段错误是什么情况？

[CGump]

很奇怪，配置和模型都是ok的，运行的时候报段错误是为啥？

yolov5_trt_create  ... 
yolov5_trt_create  cuda engine... 
[02/24/2022-16:18:59] [W] [TRT] Current optimization profile is: 0. Please ensure there are no enqueued operations pending in this context prior to switching profiles
yolov5_trt_create  buffer ... 
yolov5_trt_create  stream ... 
yolov5_trt_create  done ... 
create yolov5-trt , instance = 0x555e4d644810
[02/24/2022-16:18:59] [W] [TRT] Current optimization profile is: 0. Please ensure there are no enqueued operations pending in this context prior to switching profiles
[02/24/2022-16:18:59] [E] [TRT] INVALID_ARGUMENT: Cannot find binding of given name: input
[02/24/2022-16:18:59] [E] [TRT] INVALID_ARGUMENT: Cannot find binding of given name: output
delay_proress:3ms, delay_infer:3ms
段错误 (核心已转储)

[CGump]

定位问题可能出在cudaMemcpyAsync，目前还不是很了解这块，是输入的两个参数问题吗？

void FeatureTensor::doInference(float* inputBuffer, float* outputBuffer) {   
    cudaMemcpyAsync(buffers[inputIndex], inputBuffer, inputStreamSize * sizeof(float), cudaMemcpyHostToDevice, cudaStream);
    Dims4 inputDims{curBatchSize, 3, imgShape.height, imgShape.width};
    context->setBindingDimensions(0, inputDims);
    
    context->enqueueV2(buffers, cudaStream, nullptr);
    cudaMemcpyAsync(outputBuffer, buffers[outputIndex], outputStreamSize * sizeof(float), cudaMemcpyDeviceToHost, cudaStream);
    // cudaStreamSynchronize(cudaStream);
}

[CGump]

运行环境：rtx2080ti，ubuntu18.04，tensorrt 7.0-cuda10.2-cudnn7.6.5

[CGump]

deepsort的模型和yolov5的模型各自都是正常运行的。然后，我也查看了第一帧yolov5的结果，vector中存放了20个，yolo模型应该是正常识别的，问题就是DS这一步，这就很迷惑了……

[CGump]

在另一台tensorrt7.2.3+cuda11.1+cudnn8.1上搞定了，这个问题还是没解决。难道是cuda10和11的差异导致？

[knightdby]

@CGump 所以问题解决了吗？