This keeps us from needing to reemit all the other stage state just
because a surface changed.
Improves unoptimized glamor x11perf -f8text by 1.10201% +/- 0.489869%
(n=296).
---
src/mesa/drivers/dri/i965/brw_binding_tables.c | 51 ++++++++++++++++++++------
src/mesa/drivers/dri/i965/gen7_gs_state.c | 7 ----
src/mesa/drivers/dri/i965/gen7_vs_state.c | 7 ----
src/mesa/drivers/dri/i965/gen7_wm_state.c | 7 ----
4 files changed, 39 insertions(+), 33 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_binding_tables.c
b/src/mesa/drivers/dri/i965/brw_binding_tables.c
index b39bd10..ca42472 100644
--- a/src/mesa/drivers/dri/i965/brw_binding_tables.c
+++ b/src/mesa/drivers/dri/i965/brw_binding_tables.c
@@ -50,7 +50,7 @@
* This copies brw_stage_state::surf_offset[] into the indirect state section
* of the batchbuffer (allocated by brw_state_batch()).
*/
-static void
+static bool
brw_upload_binding_table(struct brw_context *brw,
GLbitfield brw_new_binding_table,
struct brw_stage_state *stage_state)
@@ -63,8 +63,9 @@ brw_upload_binding_table(struct brw_context *brw,
if (stage_state->bind_bo_offset != 0) {
brw->state.dirty.brw |= brw_new_binding_table;
stage_state->bind_bo_offset = 0;
+ return true;
}
- return;
+ return false;
}
if (INTEL_DEBUG & DEBUG_SHADER_TIME) {
@@ -81,6 +82,7 @@ brw_upload_binding_table(struct brw_context *brw,
memcpy(bind, stage_state->surf_offset, prog_data->binding_table.size_bytes);
brw->state.dirty.brw |= brw_new_binding_table;
+ return true;
}
/**
@@ -92,15 +94,23 @@ brw_upload_binding_table(struct brw_context *brw,
static void
brw_vs_upload_binding_table(struct brw_context *brw)
{
- brw_upload_binding_table(brw, BRW_NEW_VS_BINDING_TABLE, &brw->vs.base);
+ if (brw_upload_binding_table(brw, BRW_NEW_VS_BINDING_TABLE, &brw->vs.base))
{
+ if (brw->gen >= 7) {
+ BEGIN_BATCH(2);
+ OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_VS << 16 | (2 - 2));
+ OUT_BATCH(brw->vs.base.bind_bo_offset);
+ ADVANCE_BATCH();
+ }
+ }
}
const struct brw_tracked_state brw_vs_binding_table = {
.dirty = {
.mesa = 0,
- .brw = BRW_NEW_BATCH |
- BRW_NEW_VS_CONSTBUF |
- BRW_NEW_SURFACES,
+ .brw = (BRW_NEW_BATCH |
+ BRW_NEW_STATE_BASE_ADDRESS |
+ BRW_NEW_VS_CONSTBUF |
+ BRW_NEW_SURFACES),
.cache = CACHE_NEW_VS_PROG
},
.emit = brw_vs_upload_binding_table,
@@ -111,13 +121,22 @@ const struct brw_tracked_state brw_vs_binding_table = {
static void
brw_upload_wm_binding_table(struct brw_context *brw)
{
- brw_upload_binding_table(brw, BRW_NEW_PS_BINDING_TABLE, &brw->wm.base);
+ if (brw_upload_binding_table(brw, BRW_NEW_PS_BINDING_TABLE, &brw->wm.base))
{
+ if (brw->gen >= 7) {
+ BEGIN_BATCH(2);
+ OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS << 16 | (2 - 2));
+ OUT_BATCH(brw->wm.base.bind_bo_offset);
+ ADVANCE_BATCH();
+ }
+ }
}
const struct brw_tracked_state brw_wm_binding_table = {
.dirty = {
.mesa = 0,
- .brw = BRW_NEW_BATCH | BRW_NEW_SURFACES,
+ .brw = (BRW_NEW_BATCH |
+ BRW_NEW_STATE_BASE_ADDRESS |
+ BRW_NEW_SURFACES),
.cache = CACHE_NEW_WM_PROG
},
.emit = brw_upload_wm_binding_table,
@@ -131,15 +150,23 @@ brw_gs_upload_binding_table(struct brw_context *brw)
if (brw->geometry_program == NULL)
return;
- brw_upload_binding_table(brw, BRW_NEW_GS_BINDING_TABLE, &brw->gs.base);
+ if (brw_upload_binding_table(brw, BRW_NEW_GS_BINDING_TABLE, &brw->gs.base))
{
+ if (brw->gen >= 7) {
+ BEGIN_BATCH(2);
+ OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_GS << 16 | (2 - 2));
+ OUT_BATCH(brw->gs.base.bind_bo_offset);
+ ADVANCE_BATCH();
+ }
+ }
}
const struct brw_tracked_state brw_gs_binding_table = {
.dirty = {
.mesa = 0,
- .brw = BRW_NEW_BATCH |
- BRW_NEW_GS_CONSTBUF |
- BRW_NEW_SURFACES,
+ .brw = (BRW_NEW_BATCH |
+ BRW_NEW_STATE_BASE_ADDRESS |
+ BRW_NEW_GS_CONSTBUF |
+ BRW_NEW_SURFACES),
.cache = CACHE_NEW_GS_PROG
},
.emit = brw_gs_upload_binding_table,
diff --git a/src/mesa/drivers/dri/i965/gen7_gs_state.c
b/src/mesa/drivers/dri/i965/gen7_gs_state.c
index d2ba354..92b872f 100644
--- a/src/mesa/drivers/dri/i965/gen7_gs_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_gs_state.c
@@ -66,12 +66,6 @@ upload_gs_state(struct brw_context *brw)
/* CACHE_NEW_GS_PROG */
const struct brw_vec4_prog_data *prog_data = &brw->gs.prog_data->base;
- /* BRW_NEW_GS_BINDING_TABLE */
- BEGIN_BATCH(2);
- OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_GS << 16 | (2 - 2));
- OUT_BATCH(stage_state->bind_bo_offset);
- ADVANCE_BATCH();
-
/* CACHE_NEW_SAMPLER */
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_GS << 16 | (2 - 2));
@@ -199,7 +193,6 @@ const struct brw_tracked_state gen7_gs_state = {
.mesa = _NEW_TRANSFORM | _NEW_PROGRAM_CONSTANTS,
.brw = (BRW_NEW_CONTEXT |
BRW_NEW_GEOMETRY_PROGRAM |
- BRW_NEW_GS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = CACHE_NEW_GS_PROG | CACHE_NEW_SAMPLER
diff --git a/src/mesa/drivers/dri/i965/gen7_vs_state.c
b/src/mesa/drivers/dri/i965/gen7_vs_state.c
index 1e76eb1..f6a9e0f 100644
--- a/src/mesa/drivers/dri/i965/gen7_vs_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_vs_state.c
@@ -74,12 +74,6 @@ upload_vs_state(struct brw_context *brw)
gen7_emit_vs_workaround_flush(brw);
- /* BRW_NEW_VS_BINDING_TABLE */
- BEGIN_BATCH(2);
- OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_VS << 16 | (2 - 2));
- OUT_BATCH(stage_state->bind_bo_offset);
- ADVANCE_BATCH();
-
/* CACHE_NEW_SAMPLER */
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_VS << 16 | (2 - 2));
@@ -128,7 +122,6 @@ const struct brw_tracked_state gen7_vs_state = {
.mesa = _NEW_TRANSFORM | _NEW_PROGRAM_CONSTANTS,
.brw = (BRW_NEW_CONTEXT |
BRW_NEW_VERTEX_PROGRAM |
- BRW_NEW_VS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = CACHE_NEW_VS_PROG | CACHE_NEW_SAMPLER
diff --git a/src/mesa/drivers/dri/i965/gen7_wm_state.c
b/src/mesa/drivers/dri/i965/gen7_wm_state.c
index b6561bb..586ad67 100644
--- a/src/mesa/drivers/dri/i965/gen7_wm_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_wm_state.c
@@ -143,12 +143,6 @@ upload_ps_state(struct brw_context *brw)
const int max_threads_shift = brw->is_haswell ?
HSW_PS_MAX_THREADS_SHIFT : IVB_PS_MAX_THREADS_SHIFT;
- /* BRW_NEW_PS_BINDING_TABLE */
- BEGIN_BATCH(2);
- OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS << 16 | (2 - 2));
- OUT_BATCH(brw->wm.base.bind_bo_offset);
- ADVANCE_BATCH();
-
/* CACHE_NEW_SAMPLER */
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_PS << 16 | (2 - 2));
@@ -285,7 +279,6 @@ const struct brw_tracked_state gen7_ps_state = {
_NEW_BUFFERS |
_NEW_MULTISAMPLE),
.brw = (BRW_NEW_FRAGMENT_PROGRAM |
- BRW_NEW_PS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = (CACHE_NEW_SAMPLER |
--
1.8.5.2
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
