Skip to content

Conversation

@wanghuancoder
Copy link
Contributor

@wanghuancoder wanghuancoder commented Jul 22, 2024

PR Category

Execute Infrastructure

PR Types

Bug fixes

Description

PinnedMemory应该是CPU backend

但由于目前套件库中PinnedMemory的Tensor大多都选择了GPU做kernel运算。修改后会导致非常多套件库性能下降需要修复:

  1. 套件库Dataloader的Tensor如果是PinnedMemory需要集中转为GPU
  2. 套件库中间变量如果是PinnedMemory需要转为GPU在运算
    第二点非常难定位和修复,因此暂时通过FLGAS开关提供正确features。

Pcard-67164

@paddle-bot
Copy link

paddle-bot bot commented Jul 22, 2024

你的PR提交成功,感谢你对开源项目的贡献!
请关注后续CI自动化测试结果,详情请参考Paddle-CI手册
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

zyfncg
zyfncg previously approved these changes Jul 22, 2024
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

@wanghuancoder wanghuancoder merged commit 1ec5ead into PaddlePaddle:develop Jul 24, 2024
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.

3 participants