Skip to content

Conversation

@ZzSean
Copy link
Contributor

@ZzSean ZzSean commented Dec 30, 2021

PR types

Performance optimization

PR changes

OPs

Describe

Optimize performance of softmax_fwd when axis!=-1

case pytorch paddle 优化前 diff paddle 优化后 diff 加速比
fp32,[128,128,16,16] axis:0 0.23059 0.18179 优于 (26.84%) 0.07534 优于 (2.06x) 2.41
fp16,[128,128,16,16] axis:0 0.20328 0.14136 优于 (43.80%) 0.06305 优于 (2.22x) 2.24
fp32,[512,896,4,12] axis:1 0.76943 1.22761 差于 (59.55%) 0.56893 优于 (35.24%) 2.16
fp16,[512,896,4,12] axis:1 0.62722 1.02623 差于 (63.62%) 0.41894 优于 (49.72%) 2.45
@paddle-bot-old
Copy link

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

Liu-xiandong
Liu-xiandong previously approved these changes Jan 6, 2022
Copy link
Member

@Liu-xiandong Liu-xiandong left a comment

Choose a reason for hiding this comment

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

LGTM

@paddle-bot-old
Copy link

Sorry to inform you that e0b6286's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

@PaddlePaddle PaddlePaddle locked and limited conversation to collaborators Jan 14, 2022
@PaddlePaddle PaddlePaddle unlocked this conversation Jan 14, 2022
@paddle-bot-old
Copy link

Sorry to inform you that 418445b's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

Copy link
Contributor

@AnnaTrainingG AnnaTrainingG left a comment

Choose a reason for hiding this comment

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

LGTM

@ZzSean ZzSean merged commit 8e1b020 into PaddlePaddle:develop Feb 9, 2022
@ZzSean ZzSean deleted the opt_softmax_fwd branch November 7, 2022 03:02
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

4 participants