Skip to content

Fix abort issue in cpu multi-threads#11233

Merged
tensor-tang merged 6 commits intoPaddlePaddle:developfrom
tensor-tang:multithreads
Jun 6, 2018
Merged

Fix abort issue in cpu multi-threads#11233
tensor-tang merged 6 commits intoPaddlePaddle:developfrom
tensor-tang:multithreads

Conversation

@tensor-tang
Copy link
Contributor

@tensor-tang tensor-tang commented Jun 6, 2018 •
edited
Loading

fix #11231

The key backtrace is :

#16 0x00000000029b4640 in paddle::framework::Scope::Var (this=0x5cfd0e0, name="lstm_0.w_0")
    at /home/tangjian/paddle-docker/paddle/fluid/framework/scope.cc:55
#17 0x0000000002705d91 in paddle::framework::Executor::CreateVariables (this=0x7f07caffcab0, pdesc=...,
    scope=0x7f07b8022850, block_id=0) at /home/tangjian/paddle-docker/paddle/fluid/framework/executor.cc:94
#18 0x0000000002709942 in paddle::framework::Executor::RunPreparedContext (this=0x7f07caffcab0, ctx=0x7f07b8022990,
    scope=0x5cfd0e0, create_local_scope=true, create_vars=true)
    at /home/tangjian/paddle-docker/paddle/fluid/framework/executor.cc:316
#19 0x0000000002706597 in paddle::framework::Executor::Run (this=0x7f07caffcab0, pdesc=..., scope=0x5cfd0e0,
    block_id=0, create_local_scope=true, create_vars=true)
    at /home/tangjian/paddle-docker/paddle/fluid/framework/executor.cc:119
#20 0x00000000022e875e in paddle::inference::LoadPersistables (executor=0x7f07caffcab0, scope=0x5cfd0e0,
    main_program=..., dirname="/home/tangjian/nlp/epoch6", param_filename="")
    at /home/tangjian/paddle-docker/paddle/fluid/inference/io.cc:108
#21 0x00000000022e8c34 in paddle::inference::Load (executor=0x7f07caffcab0, scope=0x5cfd0e0,
    dirname="/home/tangjian/nlp/epoch6") at /home/tangjian/paddle-docker/paddle/fluid/inference/io.cc:124
#22 0x0000000002244eed in ThreadRunInfer (tid=5, scope=0x5cfd0e0, jobs=std::vector of length 6, capacity 6 = {...})
    at /home/tangjian/paddle-docker/paddle/fluid/inference/tests/book/test_inference_nlp.cc:111
@tensor-tang tensor-tang added the 预测 原名Inference，包含Capi预测问题等 label Jun 6, 2018
@tensor-tang
Copy link
Contributor Author

tensor-tang commented Jun 6, 2018 •
edited
Loading

https://paddleci.ngrok.io/viewLog.html?tab=buildLog&logTab=tree&filter=debug&expand=all&buildId=38034&_focus=5958

test_listen_and_serv_op .........................***Timeout 31.59 sec
@tensor-tang
Merge remote-tracking branch 'ups/develop' into multithreads
64323b1
@tensor-tang
fix conflcts
4b7b17a 
Merge remote-tracking branch 'ups/develop' into multithreads
@tensor-tang tensor-tang requested a review from panyx0718 June 6, 2018 12:48
panyx0718
panyx0718 approved these changes Jun 6, 2018
View reviewed changes
paddle/fluid/inference/tests/book/test_inference_nlp.cc
new paddle::framework::ProgramDesc(*inference_program));
// maybe framework:ProgramDesc is not thread-safe
auto& sub_scope = scope->NewScope();
auto place = paddle::platform::CPUPlace();
Copy link
Contributor

@panyx0718 panyx0718 Jun 6, 2018

Choose a reason for hiding this comment

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

paddle::platform::CPUPlace place;
paddle::framework::Executor executor(place);
paddle/fluid/inference/tests/book/test_inference_nlp.cc
executor.CreateVariables(*inference_program, scope.get(), 0);
// 1. Define place, executor, scope
auto place = paddle::platform::CPUPlace();
auto executor = paddle::framework::Executor(place);
Copy link
Contributor

@panyx0718 panyx0718 Jun 6, 2018

Choose a reason for hiding this comment

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

same as above
@tensor-tang tensor-tang merged commit 3a29404 into PaddlePaddle:develop Jun 6, 2018
@tensor-tang tensor-tang deleted the multithreads branch June 6, 2018 17:58
@tensor-tang tensor-tang mentioned this pull request Jun 7, 2018
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

预测 原名Inference，包含Capi预测问题等

2 participants

@tensor-tang @panyx0718