I am experimenting with a Xeon Phi card, and would like (eventually) run code which heavily uses Eigen for vector operations on it. Xeon Phi's have specific vector intrinsics. Are those supported by Eigen? From which version? Did not find anything on the subject. Thanks for any information, cheers!

Not implemented yet. As we already support a bunch of vector engine (SSE, AltiVec, NEON, AVX) adding support for Xeon Phi should not be that difficult, and basically it means copy/pasting the closest existing vector engine (in src/Core/arch/) and adjust to the respective intrinsics. A bit tedious, but nothing challenging to get an initial support.

Any update or progress on this front?
Thanks.