added MLP with proper documentation (wang-xinyu#874)

* [MLP]: create using core TensortRT python APIs

* [MLP]: README.md added with chart

* [MLP]: C++ TensorRT APIs added

* [MLP]: Updated Docs

* [MLP]: basic .wts added with APIs

* [MLP]: Updated minor details

* [MLP]: fix

Author: sbmalik

README.md

@@ -53,6 +53,7 @@ Following models are implemented.
 
 |Name | Description |
 |-|-|
+|[mlp](./mlp) | the very basic model for starters, properly documented |
 |[lenet](./lenet) | the simplest, as a "hello world" of this project |
 |[alexnet](./alexnet)| easy to implement, all layers are supported in tensorrt |
 |[googlenet](./googlenet)| GoogLeNet (Inception v1) |

mlp/CMakeLists.txt

@@ -0,0 +1,24 @@
+cmake_minimum_required(VERSION 3.14)  # change the version, if asked by compiler
+project(mlp)
+
+set(CMAKE_CXX_STANDARD 14)
+
+# include and link dirs of tensorrt, you need adapt them if yours are different
+include_directories(/usr/include/x86_64-linux-gnu/)
+link_directories(/usr/lib/x86_64-linux-gnu/)
+
+# include and link dirs of cuda for inference
+include_directories(/usr/local/cuda/include)
+link_directories(/usr/local/cuda/lib64)
+
+# create link for executable files
+add_executable(mlp mlp.cpp)
+
+# perform linking with nvinfer libraries
+target_link_libraries(mlp nvinfer)
+
+# link with cuda libraries for Inference
+target_link_libraries(mlp cudart)
+
+add_definitions(-O2 -pthread)
+

mlp/README.md

@@ -0,0 +1,57 @@
+# MLP
+
+MLP is the most basic net in this tensorrtx project for starters. You can learn the basic procedures of building
+TensorRT app from the provided APIs. The process of building a TensorRT engine explained in the chart below.
+
+![TensorRT Image](https://user-images.githubusercontent.com/33795294/148565279-795b12da-5243-4e7e-881b-263eb7658683.jpg)
+
+## Helper Files
+
+`logging.h` : A logger file for using NVIDIA TRT API (mostly same for all models)
+
+`mlp.wts` : Converted weight file (simple file, you can open and check it)
+
+## TensorRT C++ API
+
+```
+// 1. generate mlp.wts from https://github.com/wang-xinyu/pytorchx/tree/master/mlp -- or use the given .wts file
+
+// 2. put mlp.wts into tensorrtx/mlp (if using the generated weights)
+
+// 3. build and run
+
+    cd tensorrtx/mlp
+
+    mkdir build
+
+    cd build
+
+    cmake ..
+
+    make
+
+    sudo ./mlp -s   // serialize model to plan file i.e. 'mlp.engine'
+
+    sudo ./mlp -d   // deserialize plan file and run inference
+```
+
+## TensorRT Python API
+
+```
+# 1. Generate mlp.wts from https://github.com/wang-xinyu/pytorchx/tree/master/mlp -- or use the given .wts file
+
+# 2. Put mlp.wts into tensorrtx/mlp (if using the generated weights)
+
+# 3. Install Python dependencies (tensorrt/pycuda/numpy)
+
+# 4. Run 
+    
+    cd tensorrtx/mlp
+    
+    python mlp.py -s   # serialize model to plan file, i.e. 'mlp.engine'
+    
+    python mlp.py -d   # deserialize plan file and run inference
+```
+
+## Note
+It also supports the latest CUDA-11.4 and TensorRT-8.2.x