OS selection in GPU

Autonomous Machines Jetson & Embedded Systems Jetson TK1
1

There is a need of developing a complex computing application using some algorithms.So am thinking to choose a tegra k1 from nvidia for this purpose.Am confused to choose the OS platform Android or Linux ,where i can develop my application. Usually Android have their own api’s or services to access the GPU ,but on the other side can the android provides complete access to the hardware to perform maximum computation . One more advantage is it provide easy UI support.While choosing such platfrom for development no other unwanted background services should run causing latencies or overhead while generating the output.

In case of choosing a Linux Platform (for example Linux for tegra for Tegra processors ), what are the advantages compared to android .how is the level of complexity in developing an UI implementation for the application.

This is the initial stage where am stuck with. What are the advantages and disadvantages in choosing the right OS platform . Which platform is apt for such complex computational applications.

2

If you want to use Jetson TK1, then you can only choose Linux as Android is not supported.

Linux for Tegra on Jetson TK1 does support e.g. CUDA 6.5 so it is suitable for a computing application.

3

Seems like you are more focus on the UI rather than the computational aspect of you task. There are lots of GUI toolkit that you may be able to leverage irrespective of you OS platform decision. However, the choice of compute APIs is not that expansive…

4

So far as unwanted background services is concerned, L4T is just Ubuntu with tegra124 customizations. You can add or remove just about anything you want without much effort. It’s quite easy to do for most things. I think the only real issue anyone runs into is if they do not want a window manager, but there are threads here which provide the information needed if that is the issue.