Skip to content

[Bugfix] revert incorrect fast path for parallel layout inference #1730

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
LeiWang1999 merged 1 commit into from
Jan 26, 2026
Merged

[Bugfix] revert incorrect fast path for parallel layout inference #1730

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
20 changes: 0 additions & 20 deletions src/op/parallel.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -668,27 +668,7 @@ ParallelOpNode::ComputeLoopLayoutFromBuffer(const Buffer &buffer,

Fragment result;

// Check if access indices match loop vars AND loop ranges match layout input
// shape. We can only use src_layout directly when both conditions are met.
// Otherwise, if the loop is a sub-region of the buffer (e.g., loop is 4x128
// but buffer layout is 64x128), using the full layout would cause index
// out-of-bounds.
bool can_use_src_layout_directly = false;
if (IsCommonAccessIndice(buffer)) {
auto input_shape = src_layout->InputShape();
if (input_shape.size() == loop_vars_.size()) {
can_use_src_layout_directly = true;
for (size_t i = 0; i < loop_vars_.size(); i++) {
if (!analyzer_.CanProveEqual(loop_vars_[i]->dom->extent,
input_shape[i])) {
can_use_src_layout_directly = false;
break;
}
}
}
}

if (can_use_src_layout_directly) {
result = src_layout;
} else {
Var rep;
Expand Down
Loading