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

Pytorch后端NHWC和NCHW问题 #69

Open
Windaway opened this issue Feb 7, 2023 · 0 comments
Open

Pytorch后端NHWC和NCHW问题 #69

Windaway opened this issue Feb 7, 2023 · 0 comments

Comments

@Windaway
Copy link

Windaway commented Feb 7, 2023

New Issue Checklist

Issue Description

Pytorch后端模型定义NHWC和NCHW数据格式主要是定以数据和模型后传到设备时用.to("cuda:0", memory_format=torch.channels_last)确定。

TLX目前做法是pytorch依据nhwc格式时，全部转NCHW然后处理完转回来，这潜在是让模型用NCHW格式计算。对纯GPU应用时问题不大，但是对于一些NHWC友好的设备部署，比如未来的Mindspore，由于多次nhwc nchw切换，性能有损失。

这里可能需要框架对于pytorch这里nhwc支持改成全局变量，即输入时数据做nchw-nhwc，模型转nhwc然后计算即可。

不过Pytorch本身GPU NHWC支持稀烂，倒不是很急。
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
1 participant
@Windaway