Skip to content
This repository has been archived by the owner on Nov 17, 2023. It is now read-only.

Commit

Permalink
[1.x] Enable CUDA Graphs for TRT (#19184)
Browse files Browse the repository at this point in the history
Signed-off-by: Serge Panev <[email protected]>
  • Loading branch information
Kh4L authored Sep 19, 2020
1 parent 0496690 commit a35d568
Showing 1 changed file with 5 additions and 0 deletions.
5 changes: 5 additions & 0 deletions src/operator/subgraph/tensorrt/tensorrt.cc
Original file line number Diff line number Diff line change
Expand Up @@ -370,6 +370,11 @@ NNVM_REGISTER_OP(_TensorRT)
.set_attr<nnvm::FListInputNames>("FListInputNames", TRTListInputNames)
.set_attr<nnvm::FListOutputNames>("FListOutputNames", DefaultSubgraphOpListOutputs)
.set_attr<FCreateOpState>("FCreateOpState", TRTCreateState)
.set_attr<FIsCUDAGraphsCompatible>("FIsCUDAGraphsCompatible",
[](const NodeAttrs& attrs, const bool) {
const TRTParam& param = nnvm::get<TRTParam>(attrs.parsed);
return !param.int8_mode;
})
.set_attr<FInferStorageType>("FInferStorageType", TRTInferStorageType);

MXNET_REGISTER_SUBGRAPH_BACKEND(TensorRT);
Expand Down

0 comments on commit a35d568

Please sign in to comment.