apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-14 04:44:00 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

[Serving]support 22.12 (#1974)

Browse Source 
support 22.12
This commit is contained in:
heliqi
2023-05-22 22:27:13 +08:00
committed by GitHub
parent 1ac6e8e614
commit 3e7cb88049
4 changed files with 82 additions and 61 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
serving/src/fastdeploy_runtime.cc
View File

@@ -1145,16 +1145,16 @@ TRITONSERVER_Error* ModelInstanceState::ReadOutputTensors(
size_t total_batch_size, TRITONBACKEND_Request** requests,
const uint32_t request_count,
std::vector<TRITONBACKEND_Response*>* responses) {
// r22.03
// BackendOutputResponder responder(
// requests, request_count, responses,
// model_state_->TritonMemoryManager(), model_state_->MaxBatchSize() > 0,
// model_state_->EnablePinnedOutput(), CudaStream());
// r21.10
// r22.12
BackendOutputResponder responder(
requests, request_count, responses, StateForModel()->MaxBatchSize(),
StateForModel()->TritonMemoryManager(),
StateForModel()->EnablePinnedOutput(), CudaStream());
requests, request_count, responses,
model_state_->TritonMemoryManager(), model_state_->MaxBatchSize() > 0,
model_state_->EnablePinnedOutput(), CudaStream());
// r21.10
// BackendOutputResponder responder(
// requests, request_count, responses, StateForModel()->MaxBatchSize(),
// StateForModel()->TritonMemoryManager(),
// StateForModel()->EnablePinnedOutput(), CudaStream());
// Use to hold string output contents
bool cuda_copy = false;