projects / mesa.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 060979a) | patch
v3d: Do uniform rematerialization spilling before dropping threadcount
authorEric Anholt <eric@anholt.net>
Tue, 26 Feb 2019 18:49:25 +0000 (10:49 -0800)
committerEric Anholt <eric@anholt.net>
Tue, 5 Mar 2019 20:57:39 +0000 (12:57 -0800)
commit1e98f02d887dada530595bc0c74292d4678c5e1a
tree3bc96cc728f6f95f65edb98cc44c08f6e94d7561tree
parent060979a380be0a6149e3e875ee24fdb1e7872821commit | diff
v3d: Do uniform rematerialization spilling before dropping threadcount

This feels like the right tradeoff for threads vs uniforms, particularly
given that we often have very short thread segments right now:

total instructions in shared programs: 6411504 -> 6413571 (0.03%)
total threads in shared programs: 153946 -> 154214 (0.17%)
total uniforms in shared programs: 2387665 -> 2393604 (0.25%)
src/broadcom/compiler/vir_register_allocate.c diff | blob | history