nir/algebraic: Separate ffma lowering from fusing
authorJason Ekstrand <jason.ekstrand@intel.com>
Thu, 5 May 2016 23:58:44 +0000 (16:58 -0700)
committerJason Ekstrand <jason.ekstrand@intel.com>
Wed, 11 May 2016 18:44:35 +0000 (11:44 -0700)
commit1b72c31e1f1947123d8c236b56e230f030f60cf9
tree03ec41421f5e06d29d64af7303bc5c11084edba7
parent5886d1bad13a1c0106b7f42191bbc399fff4a0d9
nir/algebraic: Separate ffma lowering from fusing

The i965 driver has its own pass for fusing mul+add combinations that's
much smarter than what nir_opt_algebraic can do so we don't want to get the
nir_opt_algebraic one just because we didn't set lower_ffma.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
src/compiler/nir/nir.h
src/compiler/nir/nir_opt_algebraic.py
src/gallium/drivers/freedreno/ir3/ir3_nir.c