[Mesa-dev] [PATCH 14/19] i965/fs: Emit load_payload instead of multiple MOVs for large VGRFs.

[Matt Turner]

---
 src/mesa/drivers/dri/i965/brw_fs_cse.cpp | 33 ++++++++++++++++++++------------
 1 file changed, 21 insertions(+), 12 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_fs_cse.cpp b/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
index e40567f..44f1fe4 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
@@ -177,15 +177,20 @@ fs_visitor::opt_cse_local(bblock_t *block, exec_list *aeb)
                entry->tmp = tmp;
                entry->generator->dst = tmp;
 
-               for (int i = 0; i < written; i++) {
-                  fs_inst *copy = MOV(orig_dst, tmp);
+               fs_inst *copy;
+               if (written > 1) {
+                  fs_reg *sources = ralloc_array(mem_ctx, fs_reg, written);
+                  for (int i = 0; i < written; i++) {
+                     sources[i] = tmp;
+                     sources[i].reg_offset = i;
+                  }
+                  copy = new(mem_ctx) fs_inst(SHADER_OPCODE_LOAD_PAYLOAD, orig_dst, sources, written);
+               } else {
+                  copy = MOV(orig_dst, tmp);
                   copy->force_writemask_all =
                      entry->generator->force_writemask_all;
-                  entry->generator->insert_after(copy);
-
-                  orig_dst.reg_offset++;
-                  tmp.reg_offset++;
                }
+               entry->generator->insert_after(copy);
 	    }
 
 	    /* dest <- temp */
@@ -195,15 +200,19 @@ fs_visitor::opt_cse_local(bblock_t *block, exec_list *aeb)
                assert(inst->dst.type == entry->tmp.type);
                fs_reg dst = inst->dst;
                fs_reg tmp = entry->tmp;
-               fs_inst *copy = NULL;
-               for (int i = 0; i < written; i++) {
+               fs_inst *copy;
+               if (written > 1) {
+                  fs_reg *sources = ralloc_array(mem_ctx, fs_reg, written);
+                  for (int i = 0; i < written; i++) {
+                     sources[i] = tmp;
+                     sources[i].reg_offset = i;
+                  }
+                  copy = new(mem_ctx) fs_inst(SHADER_OPCODE_LOAD_PAYLOAD, dst, sources, written);
+               } else {
                   copy = MOV(dst, tmp);
                   copy->force_writemask_all = inst->force_writemask_all;
-                  inst->insert_before(copy);
-
-                  dst.reg_offset++;
-                  tmp.reg_offset++;
                }
+               inst->insert_before(copy);
             }
 
             /* Set our iterator so that next time through the loop inst->next
-- 
1.8.3.2