Skip to content

Conversation

@tensor-tang
Copy link
Contributor

When run benchmark, it spends much time on updating parameters.

This time can be improved with the OpenMP when enabled WITH_MKLDNN.

This PR adds the implement for SGD updater.

@tensor-tang tensor-tang requested a review from luotao1 September 25, 2017 09:26
Copy link
Contributor

@luotao1 luotao1 left a comment

Choose a reason for hiding this comment

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

有一个疑问:既然已经在sgdUpdateCpu函数里加了

#ifdef PADDLE_USE_MKLDNN
#pragma omp parallel for
#endif

为什么还要一个单独的sgdUpdateWithOMP呢?

@tensor-tang
Copy link
Contributor Author

两者函数申明不一样,不过确实是可以想办法综合下,是有点重复。

Copy link
Contributor

@luotao1 luotao1 left a comment

Choose a reason for hiding this comment

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

LGTM

@luotao1 luotao1 merged commit a7a82b9 into PaddlePaddle:develop Sep 26, 2017
@tensor-tang tensor-tang deleted the sgd branch September 26, 2017 03:46
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.

2 participants