Skip to content
New issue

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

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

Already on GitHub? Sign in to your account

【PIR OpTest Fix No.35】 fix test_batch_fc_op #62668

Merged
merged 15 commits into from
Mar 21, 2024

Conversation

Dmovic
Copy link
Contributor

@Dmovic Dmovic commented Mar 12, 2024

PR types

Others

PR changes

Others

Description

PIR Op单测修复
修复单测 fix test_batch_fc_op
修复后打开FLAGS_enable_pir_in_executor单测是否通过:是

Copy link

paddle-bot bot commented Mar 12, 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.

@CLAassistant
Copy link

CLAassistant commented Mar 12, 2024

CLA assistant check
All committers have signed the CLA.

@paddle-bot paddle-bot bot added the contributor External developers label Mar 12, 2024
@luotao1 luotao1 added the HappyOpenSource 快乐开源活动issue与PR label Mar 13, 2024
@kangguangli
Copy link
Contributor

如果可以通过的话,需要在 test/white_list/pir_op_test_white_list 中增加这个单测 test_batch_fc_op

@Dmovic
Copy link
Contributor Author

Dmovic commented Mar 13, 2024

如果可以通过的话,需要在 test/white_list/pir_op_test_white_list 中增加这个单测 test_batch_fc_op

好的,还想请教一个问题,PR-CI-Coverage这个ci测试失败了,看了一下报错是添加的InferShape部分,这个是因为InferShape函数没有注册上去吗

image

image

@kangguangli
Copy link
Contributor

如果可以通过的话,需要在 test/white_list/pir_op_test_white_list 中增加这个单测 test_batch_fc_op

好的,还想请教一个问题,PR-CI-Coverage这个ci测试失败了,看了一下报错是添加的InferShape部分,这个是因为InferShape函数没有注册上去吗

这倒不是,这是因为没跑到这段代码,覆盖率没过。理论上你添加单测后应该就能测试到这段代码了。

@Dmovic
Copy link
Contributor Author

Dmovic commented Mar 17, 2024

查看Q&A,为了解决RuntimeError: (PreconditionNotMet) op [pd_op.xxx] kernel output args defs should equal op outputs这个问题,在LegacyOpList加入了BatchFC

const std::unordered_set<std::string> LegacyOpList = {
    LoadCombineOp::name(),
    BatchFCOp::name(),
    BatchFCGradOp::name(),

但是在build阶段遇到了如下报错

[ 48%] Building CXX object paddle/fluid/memory/allocation/CMakeFiles/allocator.dir/cuda_virtual_mem_allocator.cc.o
/paddle/paddle/fluid/pir/dialect/operator/utils/utils.cc:40:5: error: ???BatchFCOp??? has not been declared
BatchFCOp::name(),
/paddle/paddle/fluid/pir/dialect/operator/utils/utils.cc:41:5: error: ???BatchFCGradOp??? has not been declared
BatchFCGradOp::name(),

@kangguangli
Copy link
Contributor

BatchFCOp

名字错了,按照驼峰命名,这里会是BatchFcOp。可以查看下编译生成的paddle/fluid/pir/dialect/operator/ir/pd_op.h确认下。

@Dmovic
Copy link
Contributor Author

Dmovic commented Mar 18, 2024

BatchFCOp

名字错了,按照驼峰命名,这里会是BatchFcOp。可以查看下编译生成的paddle/fluid/pir/dialect/operator/ir/pd_op.h确认下。

谢谢,fix这个问题之后已经能够通过CI测试了

Co-authored-by: kangguangli <kangguangli@hotmail.com>
@Dmovic
Copy link
Contributor Author

Dmovic commented Mar 19, 2024

谢谢,已按照建议修改

kangguangli
kangguangli previously approved these changes Mar 20, 2024
paddle/phi/infermeta/ternary.h Outdated Show resolved Hide resolved
paddle/phi/infermeta/ternary.h Outdated Show resolved Hide resolved
paddle/phi/infermeta/ternary.h Outdated Show resolved Hide resolved
@Dmovic
Copy link
Contributor Author

Dmovic commented Mar 20, 2024

好的,已按照建议修改

@kangguangli kangguangli requested a review from zyfncg March 21, 2024 11:14
@kangguangli kangguangli merged commit 49c09ed into PaddlePaddle:develop Mar 21, 2024
29 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
contributor External developers HappyOpenSource 快乐开源活动issue与PR
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants