
This SDK has been patched by Embedded Artists for the iMXRT1064 Developer's Kit.
The SDK was released on 2022-03-15 and is based on NXP's 2.11.0 SDK (SDK_2_11_0_MIMXRT1064xxxxA.zip).

This is what has been patched:
* Set CPU speed according to Commercial/Industrial CPU
* Correction of the VDD_SOC_IN voltage.
* LWIP projects - added reading of the MAC address from the onboard I2C EEPROM
* 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
* SEMC projects - changed algorithm for memory test and now test entire 32MB instead of only 4KB
* Examples using a display 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
* 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
* Added support for Embedded Artists 1ZM M.2 Module (EAR00364) in the NXP Wi-Fi examples
* 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).
* Changed reset pin for SD card examples

This has been added:
* LWIP projects - option to use 100/10Mbps Ethernet-PHY Adapter
* AWS projects - option to use 100/10Mbps Ethernet-PHY Adapter
* AzureRTOS projects - option to use 100/10Mbps Ethernet-PHY Adapter
* I2C probe 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 EVK-MIMXRT1064.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 and unless specified otherwise it is setup for 115200 8/N/1


This example has been extended to support two different network options:

Option1: (default) uses the 100/10Mbit Ethernet PHY on the iMX RT1064 uCOM board
         and the connector on the uCOM Carrier Board
Option2: uses the 100/10Mbit Ethernet-PHY Adapter board

For examples that support both options, select which option to use by changing
this define in board.h:

#define BOARD_NETWORK_USE_ONBOARD_100M_ENET_PORT (1U)

Some of the network examples have been modified to obtain the globally unique
MAC address from an EEPROM either on the 100/10Mbit Ethernet-PHY Adapter board
or on the iMX RT1176 uCOM board depending on which interface is used.


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



eIQ inference with TFLiteMicro 2.6.0

Content
-------
1. Introduction
2. Directory structure
3. eIQ inference with TFLite Example
4. Documentation
5. Release notes
6. ModelRunner for TFLite Example

1. Introduction
---------------
DeepView RT ModelRunner for TFLiteMirco is similar service as ModelRunner for TFLite, but it can only stream 
test data to modelrunner service with TFLite backend, it can return Image inferrence result after receive
test data. 
ModelRunner for TFLite can work with eIQ modeltool to profiling model.

2. Directory structure
--------------------------------------
<MCUXpresso-SDK-root>
|-- boards
|   -- <board>
|      -- eiq_examples                         - Example build projects
|         -- deepviewrt_modelrunner-tflite       - deepviewrt example
|
|-- middleware
    -- eiq
       -- tensorflow-lite          - TFLiteMirco 2.6.0

3. eIQ Inference with DeepView RT example
-----------------------------------------
3.1 Introduction
    The package contains modelrunner example applications using
    the TFLiteMirco library. The build projects can be found in
    the /boards/<board>/eiq_examples/deepviewrt* folders.

3.2 Toolchains supported
    - MCUXpresso IDE
    - IAR Embedded Workbench for ARM
    - Keil uVision MDK
    - ArmGCC - GNU Tools ARM Embedded

3.3 Supported board and settings
    - evkmimxrt1170:
      No special settings are required.
    - evkbimxrt1050:
      No special settings are required.
    - evkmimxrt1060:
      No special settings are required.
    - evkbimxrt1064:
      No special settings are required.
    - evkmimxrt1160:
      No special settings are required.

4. Documentation
----------------
    https://www.nxp.com/design/software/development-software/eiq-ml-development-environment:EIQ

5. Release notes
----------------
The library is based on TFLiteMicro 2.6.0.

6. ModelRunner for TFLite Example
---------------------------------
6.1 Running the demo

The ModelRunner for TFLite will run as one http service and can get data input. 
It dump the 

*************************************************
               TFLite Modelrunner
*************************************************
Initializing PHY...
 DHCP state       : SELECTING
 DHCP state       : REQUESTING
 DHCP state       : BOUND

 IPv4 Address     : 10.193.20.16
 IPv4 Subnet mask : 255.255.255.0
 IPv4 Gateway     : 10.193.20.254

Initialized TFliteMicro modelrunner server at port 10818
loop 1: 
CONV_2D took 342000000 us
DEPTHWISE_CONV_2D took 164000000 us
CONV_2D took 103000000 us
CONV_2D took 356000000 us
DEPTHWISE_CONV_2D took 165000000 us
CONV_2D took 95000000 us
CONV_2D took 171000000 us
DEPTHWISE_CONV_2D took 228000000 us
CONV_2D took 142000000 us
ADD took 8000000 us
CONV_2D took 171000000 us
DEPTHWISE_CONV_2D took 68000000 us
CONV_2D took 48000000 us
CONV_2D took 74000000 us
DEPTHWISE_CONV_2D took 67000000 us
CONV_2D took 67000000 us
ADD took 3000000 us
CONV_2D took 71000000 us
DEPTHWISE_CONV_2D took 67000000 us
CONV_2D took 67000000 us
ADD took 3000000 us
CONV_2D took 71000000 us
DEPTHWISE_CONV_2D took 22000000 us
CONV_2D took 76000000 us
CONV_2D took 200000000 us
DEPTHWISE_CONV_2D took 42000000 us
CONV_2D took 196000000 us
ADD took 1000000 us
CONV_2D took 198000000 us
DEPTHWISE_CONV_2D took 43000000 us
CONV_2D took 196000000 us
ADD took 1000000 us
CONV_2D took 199000000 us
DEPTHWISE_CONV_2D took 42000000 us
CONV_2D took 196000000 us
ADD took 1000000 us
CONV_2D took 198000000 us
DEPTHWISE_CONV_2D took 42000000 us
CONV_2D took 300000000 us
CONV_2D took 452000000 us
DEPTHWISE_CONV_2D took 80000000 us
CONV_2D took 457000000 us
ADD took 2000000 us
CONV_2D took 453000000 us
DEPTHWISE_CONV_2D took 82000000 us
CONV_2D took 458000000 us
ADD took 2000000 us
CONV_2D took 451000000 us
DEPTHWISE_CONV_2D took 22000000 us
CONV_2D took 190000000 us
CONV_2D took 310000000 us
DEPTHWISE_CONV_2D took 37000000 us
CONV_2D took 327000000 us
ADD took 1000000 us
CONV_2D took 311000000 us
DEPTHWISE_CONV_2D took 37000000 us
CONV_2D took 327000000 us
ADD took 1000000 us
CONV_2D took 311000000 us
DEPTHWISE_CONV_2D took 36000000 us
CONV_2D took 653000000 us
CONV_2D took 828000000 us
AVERAGE_POOL_2D took 10000000 us
CONV_2D took 57000000 us
RESHAPE took 0 us
SOFTMAX took 0 us
run ms: 10565.000000 index 905, score: 0.196136 
loop 2: 
CONV_2D took 342000000 us
DEPTHWISE_CONV_2D took 164000000 us
CONV_2D took 103000000 us
CONV_2D took 356000000 us
DEPTHWISE_CONV_2D took 165000000 us
CONV_2D took 95000000 us
CONV_2D took 171000000 us
DEPTHWISE_CONV_2D took 227000000 us
CONV_2D took 142000000 us
ADD took 8000000 us
CONV_2D took 171000000 us
DEPTHWISE_CONV_2D took 68000000 us
CONV_2D took 47000000 us
CONV_2D took 74000000 us
DEPTHWISE_CONV_2D took 67000000 us
CONV_2D took 67000000 us
ADD took 3000000 us
CONV_2D took 71000000 us
DEPTHWISE_CONV_2D took 67000000 us
CONV_2D took 67000000 us
ADD took 2000000 us
CONV_2D took 71000000 us
DEPTHWISE_CONV_2D took 22000000 us
CONV_2D took 76000000 us
CONV_2D took 200000000 us
DEPTHWISE_CONV_2D took 42000000 us
CONV_2D took 196000000 us
ADD took 2000000 us
CONV_2D took 198000000 us
DEPTHWISE_CONV_2D took 43000000 us
CONV_2D took 196000000 us
ADD took 1000000 us
CONV_2D took 199000000 us
DEPTHWISE_CONV_2D took 42000000 us
CONV_2D took 197000000 us
ADD took 1000000 us
CONV_2D took 199000000 us
DEPTHWISE_CONV_2D took 42000000 us
CONV_2D took 299000000 us
CONV_2D took 453000000 us
DEPTHWISE_CONV_2D took 81000000 us
CONV_2D took 457000000 us
ADD took 2000000 us
CONV_2D took 454000000 us
DEPTHWISE_CONV_2D took 82000000 us
CONV_2D took 458000000 us
ADD took 2000000 us
CONV_2D took 451000000 us
DEPTHWISE_CONV_2D took 22000000 us
CONV_2D took 190000000 us
CONV_2D took 310000000 us
DEPTHWISE_CONV_2D took 37000000 us
CONV_2D took 327000000 us
ADD took 1000000 us
CONV_2D took 311000000 us
DEPTHWISE_CONV_2D took 37000000 us
CONV_2D took 328000000 us
ADD took 1000000 us
CONV_2D took 311000000 us
DEPTHWISE_CONV_2D took 36000000 us
CONV_2D took 653000000 us
CONV_2D took 828000000 us
AVERAGE_POOL_2D took 9000000 us
CONV_2D took 57000000 us
RESHAPE took 0 us
SOFTMAX took 0 us
run ms: 10566.000000 index 905, score: 0.196136 
run ms: 10565.500000

6.2 Stream test data to ModelRunner for TFLite service
On Linux or Windows PC, prepare one image (jpg or png formate). Then use following
command to post data to ModelRunner for TFLite service on device. The json response show the
TFLite Version, inferrence time and label index. 

# curl -XPOST -H 'Content-Type: image/*' --data-binary "@/home/xiao/panda.jpg" 'http://10.193.20.66:10818/v1?run=1' | jq
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 51055  100    57  100 50998      4   4374  0:00:14  0:00:11  0:00:03    15
{
  "engine": "TFLiteMicro-2.6.0",
  "timing": 10566,
  "index": 389
}
