Module: Mesa
Branch: main
Commit: ed9886321cec8cb077cd33543daa887bd49b296d
URL:    
http://cgit.freedesktop.org/mesa/mesa/commit/?id=ed9886321cec8cb077cd33543daa887bd49b296d

Author: Francisco Jerez <curroje...@riseup.net>
Date:   Tue Oct 24 15:24:18 2023 -0700

intel/xehp+: Use TBIMR tile box check in order to avoid performance regressions.

This allows the hardware to behave as if TBIMR was disabled until a
polygon is processed which spans at least one tile.  This is a rather
heavy-handed heuristic meant to prevent regressions in heavily
geometry-bound workloads that render large numbers of tiny primitives
much smaller than a TBIMR tile.

A particularly bad example of this was observed in SoTR, where certain
draw calls with a long-running VS and a mostly trivial PS render more
triangles than pixels, filling up the URB and TBIMR batch pretty
quickly, which causes EU utilization to tank (since once the URB has
filled up the parallelism of the VS is limited by the number of
polygons that fit in a TBIMR batch at the completion of each tile
walk, which isn't a lot in relation to the total EU count of a DG2),
and causes the bottleneck to be the rate at which the tile sequencer
performs additional tile passes, each one processing a small number
(<1024 polygons) of the hundreds of thousands of triangles of the
draw call.

Enabling this heuristic seems effective at avoiding that scenario in
SoTR among other titles (e.g. Total War Warhammer 3), but it's a bit
of a compromise since one could imagine cases where TBIMR is helpful
even if the geometry doesn't pass the box check, so a better heuristic
or a driconf rule may be useful in the future.

Reviewed-by: Lionel Landwerlin <lionel.g.landwer...@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25493>

---

 src/gallium/drivers/iris/iris_state.c | 1 +
 src/intel/vulkan/anv_private.h        | 1 +
 src/intel/vulkan/genX_gfx_state.c     | 2 ++
 3 files changed, 4 insertions(+)

diff --git a/src/gallium/drivers/iris/iris_state.c 
b/src/gallium/drivers/iris/iris_state.c
index e0e540e12e3..206cb4f63c5 100644
--- a/src/gallium/drivers/iris/iris_state.c
+++ b/src/gallium/drivers/iris/iris_state.c
@@ -6768,6 +6768,7 @@ iris_upload_dirty_render_state(struct iris_context *ice,
             tbimr.VerticalTileCount = DIV_ROUND_UP(cso_fb->height, 
tile_height);
             tbimr.HorizontalTileCount = DIV_ROUND_UP(cso_fb->width, 
tile_width);
             tbimr.TBIMRBatchSize = util_logbase2(batch_size) - 5;
+            tbimr.TileBoxCheck = true;
          }
       }
    }
diff --git a/src/intel/vulkan/anv_private.h b/src/intel/vulkan/anv_private.h
index 3e5dfabecd4..3bb8c5918f2 100644
--- a/src/intel/vulkan/anv_private.h
+++ b/src/intel/vulkan/anv_private.h
@@ -1439,6 +1439,7 @@ struct anv_gfx_dynamic_state {
       unsigned VerticalTileCount;
       unsigned HorizontalTileCount;
       unsigned TBIMRBatchSize;
+      unsigned TileBoxCheck;
    } tbimr;
    bool use_tbimr;
 
diff --git a/src/intel/vulkan/genX_gfx_state.c 
b/src/intel/vulkan/genX_gfx_state.c
index c3d791268cd..d2cd30dfbe0 100644
--- a/src/intel/vulkan/genX_gfx_state.c
+++ b/src/intel/vulkan/genX_gfx_state.c
@@ -1280,6 +1280,7 @@ genX(cmd_buffer_flush_gfx_runtime_state)(struct 
anv_cmd_buffer *cmd_buffer)
              DIV_ROUND_UP(fb_width, tile_width));
          SET(TBIMR_TILE_PASS_INFO, tbimr.TBIMRBatchSize,
              util_logbase2(batch_size) - 5);
+         SET(TBIMR_TILE_PASS_INFO, tbimr.TileBoxCheck, true);
          SET(TBIMR_TILE_PASS_INFO, use_tbimr, true);
       } else {
          hw_state->use_tbimr = false;
@@ -1891,6 +1892,7 @@ genX(cmd_buffer_flush_gfx_hw_state)(struct anv_cmd_buffer 
*cmd_buffer)
          SET(tbimr, tbimr, VerticalTileCount);
          SET(tbimr, tbimr, HorizontalTileCount);
          SET(tbimr, tbimr, TBIMRBatchSize);
+         SET(tbimr, tbimr, TileBoxCheck);
       }
    }
 #endif

Reply via email to