[API] Add cudnn_available() function to check DNN library availability

[YqGe585]

PR Category

Custom Device

PR Types

Bug fixes

Description

Add a new Python API paddle.device.cudnn_available() to check whether cuDNN or equivalent DNN library is available. This simplifies the code that previously used get_cudnn_version() just for availability checks.

Changes:

1. C++ Layer (paddle/fluid/pybind/pybind.cc):

   • Add CudnnAvailable() function that:
     • Returns platform::DnnVersion() > 0 for CUDA/HIP builds
     • Checks custom devices via phi::DeviceManager::IsDnnAvailable(place) for custom device builds
     • Returns false otherwise
   • Add pybind binding m.def("cudnn_available", CudnnAvailable);

2. Python Layer (python/paddle/device/__init__.py):

   • Add cudnn_available() function that calls core.cudnn_available()
   • Export in __all__ list

3. Functional Conv (python/paddle/nn/functional/conv.py):

   • Replace 6 get_cudnn_version() usages with cudnn_available() in:
     • conv1d, conv2d, conv1d_transpose, conv2d_transpose, conv3d, conv3d_transpose

4. Layer Conv (python/paddle/nn/layer/conv.py):

   • Replace get_cudnn_version() usage in _ConvNd class initialization

Benefits:

• Cleaner code: use_cudnn = cudnn_available() instead of multi-line version checks
• Better custom device support: Automatically handles custom device DNN libraries
• Fixes undefined is_custom_device variable in conv1d_transpose

是否引起精度变化

否

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[YqGe585]

/re-run all-failed

[codecov-commenter]

Codecov Report

❌ Patch coverage is 92.30769% with 1 line in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (develop@06b60d2). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
python/paddle/nn/functional/vision.py	66.66%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             develop   #77880   +/-   ##
==========================================
  Coverage           ?   92.30%           
==========================================
  Files              ?        4           
  Lines              ?       13           
  Branches           ?        0           
==========================================
  Hits               ?       12           
  Misses             ?        1           
  Partials           ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[YqGe585]

/re-run all-failed