Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
OpenDAS
vllm_cscc
Commits
ce9dc02c
Unverified
Commit
ce9dc02c
authored
Jun 13, 2025
by
Wentao Ye
Committed by
GitHub
Jun 13, 2025
Browse files
[Refactor] Remove unused variables in `moe_permute_unpermute_kernel.inl` (#19573)
Signed-off-by:
yewentao256
<
zhyanwentao@126.com
>
parent
a24cb916
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
0 additions
and
5 deletions
+0
-5
csrc/moe/permute_unpermute_kernels/moe_permute_unpermute_kernel.inl
...ermute_unpermute_kernels/moe_permute_unpermute_kernel.inl
+0
-5
No files found.
csrc/moe/permute_unpermute_kernels/moe_permute_unpermute_kernel.inl
View file @
ce9dc02c
...
...
@@ -20,7 +20,6 @@ __global__ void expandInputRowsKernel(
int expert_id = sorted_experts[expanded_dest_row];
extern __shared__ int64_t smem_expert_first_token_offset[];
int64_t align_expanded_row_accumulate = 0;
if constexpr (ALIGN_BLOCK_SIZE) {
// load g2s
for (int idx = threadIdx.x; idx < num_local_experts + 1;
...
...
@@ -63,7 +62,6 @@ __global__ void expandInputRowsKernel(
using DataElem = cutlass::Array<T, ELEM_PER_THREAD>;
// Duplicate and permute rows
int64_t const source_k_rank = expanded_source_row / num_rows;
int64_t const source_row = expanded_source_row % num_rows;
auto const* source_row_ptr =
...
...
@@ -160,7 +158,6 @@ __global__ void finalizeMoeRoutingKernel(
elem_index += stride) {
ComputeElem thread_output;
thread_output.fill(0);
float row_rescale{0.f};
for (int k_idx = 0; k_idx < k; ++k_idx) {
int64_t const expanded_original_row = original_row + k_idx * num_rows;
int64_t const expanded_permuted_row =
...
...
@@ -177,8 +174,6 @@ __global__ void finalizeMoeRoutingKernel(
auto const* expanded_permuted_rows_row_ptr =
expanded_permuted_rows_v + expanded_permuted_row * num_elems_in_col;
int64_t const expert_idx = expert_for_source_row[k_offset];
ComputeElem expert_result = arrayConvert<InputElem, ComputeElem>(
expanded_permuted_rows_row_ptr[elem_index]);
thread_output = thread_output + row_scale * (expert_result);
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment