vc4: Compile the LT image helper per cpp we might load/store.
authorEric Anholt <eric@anholt.net>
Wed, 8 Aug 2018 00:53:24 +0000 (17:53 -0700)
committerEric Anholt <eric@anholt.net>
Wed, 8 Aug 2018 22:53:25 +0000 (15:53 -0700)
commit3e06b918aab3c6a4ca30f5e935aa6996f7009d56
tree6a24cfffee8ecaf22d2e446472d591310d316c92
parentd6a174669f1452e32ffd24e31cd4b1d4a6e09295
vc4: Compile the LT image helper per cpp we might load/store.

For the partial load/store support I'm about to add, we want the memcpy to
be compiled out to a single load/store.  This should also eliminate the
calls to vc4_utile_width/height().

Improves x11perf -putimage100 performance by  3.76344% +/- 1.16978% (n=15)
src/gallium/drivers/vc4/vc4_tiling_lt.c