How-to-start questions

Hi !

For a project which needs to analyse video streams using OpenCV in Python, I discovered the Nvidia Tegra K1 (as much as I have calculation and latency issues with Raspberry).
So, I want to buy a Tegra K1 and a carrier board to set up an environment, which has to allow me to :

  • develop and test my OpenCV algorithm
  • test in “real” life this hardware (I have to install this module in a warehouse)

I identify the Toradex hardware :

Even if Toradex provides its own Linux Distribution based on Nvidia L4T, I would like to confirm that I can use the Nvidia L4T and its OpenCV/CUDA integration, etc. out of the box on this hardware. Can I ?

What would you advise me to buy else, if this hardware doesn’t meet my needs ?

I obviously see the NVidia dev kit (https://www.siliconhighwaydirect.co.uk/product-p/940-7r375-0001-000.htm), but I want to develop on the targeted hardware as much as possible.

Thank you in advance.
Damien.

I doubt you could load the stock L4T without significant modification. Basically the board support package for the Tegra K1 would be the same, but the board support package for components surrounding the TK1 would differ. Changes are likely to include areas such as the device tree and firmware.

I have an Apalis T30 sitting here (uses the same carrier board) and the files used for direct hardware access (such as video hardware and USB controllers) appear to be directly from an L4T R16.x release. Much of the rest of the BSP differs.

Thank you.

In a nutshell, the differences between the Toradex and the Nvidia dev kit look like the differences between two motherboard. And the NVidia L4T doesn’t embed the drivers for the Toradex hardware. And Toradex most probably provides its own Linux package to embed these drivers.

I buy the NVidia Dev kit to test the OpenCV4Tegra performance. I will buy a Toradex in a second time, if the TK1 meets my needs.