apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-10 02:50:19 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

[feature]2.2 custom_allreduce support cudagraph recapture (#4307)

Browse Source 
* custom_allreduce support cudagraph recapture

* delete code

* add shut_down/restart default group
This commit is contained in:
chen
2025-09-29 18:14:21 +08:00
committed by GitHub
parent 9cc2c99539
commit 63a03ee152
7 changed files with 31 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
custom_ops/gpu_ops/cpp_extensions.cc
View File

@@ -616,6 +616,8 @@ int64_t open_mem_handle(paddle::Tensor& mem_handle);
void free_shared_buffer(int64_t buffer);
void clear_ipc_handles(int64_t _fa);
// speculative decoding Kernel
std::vector<paddle::Tensor> SpeculateGetPaddingOffset(
const paddle::Tensor& input_ids,
@@ -1204,6 +1206,8 @@ PYBIND11_MODULE(fastdeploy_ops, m) {
m.def("free_shared_buffer", &free_shared_buffer, "free_shared_buffer");
m.def("clear_ipc_handles", &clear_ipc_handles, "clear_ipc_handles");
m.def("open_mem_handle", &open_mem_handle, "open_mem_handle");
m.def("get_graph_buffer_ipc_meta", &get_graph_buffer_ipc_meta, "get_graph_buffer_ipc_meta");