Refine NCCL search logic

[ooooo-create]

PR Category

Environment Adaptation

PR Types

Others

Description

统一 nvidia-nccl-cu12 为 2.28.3

[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.

[Copilot]

Pull request overview

This PR refines the NCCL library search logic by adding a new flag to control whether to search for NCCL in system paths when using pip CUDA libraries. This provides more control over library loading behavior for users with pip-installed CUDA packages.

Changes:

• Added a new boolean flag search_nccl_in_system (default: false) to control system path searching for NCCL
• Extended GetDsoHandleFromSearchPath function with an optional search_system_path parameter (default: true) to conditionally skip system default path searching
• Applied the new flag specifically to NCCL loading when PADDLE_WITH_PIP_CUDA_LIBRARIES is defined

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
paddle/common/flags.cc	Defines the new search_nccl_in_system flag with appropriate documentation
paddle/phi/backends/dynload/dynamic_loader.cc	Declares the flag, adds search_system_path parameter to GetDsoHandleFromSearchPath, and applies the flag to NCCL loading in pip CUDA library builds

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[SigureMo]

默认不查找的话，是否会再次出现之前 benchmark 平台上的问题？

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
⚠️ Please upload report for BASE (develop@d5a87b6). Learn more about missing BASE report.

Additional details and impacted files

@@             Coverage Diff             @@
##             develop    #77459   +/-   ##
===========================================
  Coverage           ?   100.00%           
===========================================
  Files              ?         2           
  Lines              ?         2           
  Branches           ?         0           
===========================================
  Hits               ?         2           
  Misses             ?         0           
  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.