[Mesa-dev] [PATCH V3] i965 : Optimize atom state flag checks

[aravindan.muthukumar at intel.com]

From: Aravindan Muthukumar <aravindan.muthukumar at intel.com>

This patch improves CPI Rate(Cycles per Instruction) and branch miss predict
for i965. The function check_state() was showing CPI retired rate.

Performance stats with android:
- CPI retired lowered by 28% (lower is better)
- Branch missprediction lowered by 13% (lower is better)
- 3DMark improved by 2%

The dissassembly doesn't show difference, although above results were observed
with patch.

V2:
- Removed memset() change
- Changed commit message as per review comments

V3:
- Indentation and changes to remove check_state as function

Signed-off-by: Aravindan Muthukumar <aravindan.muthukumar at intel.com>
Signed-off-by: Yogesh Marathe <yogesh.marathe at intel.com>
Tested-by: Asish <asish at intel.com>
---
 src/mesa/drivers/dri/i965/brw_defines.h      |  4 ++++
 src/mesa/drivers/dri/i965/brw_state_upload.c | 24 +++++++++++-------------
 2 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index 2a8dbf8..8c9a510 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -1687,3 +1687,7 @@ enum brw_pixel_shader_coverage_mask_mode {
 # define CSDBG2_CONSTANT_BUFFER_ADDRESS_OFFSET_DISABLE (1 << 4)
 
 #endif
+
+/* Checking the state of mesa and brw before emitting atoms */
+#define CHECK_BRW_STATE(a,b) ((a.mesa & b.mesa) | (a.brw & b.brw))
+
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index acaa97e..1846624 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -217,12 +217,6 @@ void brw_destroy_state( struct brw_context *brw )
 /***********************************************************************
  */
 
-static bool
-check_state(const struct brw_state_flags *a, const struct brw_state_flags *b)
-{
-   return ((a->mesa & b->mesa) | (a->brw & b->brw)) != 0;
-}
-
 static void accumulate_state( struct brw_state_flags *a,
 			      const struct brw_state_flags *b )
 {
@@ -443,10 +437,8 @@ check_and_emit_atom(struct brw_context *brw,
                     struct brw_state_flags *state,
                     const struct brw_tracked_state *atom)
 {
-   if (check_state(state, &atom->dirty)) {
-      atom->emit(brw);
-      merge_ctx_state(brw, state);
-   }
+   atom->emit(brw);
+   merge_ctx_state(brw, state);
 }
 
 static inline void
@@ -541,7 +533,10 @@ brw_upload_pipeline_state(struct brw_context *brw,
 	 const struct brw_tracked_state *atom = &atoms[i];
 	 struct brw_state_flags generated;
 
-         check_and_emit_atom(brw, &state, atom);
+         /* Checking the state and emitting atoms */
+         if (CHECK_BRW_STATE(state, atom->dirty)) {
+            check_and_emit_atom(brw, &state, atom);
+         }
 
 	 accumulate_state(&examined, &atom->dirty);
 
@@ -550,7 +545,7 @@ brw_upload_pipeline_state(struct brw_context *brw,
 	  *     fail;
 	  */
 	 xor_states(&generated, &prev, &state);
-	 assert(!check_state(&examined, &generated));
+	 assert(!CHECK_BRW_STATE(examined, generated));
 	 prev = state;
       }
    }
@@ -558,7 +553,10 @@ brw_upload_pipeline_state(struct brw_context *brw,
       for (i = 0; i < num_atoms; i++) {
 	 const struct brw_tracked_state *atom = &atoms[i];
 
-         check_and_emit_atom(brw, &state, atom);
+         /* Checking the state and emitting atoms */
+         if (CHECK_BRW_STATE(state, atom->dirty)) {
+            check_and_emit_atom(brw, &state, atom);
+         }
       }
    }
 
-- 
2.7.4