductor-periodic / cuda12.1-py3.10-gcc9-sm80 (gh)
• inductor-periodic / cuda12.1-py3.10-gcc9-sm80 / build (gh)
  ##[error]The operation was canceled.
• inductor-periodic / cuda12.1-py3.10-gcc9-sm86-periodic-dynamo-benchmarks / build (gh)
  ##[error]The operation was canceled.
• Lint / lintrunner-clang / linux-job (gh)
  ##[error]The operation was canceled.
• Lint / lintrunner-noclang / linux-job (gh)
  ##[error]The operation was canceled.
• pull / linux-docs (gh)
• pull / linux-focal-cpu-py3.10-gcc9-bazel-test / build-and-test (default, 1, 1, linux.4xlarge) (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-cuda11.8-py3.10-gcc9 (gh)
• pull / linux-focal-cuda11.8-py3.10-gcc9 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-cuda12.1-py3.10-gcc9 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-cuda12.1-py3.10-gcc9-bazel-test / build-and-test (default, 1, 1, linux.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-cuda12.1-py3.10-gcc9-experimental-split-build / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-cuda12.1-py3.10-gcc9-sm86 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-cuda12.4-py3.10-gcc9-bazel-test / build-and-test (default, 1, 1, linux.4xlarge.nvidia.gpu) (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3_8-clang9-xla / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3-clang9-android-ndk-r21e-gradle-custom-build-single / build-and-test (default, 1, 1, linux.2xlarge) (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3-clang9-android-ndk-r21e-gradle-custom-build-single-full-jit / build-and-test (default, 1, 1, linux.2xlarge) (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3.11-clang10 (gh)
• pull / linux-focal-py3.11-clang10 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3.12-clang10 (gh)
• pull / linux-focal-py3.12-clang10 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3.12-clang10-experimental-split-build / test (default, 1, 3, linux.2xlarge) (gh)
• pull / linux-focal-py3.12-clang10-experimental-split-build / test (default, 2, 3, linux.2xlarge) (gh)
• pull / linux-focal-py3.12-clang10-experimental-split-build / test (default, 3, 3, linux.2xlarge) (gh)
• pull / linux-focal-py3.12-clang10-experimental-split-build / test (dynamo, 1, 3, linux.2xlarge) (gh)
• pull / linux-focal-py3.12-clang10-experimental-split-build / test (dynamo, 2, 3, linux.2xlarge) (gh)
• pull / linux-focal-py3.12-clang10-experimental-split-build / test (dynamo, 3, 3, linux.2xlarge) (gh)
  Can't find 'action.yml', 'action.yaml' or 'Dockerfile' under '/home/ec2-user/actions-runner/_work/pytorch/pytorch/.github/actions/get-workflow-job-id'. Did you forget to run actions/checkout before running your local action?
• pull / linux-focal-py3.8-clang10 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-py3.8-clang10-onnx (gh)
• pull / linux-focal-py3.8-clang10-onnx / build (gh)
  ##[error]The operation was canceled.
• pull / linux-focal-rocm6.1-py3.8 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-jammy-cuda11.8-cudnn9-py3.8-clang12 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-jammy-py3-clang12-executorch (gh)
• pull / linux-jammy-py3-clang12-executorch / build (gh)
  ##[error]The operation was canceled.
• pull / linux-jammy-py3.10-clang15-asan / build (gh)
  ##[error]The operation was canceled.
• pull / linux-jammy-py3.8-gcc11 / build (gh)
  ##[error]The operation was canceled.
• pull / linux-jammy-py3.8-gcc11-no-ops / build (gh)
  ##[error]The operation was canceled.
• pull / linux-jammy-py3.8-gcc11-pch / build (gh)
  ##[error]The operation was canceled.

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• pull / win-vs2019-cpu-py3 / build (gh) (matched win rule in flaky-rules.json)
  ##[error]The operation was canceled.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[linux-foundation-easycla]

• ✅ login: peterbell10 (6a39e2a, db150e7, adbb455, 8b989ce, d57fb50, be74674, afc5c1d, ae74cdc, a267f8d, b056613, 20657b9, 9caf1d5, 9d13415, 40fc01a, 09baca9, a922cba, 67499c1, 21c7f4e, 5d2348e, b425ee8, 9acca77, a0488af, 5187f11, 14a05c3)
• ❌ The email address for the commit (bff0618) is not linked to the GitHub account, preventing the EasyCLA check. Consult this Help Article and GitHub Help to resolve. (To view the commit's email address, add .patch at the end of this PR page's URL.) For further assistance with EasyCLA, please submit a support request ticket.