Jetson AGX 16GB with jetpack4.6 [L4T 32.6.1] can not update tensorrt

PROJECT:git@github.com:enazoe/yolo-tensorrt.git
INFER:yolov3 with int8
ERROR:ERROR: 2: [reformatRunner.cpp::onShapeChangeNONCONST::104] Error Code 2: Internal Error (Assertion mCombinedScalesSize >= channelDst failed.)
yolo-trt: /home/xavier/sample_yolo/modules/yolo.cpp:486: void Yolo::createYOLOEngine(nvinfer1::DataType, Int8EntropyCalibrator*): Assertion `m_Engine != nullptr’ failed.

Note:
I think there is some error in tensorrt8.0.1,how can i upgrade tensorrt only?

Dear @huaihuacq,
You can not upgrade tensorrt alone as the TRT libs have dependency issues with other libs like CUDA. Please upgrade to latest Jetpack version and test.

What is the latest Jetpack for AGX 16GB

I want to check if there is a bug in the TENSORRT8.0.1.6

@SivaRamaKrishnaNV
Do you know the inference result of resnet int8 no matter what input, output top5 are:
[1] wall clock
[2] switch
[3] tick
[4] black widow
[5] screw

Dear @huaihuacq,
Please test with latest release - JetPack 5.0.2(JetPack SDK | NVIDIA Developer)

Do you know the inference result of resnet int8 no matter what input, output top5 are

Are you testing the model shipped in TRT samples?

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.