i965/fs: Lower 32x32 bit multiplication on BXT.

[mesa.git] / src / mesa / drivers / dri / i965 / brw_clear.c

diff --git a/src/mesa/drivers/dri/i965/brw_clear.c b/src/mesa/drivers/dri/i965/brw_clear.c
index d9a879239ec6329c0914dfa3cd23e90207c01129..f981388ef1a2d76eee1324c22ac78e822ec89ec3 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_clear.c
+++ b/src/mesa/drivers/dri/i965/brw_clear.c
@@ -36,7 +36,6 @@
 #include "intel_blit.h"
 #include "intel_fbo.h"
 #include "intel_mipmap_tree.h"
-#include "intel_regions.h"
 
 #include "brw_context.h"
 #include "brw_blorp.h"
@@ -111,7 +110,7 @@ brw_fast_clear_depth(struct gl_context *ctx)
    struct intel_mipmap_tree *mt = depth_irb->mt;
    struct gl_renderbuffer_attachment *depth_att = &fb->Attachment[BUFFER_DEPTH];
 
-   if (brw->gen < 6 || brw->gen >= 8)
+   if (brw->gen < 6)
       return false;
 
    if (!intel_renderbuffer_has_hiz(depth_irb))
@@ -122,8 +121,9 @@ brw_fast_clear_depth(struct gl_context *ctx)
     * first.
     */
    if ((ctx->Scissor.EnableFlags & 1) && !noop_scissor(ctx, fb)) {
-      perf_debug("Failed to fast clear depth due to scissor being enabled.  "
-                 "Possible 5%% performance win if avoided.\n");
+      perf_debug("Failed to fast clear %dx%d depth because of scissors.  "
+                 "Possible 5%% performance win if avoided.\n",
+                 mt->logical_width0, mt->logical_height0);
       return false;
    }
 
@@ -155,8 +155,9 @@ brw_fast_clear_depth(struct gl_context *ctx)
        *        width of the map (LOD0) is not multiple of 16, fast clear
        *        optimization must be disabled.
        */
-      if (brw->gen == 6 && (minify(mt->physical_width0,
-                                   depth_irb->mt_level) % 16) != 0)
+      if (brw->gen == 6 &&
+          (minify(mt->physical_width0,
+                  depth_irb->mt_level - mt->first_level) % 16) != 0)
         return false;
       /* FALLTHROUGH */
 
@@ -183,12 +184,12 @@ brw_fast_clear_depth(struct gl_context *ctx)
     *      must be issued before the rectangle primitive used for the depth
     *      buffer clear operation.
     */
-   intel_batchbuffer_emit_mi_flush(brw);
+   brw_emit_mi_flush(brw);
 
    if (fb->MaxNumLayers > 0) {
-      unsigned num_layers = depth_irb->mt->level[depth_irb->mt_level].depth;
-      for (unsigned layer = 0; layer < num_layers; layer++) {
-         intel_hiz_exec(brw, mt, depth_irb->mt_level, layer,
+      for (unsigned layer = 0; layer < depth_irb->layer_count; layer++) {
+         intel_hiz_exec(brw, mt, depth_irb->mt_level,
+                        depth_irb->mt_layer + layer,
                         GEN6_HIZ_OP_DEPTH_CLEAR);
       }
    } else {
@@ -203,7 +204,7 @@ brw_fast_clear_depth(struct gl_context *ctx)
        *      by a PIPE_CONTROL command with DEPTH_STALL bit set and Then
        *      followed by Depth FLUSH'
       */
-      intel_batchbuffer_emit_mi_flush(brw);
+      brw_emit_mi_flush(brw);
    }
 
    /* Now, the HiZ buffer contains data that needs to be resolved to the depth
@@ -241,13 +242,11 @@ brw_clear(struct gl_context *ctx, GLbitfield mask)
       }
    }
 
-   /* BLORP is currently only supported on Gen6+. */
-   if (brw->gen >= 6 && brw->gen < 8) {
-      if (mask & BUFFER_BITS_COLOR) {
-         if (brw_blorp_clear_color(brw, fb, partial_clear)) {
-            debug_mask("blorp color", mask & BUFFER_BITS_COLOR);
-            mask &= ~BUFFER_BITS_COLOR;
-         }
+   /* Clear color buffers with fast clear or at least rep16 writes. */
+   if (brw->gen >= 6 && brw->gen < 9 && (mask & BUFFER_BITS_COLOR)) {
+      if (brw_meta_fast_clear(brw, fb, mask, partial_clear)) {
+         debug_mask("blorp color", mask & BUFFER_BITS_COLOR);
+         mask &= ~BUFFER_BITS_COLOR;
       }
    }