nvinfer1::ICudaEngine failure of TensorRT 10.4 when running Ubuntu 22.04 on GPU A2000

[ahmet-f-gumustas]

Environment

TensorRT Version: 10.4

NVIDIA GPU: A2000

NVIDIA Driver Version: 560

CUDA Version: Cuda 12.6

CUDNN Version: 9.4

Operating System: Ubuntu 22.04

Python Version (if applicable):

Tensorflow Version (if applicable):

PyTorch Version (if applicable):

Baremetal or Container (if so, version):

This is my Code ;

for (int i = 0; i < engine->getNbBindings(); i++) {  // getNbBindings() ile binding sayısı alınıyor
    if (engine->bindingIsInput(i)) {  // bindingIsInput ile input olup olmadığı kontrol ediliyor
        inputIndex = i;
        std::cout << "Found input binding at index " << i << std::endl;
    } else {
        outputIndex = i;
        std::cout << "Found output binding at index " << i << std::endl;
    }
}

if (inputIndex == -1 || outputIndex == -1) {
    std::cerr << "Failed to find input or output bindings!" << std::endl;
    return -1;
}

// Get the dimensions of input/output
nvinfer1::Dims inputDims = engine->getBindingDimensions(inputIndex);
nvinfer1::Dims outputDims = engine->getBindingDimensions(outputIndex);

And gave me an vs code error : "class "nvinfer1::ICudaEngine" has no member "getNbBindings"C/C++(135)"

Terminal Output:

ahmet@ahmet-Precision-3660:~/workspace-example/tensorrt$ g++ 02-running-engine-file.cpp -o engine-result -I/usr/local/cuda/include -L/usr/local/cuda/lib64 -lnvinfer -lcudart -std=c++11
02-running-engine-file.cpp: In function ‘int main()’:
02-running-engine-file.cpp:60:33: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getNbBindings’
60 | for (int i = 0; i < engine->getNbBindings(); i++) { // getNbBindings() ile binding sayısı alınıyor
| ^~~~~~~~~~~~~
02-running-engine-file.cpp:61:21: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘bindingIsInput’
61 | if (engine->bindingIsInput(i)) { // bindingIsInput ile input olup olmadığı kontrol ediliyor
| ^~~~~~~~~~~~~~
02-running-engine-file.cpp:76:40: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getBindingDimensions’
76 | nvinfer1::Dims inputDims = engine->getBindingDimensions(inputIndex);
| ^~~~~~~~~~~~~~~~~~~~
02-running-engine-file.cpp:77:41: error: ‘class nvinfer1::ICudaEngine’ has no member named ‘getBindingDimensions’
77 | nvinfer1::Dims outputDims = engine->getBindingDimensions(outputIndex);
| ^~~~~~~~~~~~~~~~~~~~

How to solve this :(

[lix19937]

And gave me an vs code error : "class "nvinfer1::ICudaEngine" has no member "getNbBindings"C/C++(135)"

Please use ICudaEngine::getNbIOTensors(). @ahmet-f-gumustas

[ahmet-f-gumustas]

Thank You I solved .

[mehmetkurt20]

What can be used for getBindingIndex and getBindingDimensions, same errors exist with you have when I try to use those from ICudaEngine class.

[santiweide]

Maybe try Dims getDimensions(char const* inputName, OptProfileSelector select) const noexcept?