Skip to content

(alternative to #2508) Fix/splitk tmp_out undersized buffer avoid double-zeroing #2551

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
rbrugaro-amd merged 8 commits into from
Mar 31, 2026
Merged

(alternative to #2508) Fix/splitk tmp_out undersized buffer avoid double-zeroing #2551

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
f7ddc47
fix: align ck_moe_stage1 split-K tmp_out buffer with CK kernel
karverma-amd Mar 27, 2026
3f50525
Merge branch 'main' into main
karverma-amd Mar 30, 2026
857c9d2
Update fused_moe.py
karverma-amd Mar 30, 2026
5063b5b
Merge branch 'main' into main
karverma-amd Mar 31, 2026
3c8ef17
tmp_out to use torch.empty vs. torch.zeros to avoid double zeroing
rbrugaro-amd Mar 31, 2026
61850e8
Merge branch 'main' into fix/splitk-tmpout-and-ck-memset
rbrugaro-amd Mar 31, 2026
7b9fe7d
tighten valid_out slice: drop redundant .contiguous() and .view(dtype…
rbrugaro-amd Mar 31, 2026
2633e68
restore .view(dtypes.fp32) on valid_out for silu_and_mul/gelu_and_mul
rbrugaro-amd Mar 31, 2026
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
24 changes: 16 additions & 8 deletions aiter/fused_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1707,13 +1707,14 @@ def ck_moe_stage1(
):
token_num = hidden_states.shape[0]
is_splitk = quant_type is aiter.QuantType.per_1x128 and splitk > 1
tmp_out = (
torch.zeros(
(token_num, topk, w1.shape[1]), dtype=dtypes.fp32, device=out.device
if is_splitk:
# CK kernel zeros this buffer via hipMemsetAsync when KBatch > 1
sorted_size = min(token_num * topk * block_m, sorted_token_ids.shape[0])
tmp_out = torch.empty(
(sorted_size, w1.shape[1]), dtype=dtypes.fp32, device=out.device
)
if is_splitk
else out
)
else:
tmp_out = out
aiter.ck_moe_stage1_fwd(
hidden_states,
w1,
Expand All @@ -1735,10 +1736,11 @@ def ck_moe_stage1(
out.dtype,
)
if is_splitk:
valid_out = tmp_out[: token_num * topk, :]
if activation == ActivationType.Silu:
aiter.silu_and_mul(out, tmp_out.view(dtypes.fp32))
aiter.silu_and_mul(out, valid_out.view(dtypes.fp32))
else:
aiter.gelu_and_mul(out, tmp_out.view(dtypes.fp32))
aiter.gelu_and_mul(out, valid_out.view(dtypes.fp32))
return out


Expand Down Expand Up @@ -1773,6 +1775,12 @@ def cktile_moe_stage1(
D = D * 8

out = torch.empty((token_num, topk, D), dtype=dtype, device=hidden_states.device)
# WARNING: when split_k > 1, this allocation has the same undersized buffer
# pattern fixed in ck_moe_stage1 (see ROCm/aiter#2508). If the CK tile
# kernel calls hipMemsetAsync with sorted_size rows, this will overflow.
# When fp32 splitk is enabled, apply the same fix: use sorted_size =
# min(token_num * topk * block_m, sorted_token_ids.shape[0]) and slice
# valid_out = tmp_out[:token_num * topk, :] before silu_and_mul/gelu_and_mul.
tmp_out = (
torch.zeros(
(token_num, topk, w1.shape[1]), dtype=hidden_states.dtype, device=out.device
Expand Down
Loading