Hello,
I have experience with C/C++, however, I have never worked with Nvidia platforms. I am trying to take advantage of the GPU computing in my projects. I have a Jetson TX2 and I am trying to develop some application in robotics. Thanks to some videos that I found on the internet, I could install the JetPack 3.2 in my Jetson. I also could run some demos. However, I cannot find a video which explains how to start to program the Jetson.
Is there some tutorial to explain how to set up the developer environment? I am confused, do I need to use a host computer to program the Jetson and then download the app to the Jetson (more or less the same concept than an Arduino)? Can I develop my applications directly on the Jetson? What things do I need to configure? Do I need to install an IDE in the Jetson? I want to use CUDA, openCV, and AI in my application. I have been looking for tutorials in Nvidia learn, but there is not a tutorial to explain in a simple way how to set the developer environment and start to create applications.
I really will appreciate your help.
Thanks.
Hi ercalleortiz, JetPack will automatically install CUDA toolkit, cuDNN, TensorRT, openCV (and VisionWorks), so you should already have these libraries installed and ready to use.
Yes, after JetPack is installed you do not technically need the host PC anymore (unless you are using Tegra System Profiler tool). Although you do retain the option of cross-compiling your code from the host and deploying to your Jetson, you can just program directly on your Jetson like it was an ordinary Linux machine. See this related thread: https://devtalk.nvidia.com/default/topic/1014825/?comment=5170657
It’s your choice if you wish to use a dedicated IDE or not. Plenty of Linux developers have their own hodgepodge of editors/terminals/toolchain to their liking - like vi/vim, GEdit, make, ect. See these topics for IDE suggestions:
https://devtalk.nvidia.com/default/topic/1035800/?comment=5261923
https://devtalk.nvidia.com/default/topic/1016158/recommended-ide-for-jetson-tx2-/
Btw, most of the installed CUDA and samples, so on, are in “/usr/local/”. Go there, run “ls” to see what is present.