Skip to content

[WebGPU EP] extend concat to handle large number of inputs #25177

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
Open
wants to merge 4 commits into
base: main
Choose a base branch
from
Open

[WebGPU EP] extend concat to handle large number of inputs #25177

wants to merge 4 commits into from
+116 −5

Conversation

prathikr
Copy link
Contributor

Resolves the following issue affecting the patchtst model:

Non-zero status code returned while running Concat node. Name:'/patching/Concat' 
Status Message: The number of storage buffer (input=32, output=1) exceeds the limit 
(8) of the device.

@prathikr prathikr changed the title [WebGPU EP] extend concat to handle >maxStorageBuffersPerShaderStage inputs [WebGPU EP] extend concat to handle large number of inputs Jun 25, 2025
@prathikr prathikr requested review from fs-eire and guschmue June 26, 2025 20:37
fs-eire
fs-eire reviewed Jun 26, 2025
View reviewed changes
onnxruntime/core/providers/webgpu/tensor/concat.cc
Comment on lines +151 to +152
// Create intermediate tensor to store result of this pass
auto intermediate = context.CreateGPUTensor(current_inputs[0]->DataType(), std::move(pass_shape));
Copy link
Contributor

Choose a reason for hiding this comment

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

Is it possible to always use prepare.output_tensor as output so that creating intermediate buffers can be avoided?

@guschmue guschmue added the ep:WebGPU ort-web webgpu provider label Jul 1, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@fs-eire fs-eire fs-eire left review comments

@guschmue guschmue Awaiting requested review from guschmue

At least 1 approving review is required to merge this pull request.

Assignees
No one assigned
Labels
ep:WebGPU ort-web webgpu provider
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@prathikr @fs-eire @guschmue