Skip to content

[Kernel][Performance] Tweak MoE Batched silu_mul_fp8_quant_deep_gemm kernel #21193

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 1 commit into from
Jul 19, 2025
Merged

[Kernel][Performance] Tweak MoE Batched silu_mul_fp8_quant_deep_gemm kernel #21193

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
9 changes: 4 additions & 5 deletions vllm/model_executor/layers/fused_moe/batched_deep_gemm_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,6 +55,7 @@ def _silu_mul_fp8_quant_deep_gemm(

# Meta ---------------------------------------------------------------
BLOCK: tl.constexpr,
NUM_STAGES: tl.constexpr,
):
G = H // GROUP_SIZE

Expand All @@ -73,8 +74,7 @@ def _silu_mul_fp8_quant_deep_gemm(
cols = cols.to(tl.int64)
mask_h = cols < BLOCK

t = tl.zeros([], tl.int64)
while t < n_tokens:
for t in tl.range(0, n_tokens, num_stages=NUM_STAGES):
base_i_offset = (e * stride_i_e + t * stride_i_t +
g * GROUP_SIZE * stride_i_h)
base_yq_offset = (e * stride_yq_e + t * stride_yq_t +
Expand Down Expand Up @@ -102,8 +102,6 @@ def _silu_mul_fp8_quant_deep_gemm(
tl.store(y_q_ptr + base_yq_offset + cols * stride_yq_h, y_q, mask=mask)
tl.store(y_s_ptr + base_ys_offset, y_s)

t += 1


def silu_mul_fp8_quant_deep_gemm(
y: torch.Tensor, # (E, T, 2*H) float32
Expand Down Expand Up @@ -180,7 +178,8 @@ def silu_mul_fp8_quant_deep_gemm(
fp8_max,
is_blackwell_deep_gemm_used(),
BLOCK=group_size,
num_warps=4,
NUM_STAGES=8,
num_warps=1,
)

return y_q, y_s
Expand Down