Skip to content

Commit

Permalink
Update csrc/moe/moe_align_sum_kernels.cu
Browse files Browse the repository at this point in the history
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
  • Loading branch information
ElizaWszola and tlrmchlsmth authored Jan 24, 2025
1 parent 4c6ca52 commit 349d986
Showing 1 changed file with 2 additions and 2 deletions.
4 changes: 2 additions & 2 deletions csrc/moe/moe_align_sum_kernels.cu
Original file line number Diff line number Diff line change
Expand Up @@ -38,8 +38,8 @@ __global__ void moe_align_block_size_kernel(scalar_t* __restrict__ topk_ids,
(cnts_byte_offset + sizeof(int32_t) - 1) / sizeof(int32_t);

extern __shared__ int32_t shared_mem[];
token_cnts_t* tokens_cnts = (token_cnts_t*)
shared_mem; // 2d tensor with shape (blockDim.x + 1, num_experts)
token_cnts_t* tokens_cnts = reinterpret_cast<token_cnts_t*>(
shared_mem); // 2d tensor with shape (blockDim.x + 1, num_experts)
int32_t* cumsum =
shared_mem + aligned_offset; // 1d tensor with shape (num_experts + 1)

Expand Down

0 comments on commit 349d986

Please sign in to comment.