mirror of
https://github.com/FEX-Emu/FEX.git
synced 2026-10-09 18:00:21 +02:00
Ryan Houdek
887200e571
These instructions aren't super amazing due to the fact that they have both a source mask and a destination duplication mask. Setup a case where we can generate more optimal code in /most/ cases. There are a few that still fall down a "bad" path for the result broadcast but in most cases they are optimal. Still to be seen what games typically use the broadcast mask as. AVX in its infinite wisdom expanded DPPS to 256-bit, while leaving DPPD to only support 128-bit still. This leaves the original implementation alone for 256-bit DPPS since I don't want to break it. This is another instruction that gets a free optimization when SVE-128bit is supported!