[Refactor] Remove dead code multiple places (#61843)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2026-10-16 15:01:07 +00:00
committed by GitHub
parent 8a6c0449b3
commit 8e25bbca19
10 changed files with 14 additions and 168 deletions
-27
View File
@@ -1059,33 +1059,6 @@ void situ_and_mul(torch::stable::Tensor& out, // [..., d]
});
}
void masked_situ_and_mul(torch::stable::Tensor& out, // [E, T, d]
torch::stable::Tensor& input, // [E, T, 2 * d]
const torch::stable::Tensor& expert_num_tokens,
double beta, double linear_beta) {
int num_experts = input.size(0);
int max_num_tokens = input.size(1);
int d = input.size(2) / 2;
if (num_experts == 0 || max_num_tokens == 0 || d == 0) {
return;
}
constexpr int block_size = 256;
dim3 grid((d + block_size - 1) / block_size, num_experts);
dim3 block(block_size);
const torch::stable::accelerator::DeviceGuard device_guard(
input.get_device_index());
const cudaStream_t stream = get_current_cuda_stream();
VLLM_STABLE_DISPATCH_FLOATING_TYPES(
input.scalar_type(), "masked_situ_and_mul_kernel", [&] {
vllm::masked_situ_and_mul_kernel<scalar_t, true>
<<<grid, block, 0, stream>>>(
out.mutable_data_ptr<scalar_t>(),
input.const_data_ptr<scalar_t>(),
expert_num_tokens.const_data_ptr<int>(), max_num_tokens, d,
(float)beta, (float)linear_beta);
});
}
#define LAUNCH_MASKED_ACT_AND_MUL(KERNEL, ACT_FIRST, HAS_CLAMP, LIMIT, ALPHA, \
BETA) \
vllm::masked_act_and_mul_kernel<scalar_t, KERNEL<scalar_t>, ACT_FIRST, \