learn to read.
I was only making a reference to assembly programming being outdated as is non-AVX systems and using to conferring the idea that the costs may not be worth it. My intent wasn't to make a direct suggestion that assembly programming is part of the issue. NGS was likely to be developed using high-level programming environment(s).
they'd have to go into the engine and write implementations that don't use any functions that utilize the instruction set which is a waste of their time
To reiterate on my position, this was what I was arguing for. When the codebase is being branched like this, the costs and issues tend to add up quite a bit. Odds are the costs far outweigh the benefits. Think of the devs that finally made breakthroughs in new content/features but only to be hindered by having to update the non-AVX code which may or may not behave.
I know the pain because I still support Windows 98/XP forks for work stuffs.