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.

Sign up for GitHub

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

Jump to bottom

Add int8 nearest_interp and nearest_interp_v2 #37985

Merged
merged 1 commit into from
Dec 22, 2021
Merged

Add int8 nearest_interp and nearest_interp_v2 #37985

merged 1 commit into from
Dec 22, 2021

Conversation

wozna
Copy link
Contributor

@wozna wozna commented Dec 8, 2021 •
edited
Loading

PR types

New features

PR changes

OPs

Describe

This PR adds:

  • int8 and uint8 support for OneDNN kernels nearest_interp and nearest_interp_v2

I prepared post-training quantization on the ocr_det model and with quantized nearest_interp operator performance improved by 11%.

@paddle-bot-old
Copy link

paddle-bot-old bot commented Dec 8, 2021

Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

@wozna
Copy link
Contributor Author

wozna commented Dec 9, 2021

@lidanqing-intel @sfraczek Could you please review?

lidanqing-intel
lidanqing-intel approved these changes Dec 12, 2021
View reviewed changes
Copy link
Contributor

@lidanqing-intel lidanqing-intel left a comment

Choose a reason for hiding this comment

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

This PR improves both frcnn-r50-fpn and ocr_det models. It will speedup for all models quantization that contains nearest_interp op.

jakpiase
jakpiase approved these changes Dec 13, 2021
View reviewed changes
Copy link
Contributor

@jakpiase jakpiase left a comment

Choose a reason for hiding this comment

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

LGTM

@PaddlePaddle PaddlePaddle deleted a comment from sfraczek Dec 15, 2021
@PaddlePaddle PaddlePaddle deleted a comment from sfraczek Dec 15, 2021
@lidanqing-intel
Copy link
Contributor

@baoachun Hi, this PR will speedup retinanet int8 model performance, Could you please ask approvers to approve PR-CI-Static-Check, Thanks !

@paddle-bot-old
Copy link

Sorry to inform you that f146a06's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

Aurelius84
Aurelius84 approved these changes Dec 16, 2021
View reviewed changes
Copy link
Contributor

@Aurelius84 Aurelius84 left a comment

Choose a reason for hiding this comment

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

LGTM for Dtype registar

@lidanqing-intel
Copy link
Contributor

@baoachun @Aganlengzi Hi, could you please merge this PR. The descriptions are given. Thanks !

Aganlengzi
Aganlengzi approved these changes Dec 22, 2021
View reviewed changes
Copy link
Contributor

@Aganlengzi Aganlengzi left a comment

Choose a reason for hiding this comment

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

LGTM

@Aganlengzi Aganlengzi merged commit 56e2a6a into PaddlePaddle:develop Dec 22, 2021
lidanqing-intel pushed a commit to lidanqing-intel/Paddle that referenced this pull request Jan 7, 2022
@wozna @lidanqing-intel
Add nearest_interp/v2 int8 and uint8 support (PaddlePaddle#37985)
4fef7f3
@wozna wozna deleted the n_interp_int8 branch February 24, 2023 16:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@Aganlengzi Aganlengzi Aganlengzi approved these changes

@Aurelius84 Aurelius84 Aurelius84 approved these changes

@lidanqing-intel lidanqing-intel lidanqing-intel approved these changes

@jakpiase jakpiase jakpiase approved these changes

Assignees
No one assigned
Labels
int8 Intel
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
5 participants
@wozna @lidanqing-intel @Aganlengzi @Aurelius84 @jakpiase