Draft: nir/algebraic: separate ffma if it allows the multiplication to be CSE'd
This can reduce register usage when RADV disables nir_op_ffma
lowering.
Separated from !8056 (merged) because it currently regresses Intel
Due to an influx of spam, we have had to impose restrictions on new accounts. Please see this wiki page for instructions on how to get full permissions. Sorry for the inconvenience.
This can reduce register usage when RADV disables nir_op_ffma
lowering.
Separated from !8056 (merged) because it currently regresses Intel