End of the road ?

[Melkhior]

Seems Andreas moved on.



Good luck with the new job. I wish there had been a sucessor the the parallella.

[dobkeratops]

hmmm..

does this mean the epiphany architecture wont be available beyond the darpa projects, .. and no 'parallela 2' ? Did he wind 'adapteva' down?

Has the world missed the opportunity? - or will it just continue internally as a DARPA only project.

AI should be possible on a simpler processor like this, without all the graphics IP (in nvidia) or big-core IP (intel/ARM).. whilst still being more flexible than dedicated neurotrophic chips.

given the GPIO on the E5 , I was hoping it would be possible for them to build something more standalone.

it would be a great shame if this has simply failed to take hold, just when we have all the tools like tensor flow and functional programming (even lambda functions in plain old C++) etc gaining popularity.

this is my big worry.. that the chips specifically suitable for AI might not get widespread availability (GPUs are out there in the open, because of demand for games)

[GreggChandler]

A public announcement here would have been nice if it is to be abandoned.

[dobkeratops]

I dithered over actually getting a parallela myself.. seeing the E5 I regret not putting time into supporting it. If it does disappear, I guess it's our fault for not getting behind it enough.

I was interested to see 'something like CELL reborn' .. but dithered over applicability; but then I got into neural nets, and am now way more convinced about the idea.

[CRosenbauer]

[dobkeratops]