[Mesa-dev] [PATCH] i965/fs: Let sat-prop ignore live ranges if producer already has sat.

[Matt Turner]

This sequence (where both x and w are used afterwards) wasn't handled.

   mul.sat x, y, z
   ...
   mov.sat w, x

We assumed that if x was used after the mov.sat, that we couldn't
propagate the saturate modifier, but in fact x was already saturated.

So ignore the live range check if the producing instruction already
saturates its result. Cuts one instruction from hundreds of TF2 shaders.

total instructions in shared programs: 1995631 -> 1994951 (-0.03%)
instructions in affected programs:     155248 -> 154568 (-0.44%)
---
 src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp b/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp
index 4b5b5ca..079eb2e 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp
@@ -47,8 +47,6 @@ opt_saturate_propagation_local(fs_visitor *v, bblock_t *block)
 
       int src_var = v->live_intervals->var_from_reg(&inst->src[0]);
       int src_end_ip = v->live_intervals->end[src_var];
-      if (src_end_ip > ip && !inst->dst.equals(inst->src[0]))
-         continue;
 
       int scan_ip = ip;
       bool interfered = false;
@@ -61,10 +59,15 @@ opt_saturate_propagation_local(fs_visitor *v, bblock_t *block)
              scan_inst->dst.reg == inst->src[0].reg &&
              scan_inst->dst.reg_offset == inst->src[0].reg_offset &&
              !scan_inst->is_partial_write()) {
-            if (scan_inst->can_do_saturate()) {
-               scan_inst->saturate = true;
+            if (scan_inst->saturate) {
                inst->saturate = false;
                progress = true;
+            } else if (src_end_ip <= ip || inst->dst.equals(inst->src[0])) {
+               if (scan_inst->can_do_saturate()) {
+                  scan_inst->saturate = true;
+                  inst->saturate = false;
+                  progress = true;
+               }
             }
             break;
          }
-- 
1.8.3.2