Skip to content

[Relax] Replaced call_pure_packed with tensor_to_shape operator #18616

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
tlopex merged 2 commits into from
Dec 30, 2025
Merged

[Relax] Replaced call_pure_packed with tensor_to_shape operator #18616

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1268299
Replaced call_pure_packed with tensor_to_shape operator
guan404ming Dec 26, 2025
5a933b0
Update python/tvm/relax/transform/legalize_ops/index.py
guan404ming Dec 26, 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
13 changes: 2 additions & 11 deletions python/tvm/relax/transform/legalize_ops/index.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@
# pylint: disable=invalid-name
"""Default legalization function for index operators."""
from tvm import topi, tir, te
from ...op import call_pure_packed
from ...op import tensor_to_shape
from ...block_builder import BlockBuilder
from ...expr import Call, Expr
from ...struct_info import ShapeStructInfo, PrimStructInfo
Expand Down Expand Up @@ -109,17 +109,8 @@ def get_length(begin, end, strides, length):
)

# 2. Convert tensor to shape and match cast with new symbolic vars
# Get shape length
ndim = int(output_shape.struct_info.shape[0])
output_shape = bb.emit(
# TODO(@relax-team): Ideally, we should use the tensor_to_shape op here to
# address the issue with purity, but that introduces a staging issue:
# we need to apply DecomposeOpsForInference in that case
# and it's unclear when in the build it should happen
call_pure_packed(
"vm.builtin.tensor_to_shape", output_shape, sinfo_args=ShapeStructInfo(ndim=ndim)
)
)
output_shape = bb.emit(tensor_to_shape(output_shape))
output_shape_vars = [tir.Var("s", "int64") for i in range(ndim)]
bb.match_cast(output_shape, ShapeStructInfo(output_shape_vars))

Expand Down
8 changes: 2 additions & 6 deletions tests/python/relax/test_transform_legalize_ops_index_linear_algebra.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -669,9 +669,7 @@ def main(
(x, begin, end, strides),
out_sinfo=R.Tensor((4,), dtype="int64"),
)
gv1: R.Shape(ndim=4) = R.call_pure_packed(
"vm.builtin.tensor_to_shape", gv, sinfo_args=(R.Shape(ndim=4),)
)
gv1: R.Shape(ndim=4) = R.tensor_to_shape(gv)
gv2: R.Shape([s, s_1, s_2, s_3]) = R.match_cast(
gv1, R.Shape([s, s_1, s_2, s_3])
)
Expand Down Expand Up @@ -868,9 +866,7 @@ def main(
(x, begin, end, strides),
out_sinfo=R.Tensor((2,), dtype="int64"),
)
gv1: R.Shape(ndim=2) = R.call_pure_packed(
"vm.builtin.tensor_to_shape", gv, sinfo_args=(R.Shape(ndim=2),)
)
gv1: R.Shape(ndim=2) = R.tensor_to_shape(gv)
gv2: R.Shape([s, s_1]) = R.match_cast(gv1, R.Shape([s, s_1]))
gv_1 = R.call_tir(
Expected.dynamic_strided_slice,
Expand Down
Loading