Skip to content

[Bugfix] Deepseek v3 fix max_num_batched_tokens #386

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

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

[Bugfix] Deepseek v3 fix max_num_batched_tokens #386

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
40 changes: 19 additions & 21 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 @@ -33,26 +33,26 @@ def fused_moe_kernel(
expert_ids_ptr,
num_tokens_post_padded_ptr,
# Matrix dimensions
N,
K,
EM,
num_valid_tokens,
N: tl.int64,
K: tl.int64,
EM: tl.int64,
num_valid_tokens: tl.int64,
# The stride variables represent how much to increase the ptr by when
# moving by 1 element in a particular dimension. E.g. `stride_am` is
# how much to increase `a_ptr` by to get the element one row down
# (A has M rows).
stride_am,
stride_ak,
stride_be,
stride_bk,
stride_bn,
stride_cm,
stride_cn,
stride_asm,
stride_ask,
stride_bse,
stride_bsk,
stride_bsn,
stride_am: tl.int64,
stride_ak: tl.int64,
stride_be: tl.int64,
stride_bk: tl.int64,
stride_bn: tl.int64,
stride_cm: tl.int64,
stride_cn: tl.int64,
stride_asm: tl.int64,
stride_ask: tl.int64,
stride_bse: tl.int64,
stride_bsk: tl.int64,
stride_bsn: tl.int64,
# Block size for block-wise quantization
group_n: tl.constexpr,
group_k: tl.constexpr,
Expand Down Expand Up @@ -114,18 +114,16 @@ def fused_moe_kernel(
num_tokens_post_padded = tl.load(num_tokens_post_padded_ptr)
if pid_m * BLOCK_SIZE_M >= num_tokens_post_padded:
return
offs_token_id = pid_m * BLOCK_SIZE_M + tl.arange(0, BLOCK_SIZE_M).to(
tl.int64)
offs_token_id = pid_m * BLOCK_SIZE_M + tl.arange(0, BLOCK_SIZE_M)
offs_token = tl.load(sorted_token_ids_ptr + offs_token_id)
token_mask = offs_token < num_valid_tokens

offs_bn = (pid_n * BLOCK_SIZE_N +
tl.arange(0, BLOCK_SIZE_N).to(tl.int64)) % N
offs_bn = (pid_n * BLOCK_SIZE_N + tl.arange(0, BLOCK_SIZE_N)) % N
offs_k = tl.arange(0, BLOCK_SIZE_K)
a_ptrs = a_ptr + (offs_token[:, None] // top_k * stride_am +
offs_k[None, :] * stride_ak)

off_experts = tl.load(expert_ids_ptr + pid_m).to(tl.int64)
off_experts = tl.load(expert_ids_ptr + pid_m)
b_ptrs = b_ptr + off_experts * stride_be + (offs_k[:, None] * stride_bk +
offs_bn[None, :] * stride_bn)
if use_int8_w8a16:
Expand Down
Loading