marcoslucianops
commented
3 years ago Hi, edit nvinfer1::ICudaEngine *Yolo::createEngine (nvinfer1::IBuilder* builder) function (lines 61-90), in yolo.cpp file, to:
nvinfer1::ICudaEngine *Yolo::createEngine (nvinfer1::IBuilder* builder)
{
assert (builder);
if (m_DeviceType == "kDLA") {
builder->setDefaultDeviceType(nvinfer1::DeviceType::kDLA);
}
std::vector<float> weights = loadWeights(m_WtsFilePath, m_NetworkType);
std::vector<nvinfer1::Weights> trtWeights;
nvinfer1::INetworkDefinition *network = builder->createNetwork();
if (parseModel(*network) != NVDSINFER_SUCCESS) {
network->destroy();
return nullptr;
}
// Build the engine
std::cout << "Building the TensorRT Engine" << std::endl;
nvinfer1::ICudaEngine * engine = builder->buildCudaEngine(*network);
if (engine) {
std::cout << "Building complete\n" << std::endl;
} else {
std::cerr << "Building engine failed\n" << std::endl;
}
// destroy
network->destroy();
return engine;
}and add these lines in config_infer_primary.txt file (in [property] section):
enable-dla=1
use-dla-core=0Note: edit these lines according to:
-
enable-dla: Indicates whether to use the DLA engine for inferencing. Boolean: 0 or 1
-
use-dla-core: DLA core to be used. Integer: ≥0
I don't have Xavier board to test. Please tell me if it works.
c7934597
satyajitghana
willosonico
https://forums.developer.nvidia.com/t/how-to-use-dla-in-deepstream-yolov5/161550/25
Hi @marcoslucianops , I used deepstream-yolov4, and I check out the engine. That been build on GPU. I saw the article before. How do I fix following program for building DLA engine?
https://github.com/marcoslucianops/DeepStream-Yolo/blob/470ed82658a5546b55185b3223f8057ecf54cf88/native/nvdsinfer_custom_impl_Yolo/yolo.cpp#L74-L81