Skip to content

[Fix]: avoid pointer mutation in group norm kernel #999

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
Tcc0403 merged 1 commit into from
Jan 12, 2026
Merged

[Fix]: avoid pointer mutation in group norm kernel #999

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
11 changes: 5 additions & 6 deletions src/liger_kernel/ops/group_norm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,15 +78,14 @@ def _group_norm_forward_kernel(
for channel_idx in tl.range(group_idx * channels_per_group, (group_idx + 1) * channels_per_group):
W = tl.load(W_ptr + channel_idx)
B = tl.load(B_ptr + channel_idx)
for i in range(0, hidden_size_per_channel, BLOCK_SIZE):
# Calculate channel offset within the group
channel_offset = (channel_idx - group_idx * channels_per_group) * hidden_size_per_channel
for i in tl.range(0, hidden_size_per_channel, BLOCK_SIZE):
hidden_size_offsets = i + block_range
mask = hidden_size_offsets < hidden_size_per_channel
X = tl.load(X_ptr + hidden_size_offsets, mask=mask, other=m)
X = tl.load(X_ptr + channel_offset + hidden_size_offsets, mask=mask, other=m)
Y = (X - m) * rstd * W + B
tl.store(Y_ptr + hidden_size_offsets, Y, mask=mask)

X_ptr += hidden_size_per_channel
Y_ptr += hidden_size_per_channel
tl.store(Y_ptr + channel_offset + hidden_size_offsets, Y, mask=mask)

tl.store(Mean_ptr + batch_idx * Mean_row_stride + group_idx * Mean_col_stride, m)
tl.store(RSTD_ptr + batch_idx * RSTD_row_stride + group_idx * RSTD_col_stride, rstd)
Expand Down
Loading