Looking for status of SIMD implementation

[Matthias Kretz]

Thank you Richard for your insight. There's a lot to discuss but I'll just 
start with the question how general the expression of data-parallelism is 
(supposed to be) for stdx::simd. Most of my answer is written up in
https://wg21.link/p0349, "Assumptions about the size of datapar".

Note that we (SG1) called it datapar instead of simd, because the intent is to 
express data-parallelism and not necessarily SIMD hardware (i.e. an abstract 
view vs. a hardware-centric view). Oh, and I wrote this paper when SVE had not 
been announced yet, but you may notice that I knew what was coming. ;)

Oh, and also relevant for what stdx::simd wants to be:
https://wg21.link/p0851, "simd<T> is neither a product type nor a container 
type".

I'll try to answer more directly later.

-- 
──────────────────────────────────────────────────────────────────────────
 Dr. Matthias Kretz                           https://mattkretz.github.io
 GSI Helmholtz Centre for Heavy Ion Research               https://gsi.de
 std::experimental::simd              https://github.com/VcDevel/std-simd
──────────────────────────────────────────────────────────────────────────