Skip to content

[Bugfix] Correct block shape logic in WNA16 MoE triton kernel #31686

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

Closed
JartX wants to merge 2 commits into from
Closed

[Bugfix] Correct block shape logic in WNA16 MoE triton kernel #31686

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
21a986b
[Bugfix] Correct block shape logic in WNA16 MoE triton kernel
JartX Jan 4, 2026
5de77ba
Merge branch 'main' into bugfix/correct_block_shape_logic_WNA16MOE
JartX Jan 5, 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
44 changes: 25 additions & 19 deletions vllm/model_executor/layers/fused_moe/fused_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -629,28 +629,32 @@ def invoke_fused_moe_wna16_triton_kernel(
EM = sorted_token_ids.size(0)
if A.size(0) < config["BLOCK_SIZE_M"]:
# optimize for small batch_size.
# We assume that top_ids of each token is unique,
# so num_valid_experts <= batch_size <= BLOCK_SIZE_M,
# and we can skip some invalid blocks.
EM = min(sorted_token_ids.size(0), A.size(0) * top_k * config["BLOCK_SIZE_M"])

config = config.copy()

block_config = get_moe_wna16_block_config(
config=config,
use_moe_wna16_cuda=False,
num_valid_tokens=num_tokens,
size_k=A.size(1),
size_n=B.size(1),
num_experts=B.size(1),
group_size=block_shape[1],
real_top_k=top_k,
block_size_m=config["BLOCK_SIZE_M"],
)
config.update(block_config)

if "BLOCK_SIZE_N" not in config:
config["BLOCK_SIZE_N"] = 64
if "BLOCK_SIZE_K" not in config:
config["BLOCK_SIZE_K"] = 32

grid = lambda META: (
triton.cdiv(EM, META["BLOCK_SIZE_M"])
* triton.cdiv(B.size(1), META["BLOCK_SIZE_N"]),
)
config = config.copy()
config.update(
get_moe_wna16_block_config(
config=config,
use_moe_wna16_cuda=False,
num_valid_tokens=num_tokens,
size_k=A.size(1),
size_n=B.size(1),
num_experts=B.size(1),
group_size=block_shape[1],
real_top_k=top_k,
block_size_m=config["BLOCK_SIZE_M"],
)
)

fused_moe_kernel_gptq_awq[grid](
A,
Expand Down Expand Up @@ -2373,12 +2377,13 @@ def apply(
topk_ids, config["BLOCK_SIZE_M"], global_num_experts, expert_map
)

invoke_fused_moe_triton_kernel(
dispatch_fused_moe_kernel(
hidden_states,
w1,
intermediate_cache1,
a1q_scale,
self.w1_scale,
getattr(self, "w1_zp", None),
None, # topk_weights
sorted_token_ids,
expert_ids,
Expand Down Expand Up @@ -2410,12 +2415,13 @@ def apply(
self.block_shape,
)

invoke_fused_moe_triton_kernel(
dispatch_fused_moe_kernel(
qintermediate_cache2,
w2,
intermediate_cache3,
a2q_scale,
self.w2_scale,
getattr(self, "w2_zp", None),
topk_weights,
sorted_token_ids,
expert_ids,
Expand Down