jetson_dla_tutorial
jetson_dla_tutorial copied to clipboard
NVIDIA-AI-IOT
A tutorial for getting started with the Deep Learning Accelerator (DLA) on NVIDIA Jetson
* Use `nsys` to profile the engine file , but I generate the file `model_gn.nvvp.nsys-rep` ``` nsys profile --trace=cuda,nvtx,cublas,cudla,cusparse,cudnn,nvmedia --force-overwrite=true --output=model_gn.nvvp /usr/src/tensorrt/bin/trtexec --loadEngine=yolov5_model.loadable --iterations=10 --inputIOFormats=int8:dla_hwc4 --outputIOFormats=fp16:chw16 --int8 --fp16 --useDLACore=0 --idleTime=500...
trtexec --onnx=truck.onnx --saveEngine=truck-detect.engine --exportProfile=model_gn.json --fp16 --useDLACore=0 --allowGPUFallback --useSpinWait --separateProfileRun > model_gn.log [W] [TRT] Layer '/model.24/Constant_1_output_0' (CONSTANT): Unsupported on DLA. Switching this layer's device type to GPU. [W] [TRT] Layer '(Unnamed...
Metadata
Owner
Metadata
A tutorial for getting started with the Deep Learning Accelerator (DLA) on NVIDIA Jetson