TensorRT Inference Engine

credits: CIFAR10 reader copied from https://github.com/prabhuomkar/pytorch-cpp/blob/master/tutorials/intermediate/deep_residual_network/src/main.cpp

NOTE: Please download and unzip the source-code of the libtorch library in lib folder before compiling

The following are the steps to build this project
All the commands below are to be executed in directoryinference/build-dir

Build

open inference/build-dir in terminal and type the following commands

 cmake ..

 make

You will get the following output

[ 33%] Building CXX object CMakeFiles/archit_inf.dir/src/main.cpp.o
[ 66%] Building CXX object CMakeFiles/archit_inf.dir/src/cifar10.cpp.o
[100%] Linking CXX executable archit_inf
[100%] Built target archit_inf

The above steps will build the executable archit_inf. (archit is my nickname :- ) )

Usage

To do the latency test type

./archit_inf batch_size # where batch_size can be between 1 and 512

for example

./archit_inf 32 #for batch size of 32

NOTE:

If no argument is given batch size is taken as 1
Accuracy is computed and printed only when batch size is 1

Contributing

Madhav Wagle
gmail: architwagle@gmail.com
TSEC, Mumbai University