fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-03 22:58:31 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

freedreno/regs: A6XX_EARLY_LRZ_LATE_Z is really A6XX_EARLY_Z_LATE_Z

Browse source 
By observing prop driver and doing some tests this mode is about
both early_z and early lrz.

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34973>
This commit is contained in:
Danylo Piliaiev 2025-05-12 16:03:09 +02:00 • committed by Marge Bot
parent cd4c0bc58e
commit de6d111d73
4 changed files with 25 additions and 41 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

42
src/freedreno/registers/adreno/a6xx.xml
View file

@ -1166,36 +1166,20 @@ to upconvert to 32b float internally?
<doc>Disable early z-test and early-lrz test (if applicable)</doc>
<value value="0x1" name="A6XX_LATE_Z"/>
<doc>
A special mode that allows early-lrz test but disables
early-z test. Which might sound a bit funny, since
lrz-test happens before z-test. But as long as a couple
conditions are maintained this allows using lrz-test in
cases where fragment shader has kill/discard:
A special mode that allows early-lrz (if applicable) or early-z
tests, but also does late-z tests at which point it writes depth.
1) Disable lrz-write in cases where it is uncertain during
binning pass that a fragment will pass. Ie. if frag
shader has-kill, writes-z, or alpha/stencil test is
enabled. (For correctness, lrz-write must be disabled
when blend is enabled.) This is analogous to how a
z-prepass works.
This mode is used when fragment can be killed (via discard or
sample mask) after early-z tests and it writes depth. In such case
depth can be written only at late-z stage, but it's ok to use
early-z to discard fragments.
2) Disable lrz-write and test if a depth-test direction
reversal is detected. Due to condition (1), the contents
of the lrz buffer are a conservative estimation of the
depth buffer during the draw pass. Meaning that geometry
that we know for certain will not be visible will not pass
lrz-test. But geometry which may be (or contributes to
blend) will pass the lrz-test.
This allows us to keep early-lrz-test in cases where the frag
shader does not write-z (ie. we know the z-value before FS)
and does not have side-effects (image/ssbo writes, etc), but
does have kill/discard. Which turns out to be a common
enough case that it is useful to keep early-lrz test against
the conservative lrz buffer to discard fragments that we
know will definitely not be visible.
However this mode is not compatible with:
- Lack of D/S attachment
- Stencil writes on stencil or depth test failures
- Per-sample shading
</doc>
<value value="0x2" name="A6XX_EARLY_LRZ_LATE_Z"/>
<value value="0x2" name="A6XX_EARLY_Z_LATE_Z"/>
<doc>Not a real hw value, used internally by mesa</doc>
<value value="0x3" name="A6XX_INVALID_ZTEST"/>
</enum>
@ -3233,9 +3217,9 @@ to upconvert to 32b float internally?
<enum name="a6xx_lrz_feedback_mask">
<value value="0x0" name="LRZ_FEEDBACK_NONE"/>
<value value="0x1" name="LRZ_FEEDBACK_EARLY_Z"/>
<value value="0x2" name="LRZ_FEEDBACK_EARLY_LRZ_LATE_Z"/>
<value value="0x2" name="LRZ_FEEDBACK_EARLY_Z_LATE_Z"/>
<!-- We don't have a flag type and this flags combination is often used -->
<value value="0x3" name="LRZ_FEEDBACK_EARLY_Z_OR_EARLY_LRZ_LATE_Z"/>
<value value="0x3" name="LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z"/>
<value value="0x4" name="LRZ_FEEDBACK_LATE_Z"/>
</enum>

12
src/freedreno/vulkan/tu_cmd_buffer.cc
View file

@ -1251,8 +1251,8 @@ tu6_emit_tile_select(struct tu_cmd_buffer *cmd,
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask =
phys_dev->info->a6xx.has_lrz_feedback
? (hw_binning ? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_LRZ_LATE_Z :
LRZ_FEEDBACK_EARLY_LRZ_LATE_Z)
? (hw_binning ? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z :
LRZ_FEEDBACK_EARLY_Z_LATE_Z)
: LRZ_FEEDBACK_NONE,
});
@ -2346,7 +2346,7 @@ tu6_sysmem_render_begin(struct tu_cmd_buffer *cmd, struct tu_cs *cs,
.buffers_location = BUFFERS_IN_SYSMEM,
.lrz_feedback_zmode_mask =
cmd->device->physical_device->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE,
});
@ -2452,7 +2452,7 @@ tu6_tile_render_begin(struct tu_cmd_buffer *cmd, struct tu_cs *cs,
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask =
phys_dev->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE
});
@ -6215,7 +6215,7 @@ tu6_build_depth_plane_z_mode(struct tu_cmd_buffer *cmd, struct tu_cs *cs)
tu_fs_reads_dynamic_ds_input_attachment(cmd, fs)) &&
(depth_write || stencil_write)) {
zmode = (cmd->state.lrz.valid && cmd->state.lrz.enabled)
? A6XX_EARLY_LRZ_LATE_Z
? A6XX_EARLY_Z_LATE_Z
: A6XX_LATE_Z;
}
@ -6234,7 +6234,7 @@ tu6_build_depth_plane_z_mode(struct tu_cmd_buffer *cmd, struct tu_cs *cs)
*/
if (!force_late_z && cmd->state.lrz.enabled && fs->variant->writes_pos &&
zmode != A6XX_LATE_Z) {
zmode = A6XX_EARLY_LRZ_LATE_Z;
zmode = A6XX_EARLY_Z_LATE_Z;
}
if ((force_late_z && !fs->variant->fs.early_fragment_tests) ||

2
src/gallium/drivers/freedreno/a6xx/fd6_emit.cc
View file

@ -98,7 +98,7 @@ compute_ztest_mode(struct fd6_emit *emit, bool lrz_valid) assert_dt
* that could discard samples that would otherwise only be
* hidden by a later draw.
*/
return lrz_valid ? A6XX_EARLY_LRZ_LATE_Z : A6XX_LATE_Z;
return lrz_valid ? A6XX_EARLY_Z_LATE_Z : A6XX_LATE_Z;
} else {
return A6XX_EARLY_Z;
}

10
src/gallium/drivers/freedreno/a6xx/fd6_gmem.cc
View file

@ -1177,7 +1177,7 @@ fd6_emit_tile_init(struct fd_batch *batch) assert_dt
.force_lrz_write_dis = !screen->info->a6xx.has_lrz_feedback,
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask = screen->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE,
});
@ -1208,7 +1208,7 @@ fd6_emit_tile_init(struct fd_batch *batch) assert_dt
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask =
screen->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE,
});
}
@ -1305,7 +1305,7 @@ fd6_emit_tile_prep(struct fd_batch *batch, const struct fd_tile *tile)
.force_lrz_write_dis = !screen->info->a6xx.has_lrz_feedback,
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask = screen->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE,
});
@ -1332,7 +1332,7 @@ fd6_emit_tile_prep(struct fd_batch *batch, const struct fd_tile *tile)
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask =
screen->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_OR_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE,
});
}
@ -1344,7 +1344,7 @@ fd6_emit_tile_prep(struct fd_batch *batch, const struct fd_tile *tile)
.force_lrz_write_dis = !ctx->screen->info->a6xx.has_lrz_feedback,
.buffers_location = BUFFERS_IN_GMEM,
.lrz_feedback_zmode_mask = ctx->screen->info->a6xx.has_lrz_feedback
? LRZ_FEEDBACK_EARLY_LRZ_LATE_Z
? LRZ_FEEDBACK_EARLY_Z_LATE_Z
: LRZ_FEEDBACK_NONE,
});