Hello Everyone,

I know the Nvidia TX is a powerful embedded board, but I want to know what is the main differences and advantages in NVIDIA TX1/2 with other embedded boards in the market, I found many embedded boards ( ARM and X86) and it’s also very powerful and has the same features comparing with Nvidia TX. Simply Why NVIDIA TX1/2 and what are the projects that really required and need Nvidia TX1/2

Many Thanks,

The TX1/2 have a powerful GPU available for compute…so far as I know nobody else has this. If you use CUDA the TK1/TX1/TX2 are king of the hill. You could use other hardware, but you’d never come close to the low power consumption…you’d need a much bigger battery on anything mobile.

Lots of image processing comes to mind, and neural networks (e.g., face recognition). To do it on another board it’ll weigh a lot more and perhaps require ten times the electrical power. You wouldn’t use a TK1/TX1/TX2 to build a new neural net, but you would use it to process an existing net in real time. See CUDA and cuDNN.

The 4k UHD video ability is also nice.

Aside from the CUDA capabilities, the TX2 is also one of the (as far as I know the) strongest ARM system available. I also use it as a full desktop system that even is fully silent until even more power is needed. It really features the latest technology.

The TX1/TX2 offer 12 MIPI lanes for camera control. They are the most capable embedded platforms on the market for running multiple video streams through an inference engine.

For me there were a few considerations which made me get a devkit for this platform. The high performance of the CPU was very nice but the main thing was the strong GPU in companion with the nice amount of camera lanes and the solid amount of video decode and encode options. The libraries Nvidia themselves has available for this platform as well as the other libraries in the Jetson pack looked like a really wide range options for me.
I myself am interested in real time image processing, having the option to work with high quality video should make my life significantly easier in this space.

Also in general, I trust Nvidia graphics drivers a lot more than the other GPU vendors for the embedded space.