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

NonZero will break cudagraph capture ! #4194

Closed
lix19937 opened this issue Oct 12, 2024 · 3 comments
Closed

NonZero will break cudagraph capture ! #4194

lix19937 opened this issue Oct 12, 2024 · 3 comments

Comments

@lix19937
Copy link

Description

NonZero will break cudagraph capture ! Because exist DevicetoShapeHostCopy

[01/01/2022-12:47:31] [I]                                                                                                                                                                                                                                                                                                         /NonZero[size][DevicetoShapeHostCopy]        1.12           0.0295             0.0289      0.0
[01/01/2022-12:47:31] [I]                                                                                                                                                                                                                                                                                                       /NonZero_2[size][DevicetoShapeHostCopy]        0.89           0.0235             0.0219      0.0
[01/01/2022-12:47:31] [I]                                                                                                                                                                                                                                                                                                       /NonZero_1[size][DevicetoShapeHostCopy]        0.74           0.0195             0.0188      0.0

Environment

TensorRT Version:8510

NVIDIA GPU:rtx2000

NVIDIA Driver Version:

CUDA Version:11.4

CUDNN Version:8.6
@lix19937
Copy link
Author

@ttyio So nozero can not support cudagraph !?

@ttyio
Copy link
Collaborator

ttyio commented Oct 14, 2024

Unfortunately, Yes, because we need device host sync to calculate the nonzero output shape ;-(

@lix19937
Copy link
Author

Thanks of your confirm. @ttyio

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
2 participants
@ttyio @lix19937