i965: Eliminate dead code more aggressively.

If an instruction writes reg but nothing later uses it, then we don't
need to bother doing it.  Before, we were just killing code that was
never read after it was ever written.

This removes many interpolation instructions for attributes with only
a few comopnents used.  Improves nexuiz high-settings performance .46%
+/- .12% (n=3) on my Ironlake.

diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
index b12a48004108f971bb82d77fa58b975307b76edd..ac795e0bda1a6f15780414e1535978dd225a0bfe 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -2801,26 +2801,17 @@ bool
 fs_visitor::dead_code_eliminate()
 {
    bool progress = false;
-   int num_vars = this->virtual_grf_next;
-   bool dead[num_vars];
-
-   for (int i = 0; i < num_vars; i++) {
-      dead[i] = this->virtual_grf_def[i] >= this->virtual_grf_use[i];
-
-      if (dead[i]) {
-        /* Mark off its interval so it won't interfere with anything. */
-        this->virtual_grf_def[i] = -1;
-        this->virtual_grf_use[i] = -1;
-      }
-   }
+   int pc = 0;
 
    foreach_iter(exec_list_iterator, iter, this->instructions) {
       fs_inst *inst = (fs_inst *)iter.get();
 
-      if (inst->dst.file == GRF && dead[inst->dst.reg]) {
+      if (inst->dst.file == GRF && this->virtual_grf_use[inst->dst.reg] <= pc) {
         inst->remove();
         progress = true;
       }
+
+      pc++;
    }
 
    return progress;