Skip to content

[TIR][Schedule] TileWithTensorIntrin skip incorrect ComputeInline for input-padding #16239

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
tqchen merged 1 commit into from
Dec 19, 2023
Merged

[TIR][Schedule] TileWithTensorIntrin skip incorrect ComputeInline for input-padding #16239

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
6 changes: 4 additions & 2 deletions src/tir/schedule/transform.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -318,10 +318,12 @@ Optional<LoopRV> TileWithTensorIntrin(const tir::Schedule& sch, const tir::Block
auto producers = sch->GetProducers(block_rv);
for (const auto& producer : producers) {
auto original_producers = sch->GetProducers(producer);
ICHECK_EQ(original_producers.size(), 1u);
// NOTICE: there may not all producers padded.
// Inline the original producer into the padding block. This ensures that the new producer
// has the padded shape.
sch->ComputeInline(original_producers[0]);
if (original_producers.size() == 1u) {
sch->ComputeInline(original_producers[0]);
}
}
auto consumers = sch->GetConsumers(block_rv);
for (const auto& consumer : consumers) {
Expand Down