[Git][xorg-team/lib/mesa][debian-unstable] 114 commits: docs: Extend calendar entries for 23.2 by 2 releases.

Timo Aaltonen pushed to branch debian-unstable at X Strike Force / lib / mesa

Commits:

6fb68b6b

by Dylan Baker at 2023-09-05T10:10:59-07:00

docs: Extend calendar entries for 23.2 by 2 releases.

ad234040
by Dylan Baker at 2023-09-05T10:11:11-07:00
```
docs: update calendar for 23.2.0-rc3
```

7270fe95

by Dylan Baker at 2023-09-22T08:46:37-07:00

.pick_status.json: Update to 10e75aae1bddee9795b1ff04ffd656b0da79b5b5

69276276

by Dylan Baker at 2023-09-22T08:46:37-07:00

.pick_status.json: Updates notes for aebe58458611e0bb585a5bce8e16c1175783f3cc

854c11ca

by Rhys Perry at 2023-09-22T08:46:37-07:00

radv: disable 64-bit color attachments

These work in some circumstances (dEQP-VK.spirv_assembly.instruction.graphics.16bit_storage.input_output_float_16_to_64.scalar9_tessc),
but I'm not sure if they work in all, blending certainly doesn't work and
this probably wasn't intended in the first place.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Fixes: 01bd012edd20 ("amd: fix 64-bit integer color image clears")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24400>
(cherry picked from commit 405f3bf9903f8a8b65da113adc9ef32e02b84fa2)

95301ca3

by Faith Ekstrand at 2023-09-22T08:46:38-07:00

nir: Handle nir_op_mov properly in opt_shrink_vectors

If the opcode is a mov, it falls into the nir_alu_src_is_trivial_ssa
case, not the vec case.

Fixes: 94eff7ccd866 ("nir: shrink phi nodes in nir_opt_shrink_vectors")
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24704>
(cherry picked from commit f9a17c6fef92b1316c9a06465712e094f33add68)

3de51743

by Faith Ekstrand at 2023-09-22T08:46:38-07:00

nir: Don't handle nir_op_mov in get_undef_mask in opt_undef

It's unnecessary because earlier parts of the pass will ensure that a
mov of undef is turned into an undef.  It's also wrong because
nir_op_mov has different semantics from nir_op_vecN when it comes to how
sources map to destination components.

Fixes: 5f26c21e6246 ("nir: Expand opt_undef to handle undef channels in a store intrinsic")
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24704>
(cherry picked from commit 408929289aadaf44e57ef60cb7c1ace4958199c7)

fa023847

by Mike Blumenkrantz at 2023-09-22T08:46:38-07:00

vk/graphics: fix CWE handling with DS3

VkPipelineColorBlendStateCreateInfo::attachmentCount cannot be used to
generate the CWE mask since it cannot be read if enough dynamic state is in use

instead just pass the max mask and let drivers figure it out

cc: mesa-stable

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24673>
(cherry picked from commit 0fb9064231687ff7af4cb48697ee24b48715b80f)

356d59d7

by Karol Herbst at 2023-09-22T08:46:38-07:00

nouveau: take glsl_type ref unconditionally

Calling into tgsi_to_nir requires it, which we are running into with vdpau
and potential other state-trackers still handing us TGSIs over.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9481
Fixes: 5889c13fcd4 ("nv50,nvc0: Use ttn for tgsi shaders by default")
Signed-off-by: Karol Herbst <git@karolherbst.de>
Reviewed-by: M Henning <drawoc@darkrefraction.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24740>
(cherry picked from commit 91029b7e87bba2a28aeb202cdf920d6ec6a70fa7)

fb774d1d

by Mike Blumenkrantz at 2023-09-22T08:46:38-07:00

Revert "vk/wsi/x11: handle geometry updating more asynchronously"

This reverts commit 36d5b58317179e5db32800743ef0faed4655b9fb.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24742>
(cherry picked from commit 8f3499bafcd32171d72bafc7ca8c3381d716aad5)

a75c0b09

by Emma Anholt at 2023-09-22T08:46:38-07:00

disk_cache: Disable the "List" test for RO disk cache.

It uses a poll function that waits for a second hoping for another thread
to catch up, which is not a reliable way to do synchronization.  The test
has been spuriously failing merges on a regular basis recently.

This is issue #9222, which I'm leaving open until the author can fix the test.

Fixes: 3b69b67545b6 ("util/fossilize_db: add runtime RO foz db loading via FOZ_DBS_DYNAMIC_LIST")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24755>
(cherry picked from commit 4dfd306454a4653a8c0b0a37a3e6bb1632abfc9e)

fbfcab3a

by Karol Herbst at 2023-09-22T08:46:38-07:00

nv50: limit max code uploads to 0x8000

I have no idea why a bigger size doesn't work, the hardware doesn't
complain, but it turns out that uploading big shaders still causes issues
with the old limit. *shrug*

Fixes: 7f63d2ebdbc ("nv50: fix code uploads bigger than 0x10000 bytes")
Signed-off-by: Karol Herbst <git@karolherbst.de>
Reviewed-by: M Henning <drawoc@darkrefraction.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24758>
(cherry picked from commit 3e0cd6dfb9b5c87ad18c3576017c6cc5c7f240a0)

6ba01077

by Sviatoslav Peleshko at 2023-09-22T08:46:38-07:00

dri: Use RGB internal formats for RGBX formats

These formats do not contain alpha channel, so their internal formats
should reflect that.

Fixes: bf576772 ("dri_util: add driImageFormatToSizedInternalGLFormat function")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9429
Signed-off-by: Sviatoslav Peleshko <sviatoslav.peleshko@globallogic.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24580>
(cherry picked from commit e62f2c48a0836495381b38b3ab055b0787545de4)

8ed55a67

by Faith Ekstrand at 2023-09-22T08:46:38-07:00

nir: Fix metadata in nir_lower_is_helper_invocation

It does not preserve everything.  It adds and removes instructions and
even adds a variable.

Fixes: f17b41ab4f01 ("nir: add lowering pass for helperInvocationEXT()")
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24760>
(cherry picked from commit 831085afa3f68cef851dca7e50f979a375dab548)

ea6d27e4

by Pavel Ondračka at 2023-09-22T08:46:38-07:00

r300: don't abort on flow control when using draw for vs

It can handle it just fine. Around 250 dEQPs go from Skip to Pass.

Fixes: 1021e2b946b18739b65b575b0770e6158ba05592
Reviewed-by: Filip Gawin <filip.gawin@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24769>
(cherry picked from commit 8657a09f9a618102d1eaad503a6c1a63d38dcb02)

055cdd28

by Eric Engestrom at 2023-09-22T08:46:38-07:00

vc4: drop duplicate .lower_ldexp

It's set 3 lines above already.

Fixes: 2a33ea95d66cd7ba83e2 ("glsl: Retire ldexp lowering in favor of the nir lowering flag.")
Signed-off-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24786>
(cherry picked from commit af28356de06d7b5adfa3fa87dac157021c3a68a9)

d28bae9c

by Eric Engestrom at 2023-09-22T08:46:38-07:00

zink: fix format in zink_make_{image,texture}_handle_resident()

`ds->db.format` is a `pipe_format`, while `buffer_infos[handle].format` is
a `VkFormat`; the conversion from one to the other was missing.

Fixes: 99ba529feed6f9917a44 ("zink: implement descriptor buffer handling of bindless texture")
Signed-off-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24790>
(cherry picked from commit ab0f0d1563365ea29175bd0b07ba34e1f96c9999)

65d3635f

by Alyssa Rosenzweig at 2023-09-22T08:46:38-07:00

nir/lower_helper_writes: Consider bindless images

These need to be handled like other image ops.

Fixes KHR-GLES31.core.shader_image_load_store.basic-allTargets-atomicFS on Asahi
with bindless image access forced.

Fixes: 586da7b3290 ("nir: Add nir_lower_helper_writes pass")
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Italo Nicola <italonicola@collabora.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24778>
(cherry picked from commit 04ba4059b7243fb9258df12154ce1e7a01e019c8)

fcbd2579

by Dylan Baker at 2023-09-22T08:46:38-07:00

.pick_status.json: Updates notes for f8cb0d8a44afb9c70f38e359ffe0ad57416e66a4

4a6ece1a

by Dylan Baker at 2023-09-22T08:46:38-07:00

Revert "Revert "intel/ci: disable iris-jsl-deqp because it always fails for an AMD MR""

This reverts commit dd98f9876498f80a34dc9dbd954385cc1a8bf267.

6de9ef12

by Alyssa Rosenzweig at 2023-09-22T08:46:38-07:00

nir/passthrough_gs: Fix array size

Triangle strips with adjacency have 6 vertices input, so we need an array big
enough for all 6 vertices to avoid overflow. Fixes passthrough GS generated for
KHR-GLES31.core.draw_indirect.basic-mode-*-triangle*adj*.

Fixes: ea14579f3dc ("nir: handle primitives with adjacency")
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Antonino Maniscalco <antonino.maniscalco@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24780>
(cherry picked from commit 558e36f64196ec734e4c7502db491f56f76118cd)

4b3549b1

by Eric Engestrom at 2023-09-22T08:46:38-07:00

v3dv: fix VK_PIPELINE_ROBUSTNESS_{BUFFER,IMAGE}_BEHAVIOR_DEVICE_DEFAULT_EXT copy/paste typo

Fixes: 24d9a80247605ac2c237 ("v3dv: implement VK_EXT_pipeline_robustness")
Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24787>
(cherry picked from commit 4dc0cb1ebea1cc8f9e1b46bb38dd21cdf2643437)

c68687e3

by Eric Engestrom at 2023-09-22T08:46:38-07:00

v3dv: fix copy/pasted type of `sample`

And use the type in functions instead of a generic `uint32_t` to make it
easier to notice the wrong type.

Fixes: 47e02a2ef11237cf4c2c ("v3dv: add a fast path for vkCmdClearAttachments")
Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24787>
(cherry picked from commit ae0cb0b998fb772fc76226d9b85b21c1fadea448)

c8a08be8

by Eric Engestrom at 2023-09-22T08:46:38-07:00

v3dv: fix shader stage name in error message

Fixes: 60145629a2bdcd4e7835 ("v3dv: initial CreateGraphicsPipeline/DestroyPipeline implementation")
Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24787>
(cherry picked from commit 8a1f3d0d73f0908d679bf8a97bafcb2dbc4661fd)

cb95664d

by Eric Engestrom at 2023-09-22T08:46:38-07:00

v3d/qpu: fix type of function argument

Fixes: 05c7d9715b8a419fd6fb ("broadcom: Add V3D 3.3 QPU instruction pack, unpack, and disasm.")
Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24787>
(cherry picked from commit 953ac9da7949a7d2333680aa77ff20288a247f19)

8084345d

by Marek Olšák at 2023-09-22T08:46:38-07:00

Revert "ac: don't call ac_query_pci_bus_info from ac_query_gpu_info"

This reverts commit a48642400be86a17d460ce2786988ec67f423f01.

Instead, add a new parameter require_pci_bus_info to control the behavior.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24759>
(cherry picked from commit 5d19a0a19ba2ffe50a7d05fcf0ff17e0078c2d8e)

657c8abc

by Samuel Pitoiset at 2023-09-22T08:46:38-07:00

Revert "radv/amdgpu: workaround a kernel bug when replacing sparse mappings"

This workaround was added temporarily but it can actually cause
stuttering in some games like Forza Horizon 5.

The kernel fix
(https://lists.freedesktop.org/archives/amd-gfx/2023-June/094648.html)
landed in some stable kernels (5.15.121+, 6.1.40+ and 6.4.5+). Sadly,
older stable kernels don't have it, so you might experiment random GPU
hangs in games that use sparse mapping. Please ensure your kernel is
up-to-date for the best experience.

This reverts commit 9b00867327c2b266fcdebcef8bc7e7497eaab06b.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9443
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24774>
(cherry picked from commit f67eb9ce07e6b19fa5cae6f14551094bf236765b)

00415298

by Samuel Pitoiset at 2023-09-22T08:46:38-07:00

Revert "radv/amdgpu: skip adding per VM BOs for sparse during CS BO list build"

This reverts commit 51caece74cd54e99a5f89521e8feacd894781ca2.

Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24774>
(cherry picked from commit e3fae0173005801ba39d50de3944bc6113e2a624)

c71b07bc

by Tatsuyuki Ishi at 2023-09-22T08:46:38-07:00

radv/amdgpu: Do not pass in a BO handle when clearing PRT VA region.

This field is invalid to access for virtual BOs.

Fixes: a931d5a4a4d ("radv/winsys: clear the PRT VA range when destroying a virtual BO")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24805>
(cherry picked from commit 6c5512568b18e5c16f695c9d4d2dd862dc3c54f8)

f36e5c85

by Mike Blumenkrantz at 2023-09-22T08:46:38-07:00

zink: wait on async fence during ctx program removal

removed=true implies that no async jobs are outstanding

fixes #9580

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24811>
(cherry picked from commit ca987c0dfb97d22d3cf2632f38d9d749162767b0)

9ddd8fff

by Chia-I Wu at 2023-09-22T08:46:38-07:00

ac/surface: limit RADEON_SURF_NO_TEXTURE to color surfaces

For z surfaces, flags.texture should be based on
RADEON_SURF_TC_COMPATIBLE_HTILE alone.  Otherwise, addrlib could pick a
_X/_T swizzle mode for a MSAA depth texture, which is said to be broken:

  When _X/_T swizzle mode was used for MSAA depth texture, TC will get zplane
  equation from wrong address within memory range a tile covered and use the
  garbage data for compressed Z reading which finally leads to corruption.

Fixes: de0885cdb89 ("amd/surface: add RADEON_SURF_NO_TEXTURE flag")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24767>
(cherry picked from commit e74c3dbb70706f1a2bcf863b5058295fcd456ae4)

ab351825

by Friedrich Vock at 2023-09-22T08:46:38-07:00

nir/load_store_vectorize: Handle intrinsics with constant base

This includes nir_load_stack and nir_store_stack, which are vectorized
in nir_lower_shader_calls. If not adjusted, we end up loading from
the wrong base.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9596
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9587
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24789>
(cherry picked from commit a28ff7f2405a4ddd0bcc309af252520e342b7ffe)

23f0ecc3

by Sagar Ghuge at 2023-09-22T08:46:38-07:00

blorp: Drop unnecessary assertions in blorp_can_hiz_clear_depth

We already checks for the alignment and the multislice surface, we don't
need to add assertions around those two.

fixes: 37fcbb375cb ("blorp: Disable unaligned partial HIZ fast clears for HIZ_CCS too")
closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9684

Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Tested-by: Mark Janes <markjanes@swizzler.org>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24837>
(cherry picked from commit 839b03cc06d63739b702e00a6f29472e7a0c058a)

32763470

by Mike Blumenkrantz at 2023-09-22T08:46:38-07:00

zink: don't start multiple cache jobs for the same program

if there's already a cache job in flight then starting a second one
is illegal

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24831>
(cherry picked from commit d5157356ceffad6161ccc773cd6a3a9219f6b4e0)

e350c853

by Rhys Perry at 2023-09-22T08:46:38-07:00

aco: fix p_bpermute_gfx6 with input at non-zero byte

Same as the other bpermute pseudo instructions.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24693>
(cherry picked from commit 85957dd6e5572dfe424f6fca2ca7a04e95084191)

f49e1ecc

by Georg Lehmann at 2023-09-22T08:46:38-07:00

aco: fix u2f16 with 32bit input

The vulkan spec says all conversions are correctly rounded, so if the input
is larger than the largest fp16 value, we need to return MAX_FLOAT/inf
instead of cutting off the msbs.

Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24826>
(cherry picked from commit 6d949e18fd514ef386a9fce6710ca8e275778ddd)

d9eb323a

by Dylan Baker at 2023-09-22T08:46:38-07:00

.pick_status.json: Updates notes for 93b4f200dead198e680991a1e95bf3d3b58f87bd

365d3dbe

by Tapani Pälli at 2023-09-22T08:46:38-07:00

mesa: fix some TexParameter and SamplerParameter cases

EXT extension was added without tests so these functions did
not work properly.

Fixes: 799710be889 ("mesa: Add EXT_texture_mirror_clamp_to_edge to extension table")
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24845>
(cherry picked from commit d65fe6eff1fbba390d885c84630a96be37ea76a0)

f3789c8c

by Dylan Baker at 2023-09-22T08:46:39-07:00

.pick_status.json: Updates notes for 7e246f7f2bde0c859269c4b81505bd0887045e7b

3cfc1274

by Dylan Baker at 2023-09-22T08:46:39-07:00

.pick_status.json: Updates notes for 9865e5dff49395543da4331a943ba5a03ce6a413

a02429d4

by Rhys Perry at 2023-09-22T08:46:39-07:00

radv: fix 128bpp comp-to-single clears

We were clearing GB to A, instead of R.

This fixes some red tinting in Overwatch 2 when shadow quality is set to
"Ultra".

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Fixes: 7451eb1d6112 ("radv: implement DCC fast clears with comp-to-single")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9446
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24400>
(cherry picked from commit e2c7ce37192cef693c7a228b997402d480674a0a)

6b99242d

by Marcin Ślusarz at 2023-09-22T08:46:39-07:00

intel/compiler/mesh: compactify MUE layout

Instead of using 4 dwords for each output slot, use only the amount
of memory actually needed by each variable.

There are some complications from this "obvious" idea:
- flat and non-flat variables can't be merged into the same vec4 slot,
  because flat inputs mask has vec4 stride
- multi-slot variables can have different layout:
   float[N] requires N 1-dword slots, but
   i64vec3 requires 1 fully occupied 4-dword slot followed by 2-dword slot
- some output variables occur both in single-channel/component split
  and combined variants
- crossing vec4 boundary requires generating more writes, so avoiding them
  if possible is beneficial

This patch fixes some issues with arrays in per-vertex and per-primitive data
(func.mesh.ext.outputs.*.indirect_array.q0 in crucible)
and by reduction in single MUE size it allows spawning more threads at
the same time.

Note: this patch doesn't improve vk_meshlet_cadscene performance because
default layout is already optimal enough.

Reviewed-by: Ivan Briano <ivan.briano@intel.com>
(cherry picked from commit a2521233631b72e7d9131243818e68fd911b94c6)

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25188>

03fc9642

by Marcin Ślusarz at 2023-09-22T08:46:39-07:00

intel/compiler,anv: put some vertex and primitive data in headers

Both per-primitive and per-vertex space is allocated in MUE in 8 dword
chunks and those 8-dword chunks (granularity of
3DSTATE_SBE_MESH.Per[Primitive|Vertex]URBEntryOutputReadLength)
are passed to fragment shaders as inputs (either non-interpolated
for per-primitive and flat vertex attributes or interpolated
for non-flat vertex attributes).

Some attributes have a special meaning and must be placed in separate
8/16-dword slot called Primitive Header or Vertex Header.

Primitive Header contains 4 such attributes (Cull Primitive,
ViewportIndex, RTAIndex, CPS), leaving 4 dwords (the rest of 8-dword
slot) potentially unused.

Vertex Header is similar - it starts with 3 unused dwords, 1 dword for
Point Size (but if we declare that shader doesn't produce Point Size
then we can reuse it), followed by 4 dwords for Position and optionally
8 dwords for clip distances.

This means we have an interesting optimization problem - we can put
some user attributes into holes in Primitive and Vertex Headers, which
may lead to smaller MUE size and potentially more mesh threads running
in parallel, but we have to be careful to use those holes only when
we need it, otherwise we could force HW to pass too much data to
fragment shader.

Example 1:
Let's assume that Primitive Header is enabled and user defined
12 dwords of per-primitive attributes.

Without packing we would consume 8 + ALIGN(12, 8) = 24 dwords of
MUE space and pass ALIGN(12, 8) = 16 dwords to fragment shader.

With packing, we'll consume 4 + 4 + ALIGN(12 - 4, 8) = 16 dwords of
MUE space and pass ALIGN(4, 8) + ALIGN(12 - 4, 8) = 16 dwords to
fragment shader.

16/16 is better than 24/16, so packing makes sense.

Example 2:
Now let's assume that Primitive Header is enabled and user defined
16 dwords of per-primitive attributes.

Without packing we would consume 8 + ALIGN(16, 8) = 24 dwords of
MUE space and pass ALIGN(16, 16) = 16 dwords to fragment shader.

With packing, we'll consume 4 + 4 + ALIGN(16 - 4, 8) = 24 dwords of
MUE space and pass ALIGN(4, 8) + ALIGN(16 - 4, 8) = 24 dwords to
fragment shader.

24/24 is worse than 24/16, so packing doesn't make sense.

This change doesn't affect vk_meshlet_cadscene in default configuration,
but it speeds it up by up to 25% with "-extraattributes N", where
N is some small value divisible by 2 (by default N == 1) and we
are bound by URB size.

Reviewed-by: Ivan Briano <ivan.briano@intel.com>
(cherry picked from commit c1685f08dd4edcf505d5e9a150b70ccbbb989fa5)

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25188>

7f766cfb

by Marcin Ślusarz at 2023-09-22T08:46:39-07:00

intel/compiler: load debug mesh compaction options once

Reviewed-by: Ivan Briano <ivan.briano@intel.com>
(cherry picked from commit 48885c7fe347b61555ddae247a7c7fdc6834a6a1)

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25188>

73ccbcab