Skip to content

register quantized_linear.per_tensor in lib #6563

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
merged 1 commit into from
Nov 8, 2024
Merged

register quantized_linear.per_tensor in lib #6563

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
28 changes: 27 additions & 1 deletion backends/cadence/aot/ops_registrations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,7 +50,11 @@
"quantized_linear.out(Tensor src, Tensor weight, Tensor bias, int src_zero_point, Tensor weight_zero_point, Tensor out_multiplier, Tensor out_shift, int out_zero_point, Tensor? offset, *, Tensor(a!) out) -> Tensor(a!)"
)
lib.define(
"cadence::quantized_linear.per_tensor_out(Tensor src, Tensor weight, Tensor bias, SymInt src_zero_point, SymInt weight_zero_point, SymInt out_multiplier, SymInt out_shift, SymInt out_zero_point, Tensor? offset, *, Tensor(a!) out) -> Tensor(a!)"
"quantized_linear.per_tensor_out(Tensor src, Tensor weight, Tensor bias, SymInt src_zero_point, SymInt weight_zero_point, SymInt out_multiplier, SymInt out_shift, SymInt out_zero_point, Tensor? offset, *, Tensor(a!) out) -> Tensor(a!)"
)
lib.define(
"quantized_linear.per_tensor(Tensor src, Tensor weight, Tensor bias, SymInt src_zero_point, "
"SymInt weight_zero_point, SymInt out_multiplier, SymInt out_shift, SymInt out_zero_point, Tensor? offset) -> Tensor"
)

lib.define(
Expand Down Expand Up @@ -129,6 +133,28 @@ def quantized_linear_meta(
return src.new_empty(out_size, dtype=src.dtype)


@register_fake("cadence::quantized_linear.per_tensor")
def quantized_linear_per_tensor_meta(
src: torch.Tensor,
weight: torch.Tensor,
bias: torch.Tensor,
in_zero_point: torch.SymInt,
weight_zero_point: torch.SymInt,
out_multiplier: torch.SymInt,
out_shift: torch.SymInt,
out_zero_point: torch.SymInt,
offset: Optional[torch.Tensor],
) -> torch.Tensor:
# src comes in shape [leading_dims, in_dim]
# weight comes in shape [out_dim, in_dim]
# output comes in empty with shape [leading_dims, out_dim]
out_size = list(src.size())
weight_size = list(weight.size())
assert len(weight_size) == 2
out_size[-1] = weight_size[0]
return src.new_empty(out_size, dtype=src.dtype)


@register_fake("cadence::quantized_conv")
def quantized_conv_meta(
input: torch.Tensor,
Expand Down
Loading