Seems to be the case at least in built from recent commit, seen in various benchmarks and fraps.
Can anyone confirm? GCC is 4.7.2.
-funroll-loops resulting in 3-5% performance increase?
Re: -funroll-loops resulting in 3-5% performance increase?
For making a quick build to try I've found that -flto helps a lot (link-time optimization). Not as good as profile generated optimization but better than nothing.
From my experience, -funroll-loops brings a lot of cache hiccups as it causes some functions to really bloat up in size
From my experience, -funroll-loops brings a lot of cache hiccups as it causes some functions to really bloat up in size
Re: -funroll-loops resulting in 3-5% performance increase?
PGO cannot optimize everything. voodoo.o is broken when PGO used.
Re: -funroll-loops resulting in 3-5% performance increase?
Then you're doing it wrong. You should profile it with 1 render thread
- SarahWalker
- Site Admin
- Posts: 2054
- Joined: Thu 24 Apr, 2014 4:18 pm
Re: -funroll-loops resulting in 3-5% performance increase?
Or build with '-fprofile-use -fprofile-correction'