Skip to content

Cortex-M: Don't replace channels-first conv with channels-last kernel #23943

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

Open
rahimuddin140 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Cortex-M: Don't replace channels-first conv with channels-last kernel #23943

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
11 changes: 11 additions & 0 deletions backends/cortex_m/passes/aten_to_cortex_m_pass.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@

from executorch.backends.cortex_m.utils import (
build_activation_lut,
is_channels_last,
is_foldable_alpha,
quantize_multiplier_aot,
quantize_val,
Expand Down Expand Up @@ -461,6 +462,16 @@ def _get_convolution_replacement(
cast(int, conv_args[8]),
)

if not explicit_nhwc:
input_tensor = get_first_fake_tensor(cast(Node, x))
if input_tensor is None or not is_channels_last(input_tensor):
# cortex_m::quantized_conv2d / quantized_depthwise_conv2d require
# channels-last inputs and outputs; replacing a channels-first conv
# produces a graph that fails at runtime with
# "input and output must be channels_last" (#23812).
# Leave the node un-replaced so it falls back to portable ops.
return None

if transposed:
return _get_transpose_conv2d_replacement(node, dialect_pass)

Expand Down