i965: Shrink Gen5 VUE map layout to be the same as Gen4.

The PRM suggests a larger layout, mostly to support having
gl_ClipDistance[] somewhere predictable for the fixed-function clipper
-- but it didn't actually arrive in Gen5.

Just use the same layout for both Gen4 and Gen5.

No Piglit regressions.

Improves performance in CS:S Video Stress Test by ~3%.

V2: - Remove now-useless function for determining the SF URB read offset
    - Remove now-unused BRW_VARYING_SLOT_POS_DUPLICATE

Signed-off-by: Chris Forbes <[email protected]>
Reviewed-by: Paul Berry <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>

6 files changed, 7 insertions, 40 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index ae6f81ae6c4..0db1a1802b8 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -333,7 +333,6 @@ struct brw_wm_prog_data {
 typedef enum
 {
    BRW_VARYING_SLOT_NDC = VARYING_SLOT_MAX,
-   BRW_VARYING_SLOT_POS_DUPLICATE,
    BRW_VARYING_SLOT_PAD,
    /**
     * Technically this is not a varying but just a placeholder that
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index 5870b60d38e..ba8782b6d0c 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -76,7 +76,7 @@ static void compile_sf_prog( struct brw_context *brw,
       c.vue_map.varying_to_slot[BRW_VARYING_SLOT_PNTC] = c.vue_map.num_slots;
       c.vue_map.slot_to_varying[c.vue_map.num_slots++] = BRW_VARYING_SLOT_PNTC;
    }
-   c.urb_entry_read_offset = brw_sf_compute_urb_entry_read_offset(intel);
+   c.urb_entry_read_offset = BRW_SF_URB_ENTRY_READ_OFFSET;
    c.nr_attr_regs = (c.vue_map.num_slots + 1)/2 - c.urb_entry_read_offset;
    c.nr_setup_regs = c.nr_attr_regs;
 
diff --git a/src/mesa/drivers/dri/i965/brw_sf.h b/src/mesa/drivers/dri/i965/brw_sf.h
index f908fc0667b..caeb0d06b1c 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.h
+++ b/src/mesa/drivers/dri/i965/brw_sf.h
@@ -103,6 +103,7 @@ void brw_emit_line_setup( struct brw_sf_compile *c, bool allocate );
 void brw_emit_point_setup( struct brw_sf_compile *c, bool allocate );
 void brw_emit_point_sprite_setup( struct brw_sf_compile *c, bool allocate );
 void brw_emit_anyprim_setup( struct brw_sf_compile *c );
-int brw_sf_compute_urb_entry_read_offset(struct intel_context *intel);
+
+#define BRW_SF_URB_ENTRY_READ_OFFSET 1
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index 7c29ba27d1a..4b5e7cc93f1 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -124,19 +124,6 @@ const struct brw_tracked_state brw_sf_vp = {
    .emit = upload_sf_vp
 };
 
-/**
- * Compute the offset within the URB (expressed in 256-bit register
- * increments) that should be used to read the VUE in th efragment shader.
- */
-int
-brw_sf_compute_urb_entry_read_offset(struct intel_context *intel)
-{
-   if (intel->gen == 5)
-      return 3;
-   else
-      return 1;
-}
-
 static void upload_sf_unit( struct brw_context *brw )
 {
    struct intel_context *intel = &brw->intel;
@@ -163,9 +150,7 @@ static void upload_sf_unit( struct brw_context *brw )
    sf->thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
 
    sf->thread3.dispatch_grf_start_reg = 3;
-
-   sf->thread3.urb_entry_read_offset =
-      brw_sf_compute_urb_entry_read_offset(intel);
+   sf->thread3.urb_entry_read_offset = BRW_SF_URB_ENTRY_READ_OFFSET;
 
    /* CACHE_NEW_SF_PROG */
    sf->thread3.urb_entry_read_length = brw->sf.prog_data->urb_read_length;
diff --git a/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp b/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
index 02ba603d018..162fd55f429 100644
--- a/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
@@ -2746,7 +2746,6 @@ vec4_visitor::emit_urb_slot(int mrf, int varying)
       current_annotation = "NDC";
       emit(MOV(reg, src_reg(output_reg[BRW_VARYING_SLOT_NDC])));
       break;
-   case BRW_VARYING_SLOT_POS_DUPLICATE:
    case VARYING_SLOT_POS:
       current_annotation = "gl_Position";
       emit(MOV(reg, src_reg(output_reg[VARYING_SLOT_POS])));
diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index 720325dec5c..d173d2e31b8 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -85,34 +85,17 @@ brw_compute_vue_map(struct brw_context *brw, struct brw_vue_map *vue_map,
     */
    switch (intel->gen) {
    case 4:
+   case 5:
       /* There are 8 dwords in VUE header pre-Ironlake:
        * dword 0-3 is indices, point width, clip flags.
        * dword 4-7 is ndc position
        * dword 8-11 is the first vertex data.
-       */
-      assign_vue_slot(vue_map, VARYING_SLOT_PSIZ);
-      assign_vue_slot(vue_map, BRW_VARYING_SLOT_NDC);
-      assign_vue_slot(vue_map, VARYING_SLOT_POS);
-      break;
-   case 5:
-      /* There are 20 DWs (D0-D19) in VUE header on Ironlake:
-       * dword 0-3 of the header is indices, point width, clip flags.
-       * dword 4-7 is the ndc position
-       * dword 8-11 of the vertex header is the 4D space position
-       * dword 12-19 of the vertex header is the user clip distance.
-       * dword 20-23 is a pad so that the vertex element data is aligned
-       * dword 24-27 is the first vertex data we fill.
        *
-       * Note: future pipeline stages expect 4D space position to be
-       * contiguous with the other varyings, so we make dword 24-27 a
-       * duplicate copy of the 4D space position.
+       * On Ironlake the VUE header is nominally 20 dwords, but the hardware
+       * will accept the same header layout as Gen4 [and should be a bit faster]
        */
       assign_vue_slot(vue_map, VARYING_SLOT_PSIZ);
       assign_vue_slot(vue_map, BRW_VARYING_SLOT_NDC);
-      assign_vue_slot(vue_map, BRW_VARYING_SLOT_POS_DUPLICATE);
-      assign_vue_slot(vue_map, VARYING_SLOT_CLIP_DIST0);
-      assign_vue_slot(vue_map, VARYING_SLOT_CLIP_DIST1);
-      assign_vue_slot(vue_map, BRW_VARYING_SLOT_PAD);
       assign_vue_slot(vue_map, VARYING_SLOT_POS);
       break;
    case 6: