Skip to content

Conversation

@zyfncg
Copy link
Contributor

@zyfncg zyfncg commented Feb 5, 2024

PR types

Others

PR changes

Others

Description

Pcard-73448

修复generate_shape拆分为普通shape类算子时使用cache导致Value跨block引用的问题。本PR中移除了cache的使用,可能会有一定的性能损失,但预计影响不大,后续出现性能问题时再进行专项优化。

@paddle-bot
Copy link

paddle-bot bot commented Feb 5, 2024

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.

Comment on lines -380 to +366
return op->isa<pir::ModuleOp>() && op->num_regions() > 0;
return op->num_regions() > 0;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这里为什么不需要判断 ModuleOp 了呢?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这个pass也要进到if的子block中处理,需要去掉ModuleOp的判断才能进去

@tc20042008 tc20042008 merged commit 92eaa7d into PaddlePaddle:develop Feb 6, 2024
@zyfncg zyfncg deleted the fix_split_generate_shape_into_shape_ops_pass branch February 6, 2024 02:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants