Skip to content

Enable QuantFusionPass in compiler pipeline (#19728) #19728

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
SS-JIA merged 7 commits into from
May 30, 2026
Merged

Enable QuantFusionPass in compiler pipeline (#19728) #19728

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
845eace
Add shared fusion infrastructure and QuantFusionPass (#19724)
ethansfng May 28, 2026
ac84526
Add fuse() to QuantizationPatterns (#19726)
ethansfng May 28, 2026
60b0da7
Add fuse() to remaining QuantizationPatterns (#19727)
ethansfng May 28, 2026
2cf3fd2
Enable QuantFusionPass in compiler pipeline (#19728)
ethansfng May 28, 2026
facfb2f
Merge branch 'main' into export-D105728219
SS-JIA May 29, 2026
d98b427
Merge branch 'main' into export-D105728219
SS-JIA May 30, 2026
01e6816
Merge branch 'main' into export-D105728219
SS-JIA May 30, 2026
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: 0 additions & 2 deletions backends/cadence/aot/BUCK
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,7 +44,6 @@ fbcode_target(_kind = runtime.python_library,
":compiler_funcs",
":utils",
"//caffe2:torch",
"//executorch/backends/cadence/aot/quantizer:fusion_pass",
"//executorch/backends/cadence/aot/quantizer/passes:fuse_ops",
"//executorch/backends/cadence/aot/quantizer:quantizer",
"//executorch/backends/transforms:decompose_sdpa",
Expand All @@ -65,7 +64,6 @@ fbcode_target(_kind = runtime.python_library,
":replace_ops",
":utils",
"//caffe2:torch",
"//executorch/backends/cadence/aot/quantizer:fusion_pass",
"//executorch/backends/cadence/aot/quantizer:quantizer",
"//executorch/backends/cadence/runtime:runtime",
"//executorch/backends/transforms:decompose_sdpa",
Expand Down
8 changes: 4 additions & 4 deletions backends/cadence/aot/compiler.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,14 +14,14 @@
import torch
from executorch.backends.cadence.aot.compiler_funcs import (
prepare as prepare_fn,
QuantFusionPass,
QuantizedInputWrapper,
trace as trace_fn,
)
from executorch.backends.cadence.aot.memory_planning import (
CadenceMemoryPlanning,
print_memory_planning_info,
)
from executorch.backends.cadence.aot.quantizer.fusion_pass import QuantFusion
from executorch.backends.cadence.aot.quantizer.passes.fuse_ops import FuseQATConvBN
from executorch.backends.cadence.aot.quantizer.quantizer import (
CadenceDefaultQuantizer,
Expand Down Expand Up @@ -154,9 +154,9 @@ def apply_pre_edge_transform_passes(
quantizer: CadenceQuantizer,
) -> ExportedProgram:
"""
Apply pre-edge transform passes including QuantFusion and torch ops passes.
Apply pre-edge transform passes including QuantFusionPass and torch ops passes.
This mirrors the Cadence AOT compiler flow:
1. QuantFusion - fuses dq->op->q patterns
1. QuantFusionPass - fuses dq->op->q patterns
2. apply_torch_ops_passes - applied just before to_edge()

The quantizer must be the same as the one used to convert the model.
Expand All @@ -169,7 +169,7 @@ def apply_pre_edge_transform_passes(
PassManager(
[
FuseQATConvBN(converted_program),
QuantFusion(patterns),
QuantFusionPass(patterns),
]
)(converted_program.graph_module)

Expand Down
Loading