i965/fs: Perform CSE on MOV ..., VF instructions.

Safe from causing optimization loops, since we don't constant propagate
VF arguments.

(for this and the previous patch):
total instructions in shared programs: 4289075 -> 4271932 (-0.40%)
instructions in affected programs:     1616779 -> 1599636 (-1.06%)

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

diff --git a/src/mesa/drivers/dri/i965/brw_fs_cse.cpp b/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
index b1c433e45c7ddae53d13cc0cd8f9dfe87287b0fd..38fae1781142f2adacc4c2e06af651475e949fb2 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
@@ -62,6 +62,7 @@ static bool
 is_expression(const fs_inst *const inst)
 {
    switch (inst->opcode) {
+   case BRW_OPCODE_MOV:
    case BRW_OPCODE_SEL:
    case BRW_OPCODE_NOT:
    case BRW_OPCODE_AND:
@@ -194,11 +195,16 @@ fs_visitor::opt_cse_local(bblock_t *block)
          }
 
          if (!found) {
-            /* Our first sighting of this expression.  Create an entry. */
-            aeb_entry *entry = ralloc(cse_ctx, aeb_entry);
-            entry->tmp = reg_undef;
-            entry->generator = inst;
-            aeb.push_tail(entry);
+            if (inst->opcode != BRW_OPCODE_MOV ||
+                (inst->opcode == BRW_OPCODE_MOV &&
+                 inst->src[0].file == IMM &&
+                 inst->src[0].type == BRW_REGISTER_TYPE_VF)) {
+               /* Our first sighting of this expression.  Create an entry. */
+               aeb_entry *entry = ralloc(cse_ctx, aeb_entry);
+               entry->tmp = reg_undef;
+               entry->generator = inst;
+               aeb.push_tail(entry);
+            }
          } else {
             /* This is at least our second sighting of this expression.
              * If we don't have a temporary already, make one.