https://bugs.freedesktop.org/show_bug.cgi?id=34156
--- Comment #8 from Dave Witbrodt dawitbro@sbcglobal.net 2011-03-02 04:47:31 PST --- (In reply to comment #7)
I'm seeing some performance impact from 077c448d, mostly due to the extra calls to evergreen_context_pipe_state_set_vs_resource() / evergreen_context_pipe_state_set_ps_resource(). The earlier commits don't seem to make much of a difference for me.
Hmmm... for me performance was first impacted with a77e813d, and building Mesa from the commit before that one (3b1c1f02) still gave me the faster performance Mesa had jumped to in January. (I only reverted the other 3 commits because they depended on a77e813d, and I did not know how to preserve them without it.) Because of this, I was never really able to separate the impact of 077c448d by itself.
As described above, just moving from 3b1c1f02 to a77e813d reduces (for me) the frame rate in 'torcs' by roughly 30%.