intel/gen11+: Enable Hardware filtering of Semi-Pipelined State in WM

Initial benchmarking didn't show any performance benefits. But it might eventually.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

diff --git a/src/gallium/drivers/iris/iris_state.c b/src/gallium/drivers/iris/iris_state.c
index 1c70b2ccbff6cc433d806647327146621847bd57..4511a075ffcfd63edddd460f03185c752e4d874e 100644 (file)
--- a/src/gallium/drivers/iris/iris_state.c
+++ b/src/gallium/drivers/iris/iris_state.c
@@ -824,6 +824,15 @@ iris_init_render_context(struct iris_screen *screen,
       iris_upload_slice_hashing_state(batch);
 #endif
 
+#if GEN_GEN >= 11
+      /* WA_220160979: Enable Hardware filtering of Semi-Pipelined State in WM */
+      iris_pack_state(GENX(COMMON_SLICE_CHICKEN4), &reg_val, reg) {
+         reg.EnableHardwareFilteringinWM = true;
+         reg.EnableHardwareFilteringinWMMask = true;
+      }
+      iris_emit_lri(batch, COMMON_SLICE_CHICKEN4, reg_val);
+#endif
+
    /* 3DSTATE_DRAWING_RECTANGLE is non-pipelined, so we want to avoid
     * changing it dynamically.  We set it to the maximum size here, and
     * instead include the render target dimensions in the viewport, so

diff --git a/src/intel/vulkan/genX_state.c b/src/intel/vulkan/genX_state.c
index df76b33a7c35a24a5d95e2e933a1269b2954b2a0..06b9d497cb05e679e94d158d62c956b02dbd5d7f 100644 (file)
--- a/src/intel/vulkan/genX_state.c
+++ b/src/intel/vulkan/genX_state.c
@@ -292,6 +292,17 @@ genX(init_device_state)(struct anv_device *device)
          lri.DataDWord      = cache_mode_0;
       }
    }
+
+   /* WA_220160979: Enable Hardware filtering of Semi-Pipelined State in WM. */
+   uint32_t common_slice_chicken4;
+   anv_pack_struct(&common_slice_chicken4, GENX(COMMON_SLICE_CHICKEN4),
+                   .EnableHardwareFilteringinWM = true,
+                   .EnableHardwareFilteringinWMMask = true);
+
+   anv_batch_emit(&batch, GENX(MI_LOAD_REGISTER_IMM), lri) {
+      lri.RegisterOffset = GENX(COMMON_SLICE_CHICKEN4_num);
+      lri.DataDWord      = common_slice_chicken4;
+   }
 #endif
 
    /* Set the "CONSTANT_BUFFER Address Offset Disable" bit, so

diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index 76ec9a26a27a8f9c1e0a873b5f956edd6cd69303..5a9e77576ecd13a84a7bf56ae165a8a392002ebc 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -1660,6 +1660,10 @@ enum brw_pixel_shader_coverage_mask_mode {
 # define GLK_SCEC_BARRIER_MODE_MASK        REG_MASK(1 << 7)
 # define GEN11_STATE_CACHE_REDIRECT_TO_CS_SECTION_ENABLE (1 << 11)
 
+
+#define COMMON_SLICE_CHICKEN4                               0x7300
+# define GEN11_ENABLE_HARDWARE_FILTERING_IN_WM              (1 << 5)
+
 #define HALF_SLICE_CHICKEN7                0xE194
 # define TEXEL_OFFSET_FIX_ENABLE           (1 << 1)
 # define TEXEL_OFFSET_FIX_MASK             REG_MASK(1 << 1)

diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index 87e459376a80ed33dbb36dff487ed1e019b2327f..dfbcea586cce3cfdf49e2a3796ba026aa4b9aafe 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -189,6 +189,11 @@ brw_upload_initial_gpu_state(struct brw_context *brw)
        */
       brw_load_register_imm32(brw, GEN8_L3CNTLREG,
                               GEN8_L3CNTLREG_EDBC_NO_HANG);
+
+      /* WA_220160979: Enable Hardware filtering of Semi-Pipelined State in WM */
+      brw_load_register_imm32(brw, COMMON_SLICE_CHICKEN4,
+                              GEN11_ENABLE_HARDWARE_FILTERING_IN_WM |
+                              REG_MASK(GEN11_ENABLE_HARDWARE_FILTERING_IN_WM));
    }
 
    /* hardware specification recommends disabling repacking for