[Mesa-dev] [PATCH] i965/vec4: check writemask when bailing out at register coalesce

[Alejandro Piñeiro]

opt_register_coalesce stopped to check previous instructions to
coalesce with if somebody else was writing on the same
destination. This can be optimized to check if somebody else was
writing to the same channels of the same destination using the
writemask.

Shader DB results (taking into account only vec4):

total instructions in shared programs: 1781593 -> 1734957 (-2.62%)
instructions in affected programs:     1238390 -> 1191754 (-3.77%)
helped:                                12782
HURT:                                  0
GAINED:                                0
LOST:                                  0
---

piglit run on both IR and NIR path without any regression.


 src/mesa/drivers/dri/i965/brw_vec4.cpp | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_vec4.cpp b/src/mesa/drivers/dri/i965/brw_vec4.cpp
index 85dc372..998dbf2 100644
--- a/src/mesa/drivers/dri/i965/brw_vec4.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4.cpp
@@ -1089,11 +1089,12 @@ vec4_visitor::opt_register_coalesce()
 	 if (interfered)
 	    break;
 
-         /* If somebody else writes our destination here, we can't coalesce
-          * before that.
+         /* If somebody else writes the same channels of our destination here,
+          * we can't coalesce before that.
           */
-         if (inst->dst.in_range(scan_inst->dst, scan_inst->regs_written))
-	    break;
+         if ((inst->dst.in_range(scan_inst->dst, scan_inst->regs_written) &&
+              (inst->dst.writemask & scan_inst->dst.writemask)))
+            break;
 
          /* Check for reads of the register we're trying to coalesce into.  We
           * can't go rewriting instructions above that to put some other value
-- 
2.1.4