DeepSeek/DeepEP
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepEP synced 2025-06-26 18:28:11 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Code cleanup and bug fixed

Browse Source
This commit is contained in:
Chenggang Zhao
2025-05-23 11:14:16 +08:00
parent 68ae8b3d07
commit 92405ddf30
5 changed files with 45 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
csrc/deep_ep.cpp
View File

@@ -1190,8 +1190,9 @@ Buffer::low_latency_combine(const torch::Tensor& x, const torch::Tensor& topk_id
}
torch::Tensor
Buffer::get_next_low_latency_combine_buffer(int num_max_dispatch_tokens_per_rank, int hidden, int num_experts) {
Buffer::get_next_low_latency_combine_buffer(int num_max_dispatch_tokens_per_rank, int hidden, int num_experts) const {
LowLatencyLayout layout(rdma_buffer_ptr, num_max_dispatch_tokens_per_rank, hidden, num_ranks, num_experts);
auto buffer = layout.buffers[low_latency_buffer_idx];
auto dtype = torch::kBFloat16;
auto num_msg_elems = static_cast<int>(buffer.num_bytes_per_combine_msg / elementSize(torch::kBFloat16));