freedreno/a4xx: better occlusion/sample counting
authorRob Clark <robclark@freedesktop.org>
Sun, 14 Feb 2016 14:59:41 +0000 (09:59 -0500)
committerRob Clark <robclark@freedesktop.org>
Wed, 17 Feb 2016 15:41:55 +0000 (10:41 -0500)
commit62fa868728c729152af0d7cecd1d3e47e831cb7d
tree9539628de9b069748ef79a6326fcdddc3b3ab792
parent87eb4067912921ee3a260792b82fa5d4501ae8a9
freedreno/a4xx: better occlusion/sample counting

This seems to give more reliable results.  More similar to what we do on
a3xx, although I think it breaks the a3xx theory that the four sets of
results map to each MRT (since we appear to still only have four sets on
a4xx).  The divide-by-two is a bit odd, but seems to be needed for some
reason.

Signed-off-by: Rob Clark <robclark@freedesktop.org>
src/gallium/drivers/freedreno/a4xx/fd4_query.c