Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

[BugFix]Fix OneDNN Kernels Bug when use pass #48364

Merged
merged 2 commits into from
Nov 28, 2022

Conversation

YuanRisheng
Copy link
Contributor

@YuanRisheng YuanRisheng commented Nov 25, 2022

PR types

Others

PR changes

Others

Describe

This PR Fix bugs of ExtraAttr's mechanism.
Because ExtraAttrs are stored in dev_ctx and dev_ctx is a global object. So, it is easy to use old ExtraAttrs in next program's iteration or next kernel execution. This will generate incorrect results. This PR fix this bug.
Related PR:#47657 #48162

@paddle-bot
Copy link

paddle-bot bot commented Nov 25, 2022

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.

Silv3S added a commit to Silv3S/Paddle that referenced this pull request Nov 25, 2022
Copy link
Member

@Silv3S Silv3S left a comment

Choose a reason for hiding this comment

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

LGTM 👍
Thanks for resolving this issue

@YuanRisheng YuanRisheng merged commit df82fd3 into PaddlePaddle:develop Nov 28, 2022
YuanRisheng pushed a commit that referenced this pull request Nov 29, 2022
* cleanup unused code

* unify is_int8 is_bfloat16

* Simplify matmul_v2 FWD kernel

* remove RunKernel methods

* remove import namespace

* remove headers

* clean fluid/phi cross imports

* remove fluid axpy_handler

* delete fluid methods

* activations

* OneDNNMemDesc

* MKLDNNFormatForSize

* MatchShapeToLayout

* MKLDNNMemoryFormat

* MKLDNNFormat

* ReorderMKLDNNHandler

* to_void_cast

* review suggestions

* interpolate

* remove fluid depedency

* init

* ExecuteMatMulV2

* rm fluid kernel

* matmul_grad

* remove mutable_data

* mul_grad

* matmul fwd

* add extra attr

* temp disable passes

* re-enable passes

* workaround for matmul+act

* fix for matmul+eltwise_add

* fix typo

* merge bugfix #48364

* remove merge conflict
lxsbupt pushed a commit to lxsbupt/Paddle that referenced this pull request Dec 17, 2022
* cleanup unused code

* unify is_int8 is_bfloat16

* Simplify matmul_v2 FWD kernel

* remove RunKernel methods

* remove import namespace

* remove headers

* clean fluid/phi cross imports

* remove fluid axpy_handler

* delete fluid methods

* activations

* OneDNNMemDesc

* MKLDNNFormatForSize

* MatchShapeToLayout

* MKLDNNMemoryFormat

* MKLDNNFormat

* ReorderMKLDNNHandler

* to_void_cast

* review suggestions

* interpolate

* remove fluid depedency

* init

* ExecuteMatMulV2

* rm fluid kernel

* matmul_grad

* remove mutable_data

* mul_grad

* matmul fwd

* add extra attr

* temp disable passes

* re-enable passes

* workaround for matmul+act

* fix for matmul+eltwise_add

* fix typo

* merge bugfix PaddlePaddle#48364

* remove merge conflict
# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants