Skip to content

Add dim order assert #7561

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

Merged
digantdesai merged 4 commits into from
Jan 14, 2025
Merged

Add dim order assert #7561

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
45df002
Add dim order assert for MTK backend
Dec 25, 2024
4d86272
Add dim order guard for mediatek backend AOT part
neuropilot-captain Jan 9, 2025
f0cc259
Fix lintrunner error
neuropilot-captain Jan 13, 2025
3c7c854
Remove unnecessary code
MediaTek-ai Jan 14, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion backends/mediatek/partitioner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,7 +46,7 @@ def is_node_supported(self, _, node: torch.fx.Node) -> bool:
op_type = node.target.__name__

# Skip until we can handle the dimension order representation
if op_type == 'aten._to_copy.default':
if op_type == "aten._to_copy.default":
Copy link
Contributor

@digantdesai digantdesai Jan 14, 2025

Choose a reason for hiding this comment

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

I assume you don't partition empty - that is another one of the dim_order ops. Just FYI.

return False

if op_type in self._op_types_to_skip or node.name in self._op_names_to_skip:
Expand Down
2 changes: 1 addition & 1 deletion backends/mediatek/preprocess.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,7 @@ def preprocess(
) -> PreprocessResult:

# Make sure all inputs are contiguous_format or NCHW or default dim order
print('here')
print("here")
assert_default_dim_order(edge_program.graph_module)

name_to_node_mappings = {node.name: node for node in edge_program.graph.nodes}
Expand Down
3 changes: 2 additions & 1 deletion backends/mediatek/runtime/NeuronBackend.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -114,7 +114,8 @@ Error NeuronExecuTorchDelegate::execute(
for (int i = 0; i < inputCount; i++) {
auto tensor_in = args[i]->toTensor();
ET_CHECK_OR_RETURN_ERROR(
runtime::is_contiguous_dim_order(tensor_in.dim_order().data(), tensor_in.dim()),
runtime::is_contiguous_dim_order(
tensor_in.dim_order().data(), tensor_in.dim()),
Internal,
"Expecting default dim_order but got a non default dim_order tensor for external input %u",
i);
Expand Down
Loading