Skip to content

Fix torch compile run #7391

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
merged 4 commits into from
Jun 22, 2025
Merged

Fix torch compile run #7391

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
d00dec7
Fix torch-compile run will hit compiling error
Jun 20, 2025
293223c
fix lint error
Jun 20, 2025
2c23e4e
Merge branch 'main' into fix-torch-compile-run
zhyncs Jun 20, 2025
5c4e1cd
Update image to include aiter update.
saienduri Jun 21, 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
2 changes: 1 addition & 1 deletion docker/Dockerfile.rocm
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ ARG TRITON_COMMIT="improve_fa_decode_3.0.0"


ARG AITER_REPO="https://github.com/ROCm/aiter.git"
ARG AITER_COMMIT="v0.1.2"
ARG AITER_COMMIT="v0.1.3"

RUN git clone ${SGL_REPO} \
&& cd sglang \
Expand Down
3 changes: 2 additions & 1 deletion python/sglang/srt/layers/moe/fused_moe_triton/layer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,6 +32,7 @@

if _use_aiter:
from aiter import ActivationType
from aiter.fused_moe import fused_moe
from aiter.fused_moe_bf16_asm import ck_moe_2stages
from aiter.ops.shuffle import shuffle_weight

Expand Down Expand Up @@ -204,7 +205,7 @@ def forward_cuda(
topk_weights, dtype=torch.float32
) # topk_weights must be FP32 (float32)

return ck_moe_2stages(
return fused_moe(
x,
layer.w13_weight,
layer.w2_weight,
Expand Down
16 changes: 8 additions & 8 deletions python/sglang/srt/layers/quantization/fp8.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1052,15 +1052,15 @@ def maybe_apply_hip_fused_experts(
if _use_hip_int4:
# TODO: add triton kernel and add check _use_aiter
assert not no_combine, f"{no_combine=} is not supported."
return ck_moe_2stages(
return fused_moe(
x,
layer.w13_weight,
layer.w2_weight,
topk_weights,
topk_ids,
QuantType.per_Token,
layer.w13_weight_scale1,
layer.w2_weight_scale1,
quant_type=QuantType.per_Token,
w1_scale=layer.w13_weight_scale1,
w2_scale=layer.w2_weight_scale1,
activation=(
ActivationType.Silu if activation == "silu" else ActivationType.Gelu
),
Expand All @@ -1086,15 +1086,15 @@ def maybe_apply_hip_fused_experts(
expert_mask=None,
)
else:
return ck_moe_2stages(
return fused_moe(
x,
layer.w13_weight,
layer.w2_weight,
topk_weights,
topk_ids,
QuantType.per_Token,
layer.w13_weight_scale1,
layer.w2_weight_scale1,
quant_type=QuantType.per_Token,
w1_scale=layer.w13_weight_scale1,
w2_scale=layer.w2_weight_scale1,
activation=(
ActivationType.Silu
if activation == "silu"
Expand Down
Loading