The thing just is that TI didn't exactly retain the people that know the internals of the OMAP series. If you hit any bugs you're on your own unfortunately. (Your milage with the existing documentation may vary also).

OMAP's been dead a long time. Sitara is OMAP with the media blocks stripped out, and it sure looks like it's on their roadmap for a while to come.

https://training.ti.com/sites/default/files/docs/TI%20Missio...

Of course it will be produced for as long as customers want it. But the roadmap makes it also crystal clear where they are going: More DSP cores. We likely won't see anything newer than the Cortex-A15. TI builds what enough customers want, and if you're not locked in enough to their DSPs or general ecosystem, you'll move away. That in return means there won't be anything new. No single customer is big enough to warrant bigger cores alone.

A nice quote I remember "I'd rather buy TI stocks than their products".

Nice spec. 2x Ethernet?! Custom router possibilities abound.

More expensive than Pi however. I can get an entire pi for the same price as this module,

The 2x Ethernet is a little quirky in implementation: Internally, the CPU only has one ethernet MAC, which is connected to a switch on the same die. In practice, linux sees two ethernet ports, but this is implemented by VLAN tagging on the internal switch. So if you think you have two actual ports and want to interact with VLANs, you're gonna have a bad time.

And the performance isn't very good either from what I have heard. It is a good candidate if you want to try and make something cheap, but few (none?) of the offerings really are (olimex has one if anyone is interested). Personally I am pretty tired of the Linux ecosystem in general. It is pretty well known that you can ship broken stuff by slapping an open source sticker on it.

Thanks for the heads up.

Lots of compromises in the sbc space. A lot aren’t required but cost is the most common factor.