Skip to content

GitLab

Commit a75021db authored by zhuwenwen's avatar zhuwenwen
Browse files

Merge branch 'v0.7.2-dev' of http://10.6.10.68/dcutoolkit/deeplearing/vllm into v0.7.2-dev

parents a8b2d878 a9c16183
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
csrc/moe/moe_align_sum_kernels.cu
View file @ a75021db
...@@ -307,7 +307,7 @@ __global__ void ep_moe_align_block_size_kernel(scalar_t* __restrict__ topk_ids, ...@@ -307,7 +307,7 @@ __global__ void ep_moe_align_block_size_kernel(scalar_t* __restrict__ topk_ids,
* assigned to expert expert_index. * assigned to expert expert_index.
*/ */
for (int i = start_idx; i < numel && i < start_idx + tokens_per_thread; ++i) { for (int i = start_idx; i < numel && i < start_idx + tokens_per_thread; ++i) {
expert_id = topk_ids[i]; int32_t expert_id = topk_ids[i];
if (expert_id >= start_expert && expert_id < end_expert) { if (expert_id >= start_expert && expert_id < end_expert) {
++tokens_cnts[index(num_experts, threadIdx.x + 1, expert_id - start_expert)]; ++tokens_cnts[index(num_experts, threadIdx.x + 1, expert_id - start_expert)];
} }
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment