Skip to content

Conversation

@mozga-intel
Copy link
Contributor

@mozga-intel mozga-intel commented May 16, 2018

The mkldnn mul operator was withdraw because the mul operator + (python) transpose matrix gets worse performance than the version of mklml.

@mozga-intel mozga-intel added Intel and removed Intel labels May 16, 2018
@mozga-intel mozga-intel requested a review from luotao1 May 17, 2018 15:22
@luotao1
Copy link
Contributor

luotao1 commented May 18, 2018

@mozga-intel Where is the (python) transpose matrix?

@mozga-intel
Copy link
Contributor Author

mozga-intel commented May 28, 2018

@luotao1 We must withdraw this operator because the mkldnn version of multiplication is different than the mul of PaddlePaddle. The operator of transpose is used in python version.

@luotao1 So, therefore please to withdraw this version of code.

@mozga-intel mozga-intel force-pushed the mozga-intel/Mul_mkldnn_withdraw branch 2 times, most recently from 6809be3 to f92fa2d Compare May 29, 2018 14:03
luotao1
luotao1 previously approved these changes May 30, 2018
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!Could you please merge the latest develop code to pass the teamcity?

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! Thanks very much!

@luotao1 luotao1 merged commit ccf61b3 into PaddlePaddle:develop May 31, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants