TensorRT's nvinfer1::INetworkDefinition::addFullyConnected() does not work as expected for C3D network

[akachen]

But I never saw memory occupation increased very much when I did inference with my this network implemented with TensorRT API. I saw an about 700M more memory was occupied if I forcibly added a call on cudnnxxx() in the inference code of my this network, just like video-caffe does as it calls cudnnxxx() in its convolution layers.
I observed the memory increasement by jtop or by adding code suggested here :

CUDA memory release

Hi, It varies from the cuDNN/TensorRT version. You can test it with this sample directly: topic_1055977.cu #include <stdio.h> #include "cuda.h" #include "cudnn.h" #define ONE_MBYTE (1024*1024) void printMemInfo() { size_t free_byte ; size_t total_byte ; cudaError_t cuda_status = cudaMemGetInfo( &free_byte, &total_byte ) ; if ( cudaSuccess != cuda_status ){ printf("Error: cudaMemGetInfo fails, %s\n", cudaGetErrorString(cuda_status)); exit(1); } double…

CUDA memory release

Hi, Please check this sample: #include <cuda_runtime.h> #include <iostream> #include <sstream> #include <fstream> #include <opencv2/core/core.hpp> #include <opencv2/highgui/highgui.hpp> #include <opencv2/imgproc/imgproc.hpp> #include "NvInfer.h" #include "NvCaffeParser.h" #include <cublas_v2.h> #include <cudnn.h> #define INPUT_BLOB "data" #define OUTPUT_BLOB "prob" #define ONE_MBYTE (1024*1024) class Logger : public nvinfer1::ILogger { void log(Severity severity, const char* msg) overr…