
This SDK has been patched by Embedded Artists for the iMXRT1176 Developer's Kit.
The SDK was released on 2022-03-02 and is based on NXP's 2.11.0 SDK (SDK_2_11_0_MIMXRT1176xxxxx.zip).

This is what has been patched:
* LWIP projects - added reading of the MAC address from I2C EEPROM either on the 100Mbit adapter
  or on the uCOM board
* Added an I2C driver for the gpio expander (PCA6416) and code to use it
* Added an I2C driver for the PWM gpio expander (PCA9530) and code to use it
* SDRAM size has been corrected to 32Mb (including linker files, MPU and DCD)
* 1G Ethernet PHY has been changed from RTL8211F to AR8031DS
* SEMC projects - changed to correct settings for the SDRAM
* SEMC projects - changed algorithm for memory test and now test entire 32MB instead of only 4KB
* Examples using eLCDIF/LCDIFv2 have been updated to use PCA6416/PCA9530 for
  RST/PWR/BL signals
* BOARD_USER_BUTTON has been redirected to SW5/WAKEUP button on the uCOM Carrier Board
* USER_LED has been changed to the blue RGB LED using PCA6416
* Touch: I2C bus and GPIOs have been changed for RST/INT
* Camera pins
* Adjusted the USB interface number (it is different for host and device examples)
* Changed the Wi-Fi examples to use the Embedded Artists 1XK M.2 Module (EAR00385) as default
* Corrected the ethernet PHY addresses
* Changed CORE clock depending on speed grading of MCU (798MHz for Industrial, 996MHz for Commercial)
* Changed SEMC clock to be within maximum speed for SDRAM (now 148.5MHz, was 198MHz)
* Many of the projects have been updated to use a more complete pin_mux.c file where all
  necessary pins have been initialized. The SDK examples used to only configure the pins
  that they use (and often not every pin) and most of the time the configuration was only
  for MUX:ing and not the PAD settings (pull up/down/none, drive strength and slew).
* Converted the AzureRTOS examples to use the BOARD_NETWORK_USE_100M_ENET_PORT (same as all
  other networking examples) instead of using the old EXAMPLE_USE_1G_ENET_PORT

This has been added:
* HDMI support to most GUI examples. HDMI at 1024x768@60 is the default resolution but
  that can be changed per project in display_support.h/elcdif_support.h/lcdifv2_support.h.
* Added ADT example for TensorFlow Lite
* I2C probe example
* EDID reader example
* Wi-Fi (serial) examples for the CMWC1ZZABR-107-EVB (a.k.a ABR Module)

This has been removed:
* All projects for the expansion board AGM01

Important things to note:
* Read section "8 - Known Issues" in docs/MCUXpresso SDK Release Notes for MIMXRT1170-EVK.pdf
  to see known issues with the current version of the SDK.
* For Iperf examples, set compiler optimization to -O3 or similar to improve performance.
* If the hardware seems unresponsive and the debugger cannot connect/flash/erase the current program
  then the most likely cause is the running program preventing the access. To stop the currently
  running program and regain control:
  1) Press and hold down the ISP_ENABLE button (SW1)
  2) Press and hold down the RESET button (SW3)
  3) Let go of the RESET button
  4) Wait an extra second or two
  5) Release the ISP_ENABLE button
  6) The hardware is now in a mode where programming/erasing it should work


Connectors:
* J29 (micro USB) is the default UART for the CM7 core and unless specified otherwise it is setup for 115200 8/N/1
* J30 (micro USB) is the default UART for the CM4 core and unless specified otherwise it is setup for 115200 8/N/1
* For 1Gbit Ethernet examples, use connector J25 on uCOM Carrier Board
* For 100Mbit Ethernet examples, use ethernet adapter connected between J12 on uCOM Carrier Board
  and J37 on the adapter. These four connections are also needed:
    1) uCOM Carrier Board, JP38:1 -> adapter JP37:1
    2) uCOM Carrier Board, JP38:2 -> adapter JP37:2
    3) uCOM Carrier Board, JP27:1 -> adapter JP39:2
    4) uCOM Carrier Board, JP27:2 -> adapter JP39:1
* The two CSI examples can use either an OV5640 camera in connector J23 or a camera in connector J24
* The EIQ examples that use a camera expects the camera in connector J24 (J23 might work for some
  of the examples but runs much slower)
* Some GUI examples are configured for the RK055AHD091 display which should be in connector "C" on
  the uCOM Carrier Board. Ignore the readme text about connecting extra 5V power.
* The default for GUI examples is to use an HDMI adapter in connector "C" on
  the uCOM Carrier Board. Ignore the readme text about connecting extra 5V power.


Everything below this line is the original content of the readme file.
=======================================================================



Overview
========
This example project provides an inference example using the Lenet model
compiled with the Glow AOT software tools. The model is capable to perform
hand-written digit classification. The model is using 28 x 28 grayscale
input images and provides the confidence scores for the 10 output classes:
digits "0" to "9". The application will run the inference on a sample
image and display the top1 classification results and the inference time.
The project example will walk through all the steps of downloading and
compiling the model, pre-processing a sample image, building and running
the project.

Files:
  main.c  - example source code
  timer.c - implementation of helper functions for measuring inference time
  timer.h - declarations of helper functions for measuring inference time


Toolchains supported
====================
- MCUXpresso IDE
- IAR Embedded Workbench for ARM
- ArmGCC - GNU Tools ARM Embedded

Hardware requirements
=====================
- Mini/micro USB cable
- EVK-IMXRT1170 board
- Personal computer (Windows)
- Python 3.x installed
- Glow AOT tools installed for Windows and added to system PATH variable:
  - image-classifier.exe
  - model-compiler.exe

You can find the Glow Windows Installer on the NXP eIQ mirror:
https://www.nxp.com/design/software/development-software/eiq-ml-development-environment:EIQ

Board settings
==============
No special settings are required.

Prepare the demo
================
0. Create a folder "model" in the root directory of the project with two subfolders:
   "lenet_mnist" and "dataset".

1. Download the Lenet model files (Caffe2 format) using your browser from the
   following links into the "model\lenet_mnist" project directory:

   http://fb-glow-assets.s3.amazonaws.com/models/lenet_mnist/predict_net.pb
   http://fb-glow-assets.s3.amazonaws.com/models/lenet_mnist/init_net.pb

2. Download the Mnist image samples using your browser from the official Glow
   repository into the "model\dataset" project directory. The images contain
   one sample image for each digit from "0" to "9".

   https://github.com/pytorch/glow/blob/master/tests/images/mnist/0_1009.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/1_1008.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/2_1065.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/3_1020.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/4_1059.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/5_1087.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/6_1099.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/7_1055.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/8_1026.png?raw=true
   https://github.com/pytorch/glow/blob/master/tests/images/mnist/9_1088.png?raw=true

3. Perform profiling on the Mnist dataset using the "image-classifier" Glow tool.
   Open a Windows Command Prompt with the current path set within the project 
   directory and run the following command:

image-classifier model\dataset\*.png ^
    -image-mode=0to1 -image-layout=NCHW ^
    -model=model\lenet_mnist -model-input-name=data ^
    -dump-profile=model\profile.yml ^
    -quantization-schema=symmetric_with_power2_scale ^
    -quantization-precision-bias=Int8

4. Compile the model and generate the bundle using the "model-compiler" Glow tool.
   Run the following command:

model-compiler -model=model\lenet_mnist -model-input=data,float,[1,1,28,28] ^
    -backend=CPU -target=arm -mcpu=cortex-m7 -float-abi=hard ^
    -emit-bundle=source -load-profile=model\profile.yml ^
    -quantization-schema=symmetric_with_power2_scale ^
    -quantization-precision-bias=Int8 ^
    -use-cmsis

5. Use the python script "scripts\glow_process_image.py" to transform an input
   image into a C array data and save it as "source\input_image.inc" which is
   included in main.cpp. As an example we will use the sample image "9_1088.png"
   which contains the digit "9":

python scripts\glow_process_image.py ^
    -input-path=model\dataset\9_1088.png ^
    -output-path=source\input_image.inc ^
    -image-mode=0to1 -image-layout=NCHW -image-channel-order=RGB

6. Import the project in IDE and link the Glow bundle by performing the following steps:

   For MCUXpresso IDE:
     - Right click on the project -> "Properties" 
     - Select "C/C++ Build" -> "Settings" 
     - In the "Tool Setting" tab select "MCU C++ Linker" -> "Miscellaneous" 
     - Add the bundle to "Other objects" panel by clicking "Add..." and specify
       the relative path to the bundle in the project: "../source/lenet_mnist.o"

   For IAR Embedded Workbench:
     - Right click on the "source" folder -> Add -> Add Files...
     - Choose file category "Library/Object Files (*.r;*.a;*.lib;*.o)"
     - Select the object file "lenet_mnist.o" located in "source" folder

   For Keil uVision MDK:
     - Right click on the target -> "Options for Target ..."
     - Select "Linker" tab
     - In the "Misc controls" section add to the existing string with a 
       separating space (" ") character the string "../source/lenet_mnist.o"

   For Armgcc:
     - Edit the "CMakeLists.txt" file from the "armgcc" directory by adding the
       following line at the end of the file:
         target_link_libraries(glow_lenet_mnist.elf ${ProjDirPath}/../source/lenet_mnist.o)

7. Build the project.

8. Connect a USB cable between the host PC and the OpenSDA USB port on the target board.

9. Open a serial terminal with the following settings:
   - 115200 baud rate
   - 8 data bits
   - No parity
   - One stop bit
   - No flow control

10. Download the program to the target board.

Running the demo
================
The log below shows the output of the Release version in the terminal window:

Top1 class = 9
Confidence = 0.942
Inference time = 6 (ms)
