Jump to content
Tinker Board

Getting Started

Recommended Posts

[Image: RiqlldHm.jpg]
ASUS Tinker Edge R is an Single Board Computer (SBC) specially designed for AI applications. It uses Rockchip NPU, a Machine Learning (ML) accelerator that speeds up processing efficiency, lowers power demands and makes it easier to build connected devices and intelligent applications. With this integrated Machine Learning (ML) accelerator, the Tinker Edge R is capable of performing 3 tera-operations per second (TOPS), using low power consumption. And it’s optimized for Neural Network (NN) architecture, which means Tinker Edge R can support multiple Machine Learning (ML) frameworks and let lots common Machine Learning (ML) models can be easily compiled and run on the Tinker Edge R.

Requirement

  • 1 x USB Type-C® cable with data transfer function (to connect your PC to the board’s data port)
  • 1 x 12~19V power supply*
  • 1 x Monitor with HDMI cable or USB Type-C® (DP) cable
  • 1 x Keyboard and Mouse set
    * The power supply is purchased separately.

Software preparation

  • Get Tinker Edge R ROM Image
    Check ASUS Tinker Edge R official website to get latest image. https://tinker-board.asus.com/download-list.html, select Tinker Edge R from dropdown menu.
     
  • Get Edge R Flash tool (Windows GUI version)
    Check ASUS Tinker Edge R official website to get newest version. https://tinker-board.asus.com/download-list.html, select Tinker Edge R from dropdown menu.
     
  • Get Edge R Flash tool (Windows/Linux Command line)
    Find the command line flash tool in ROM image directory.
     
  • [Windows] Install Rockchip Driver
    Find the DriverAssitant zip package in ROM image directory, unzip it and execute DriverInstall.exe to install driver.

Flashing the Tinker Edge R

Initiating MASKROM mode

  1. Connect the USB Type-C® cable to the USB Type-C® ports on the Tinker Edge R and your host computer.
  2. Before you begin the flashing procedure, please ensure of the following:
    • The board is completely powered off, and the power cord and cables connecting the board to your computer are all disconnected.
    • In order to set boot mode to MASKROM mode, use metal object or a jumper cap to short recovery header.
  3. Power on the Tinker Edge R, board should automatically be booted into MASKROM mode.
  4. Remember to remove the jumper cap upon power on.

[Image: a8Whdfym.png]
Figure1: Recovery header pin position
(Notes: Remember to remove jumper upon power on)

Executing the flash tool

  1. Download the OS image from the Tinker Edge R website, then unzip the image file.
  2. Run the GUI flash tool (Windows OS) or command line (Linux) to start up the flash process. The flash process should take a few minutes.
  3. Once flash completed, Tinker board will be automatically rebooted.

  A. Windows Flash tool (GUI)

  1. Check Device Manager to ensure “Rockusb device” is detected. if problem encountered:
    • Try to reconnect cable directly to PC’s USB port without hub
    • Short Recovery header then power on again
    • Try to re-install Rockchip driver.
  2. Unzip GUI flash tool package, run Tinker-Flash-Tool.exe (Run as Administrator option)
  3. Follow the instruction, select OS image file, pick the target Tinker board and execute the flash.
    [Image: ukwlSjym.jpg]
    Figure2: Main user interface of ASUS GUI flash tool

  B. Flash tool-command Line (Windows, Linux)

  1. Make sure Recovery header is no longer being shorted after power on.
  2. Run the flash script flash.cmd for Windows or flash.sh for Linux to start the flash process.
  3. Refer to README.txt for more information.

web: https://tinker-board.asus.com/doc_er.html#started
pdf: https://tinker-board.asus.com/images/doc...tarted.pdf

Share this post


Link to post
Share on other sites

Hi, 

 

I am not sure if  the gpu works properly.  

 

The command 

cat /sys/class/devfreq/ff9a0000.gpu/cur_freq gives me a constant value. 

 

I don't understand  neither where is the Asus API that I can use to train models. For instance, tensorflow does not support gpu on arm. What is the workaroud?

 

Best

Share this post


Link to post
Share on other sites

Please tell: why command "sudo /usr/bin/npu_powerctrl -i" fails with message "Error writing to /sys/class/gpio/export value 4:"?

And hardware NPU does not work (benchmarks of neural tests is very slow - looks like test are running on cpu, not NPU)?

 

Share this post


Link to post
Share on other sites
6 hours ago, Hoshino said:

I could not start Tinker Edge R. 

Flushing Failed.

What should I do?

Hello Hoshino, 

is there any error logs? could you describe how you flash the board?

also, related information can be found at:

Documentation > Tinker Edge R > Getting Started > Flashing the Tinker Edge R https://tinker-board.asus.com/doc_er.html#started

ROM image can be downloaded here: https://tinker-board.asus.com/download-list.html

hope it helps!

 

Share this post


Link to post
Share on other sites
On 12/26/2020 at 3:02 AM, RubberBigPepper said:

Please tell: why command "sudo /usr/bin/npu_powerctrl -i" fails with message "Error writing to /sys/class/gpio/export value 4:"?

And hardware NPU does not work (benchmarks of neural tests is very slow - looks like test are running on cpu, not NPU)?

 

Hi RubberBigPepper,
"Error writing to /sys/class/gpio/export value=4" <= "ls /sys/class/gpio/gpio4", Initialize again cause Error.
You can use "lsusb" to confirm the NPU id "2207:0019"
Benchmarks need "rknn" conversion, execute on NPU.
hope it helps!

Share this post


Link to post
Share on other sites

Hi 

Can you provide your download URL?
Is Jump removed during burning?
Thank you

Share this post


Link to post
Share on other sites

Hi guys, 

I have just bought an Asus Tinker Edge R to test and see for my project. I've installed a Linux aarch64 GNU/Linux. I am planning to run tensorflow object detection model using either frozen one, tlifte. I've tested with the tlite model, and the FP is very low like 3-5FPS( frame per Second). base on what I've seen it could run up 100FPS. and I also want to know how to setup and utilize the NPU capabilities.  if you've documentations available on that pls do share. I've seen this one https://github.com/rockchip-linux/rknn-toolkit and test as well, but it doesn't answer what am looking. if any of you has experienced testing tensorflow object detection, and pls do share ur experience. Thank you. 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...