cuda : move neu_padded to where it is used (#29940)
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
This commit is contained in:
1 parent
2e7c58c547
commit
0eb6d9a813
1 file changed
+1
-3
@@ -37,9 +37,6 @@ static __global__ void mm_ids_helper(
|
||||
const int n_expert_used = n_expert_used_template == 0 ? n_expert_used_var : n_expert_used_template;
|
||||
const int expert = blockIdx.x;
|
||||
|
||||
// token slots per warp lane group, padded to a power of 2 so a warp divides evenly
|
||||
constexpr int neu_padded = mm_ids_pow2<n_expert_used_template>::value;
|
||||
|
||||
extern __shared__ char data_mm_ids_helper[];
|
||||
mm_ids_helper_store * store = (mm_ids_helper_store *) data_mm_ids_helper;
|
||||
|
||||
@@ -69,6 +66,7 @@ static __global__ void mm_ids_helper(
|
||||
} else {
|
||||
// Implementation optimized for specific numbers of experts used:
|
||||
// a warp holds a whole number of token slots, so the slot count is padded to a power of 2
|
||||
constexpr int neu_padded = mm_ids_pow2<n_expert_used_template>::value;
|
||||
static_assert(neu_padded <= warp_size && warp_size % neu_padded == 0, "bad n_expert_used");
|
||||
for (int it0 = 0; it0 < n_tokens; it0 += warp_size/neu_padded) {
|
||||
const int it = it0 + threadIdx.x / neu_padded;
|
||||
|
||||
Reference in new issue
Block a user