Commit graph

201789 commits

Author SHA1 Message Date
Mel Henning
11b8c8b8e6 nak,nir: Add 64-bit lea_nv
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32517>
2025-02-13 17:36:41 +00:00
Mel Henning
c92a92e72b nak: Add OpLeaX
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32517>
2025-02-13 17:36:41 +00:00
Mel Henning
0470643047 nak,nir: Add 32-bit nir_op_lea_nv and use it
Changes code size by -0.80% on shaderdb.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32517>
2025-02-13 17:36:41 +00:00
Mel Henning
54fcc63d3e nak: Add OpLea
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32517>
2025-02-13 17:36:41 +00:00
Yiwei Zhang
ff64092ff3 venus: support VK_EXT_sample_locations
Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33520>
2025-02-13 17:12:58 +00:00
Yiwei Zhang
ae3bc10d58 venus: refactor to share more codes between pipeline state fillings
A lot of codes can be shared for filling states of fragment shader and
fragment output interface.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33520>
2025-02-13 17:12:58 +00:00
Yiwei Zhang
7c28f614a4 venus: support VK_EXT_blend_operation_advanced
Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33520>
2025-02-13 17:12:58 +00:00
Yiwei Zhang
a7fccbbf85 venus: fix to handle pipeline flags2 from maint5
Fixes: be6fece6e1 ("venus: enable VK_KHR_maintenance5")
Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33520>
2025-02-13 17:12:58 +00:00
Yiwei Zhang
a7adf8db67 venus: group private data together with other 1.3 exts
Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33520>
2025-02-13 17:12:58 +00:00
Yiwei Zhang
f656ca6a13 venus: sync venus protocol for below extensions
VK_EXT_blend_operation_advanced
VK_EXT_sample_locations

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33520>
2025-02-13 17:12:58 +00:00
Andrew Wolfers
92370f63ab vulkan: Add BGRA format support
This change adds handling to convert between the VULKAN and
AHARDWAREBUFFER enum values for the BGRA_8888 buffer format.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33485>
2025-02-13 16:10:57 +00:00
Mohamed Ahmed
8a71b21c2e nvk: Set NIL_IMAGE_USAGE_VIDEO_BIT for images meant for video usage
Reviewed-by: Faith Ekstrand <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33453>
2025-02-13 15:16:36 +00:00
Mohamed Ahmed
7d21d17b01 nil: Force smallest block size for images meant for Vulkan Video
Reviewed-by: Faith Ekstrand <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33453>
2025-02-13 15:16:36 +00:00
Mohamed Ahmed
54fed0536d nil: Use multiplanar init_info during image creation
Reviewed-by: Faith Ekstrand <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33453>
2025-02-13 15:16:36 +00:00
Mohamed Ahmed
b2ca04ec78 nvk: Add NVK_MAX_SAMPLER_PLANES for samplers
Reviewed-by: Faith Ekstrand <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33453>
2025-02-13 15:16:36 +00:00
Mohamed Ahmed
8c70c8f60f nvk: Add NVK_MAX_IMAGE_PLANES for images
Reviewed-by: Faith Ekstrand <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33453>
2025-02-13 15:16:36 +00:00
Lionel Landwerlin
db53e53bf6 brw: add documentation about slot compaction & component packing
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
6845dede59 brw: add support for no VF input slot compaction
Normally the driver & compiler work together to use as few
3DSTATE_VERTEX_ELEMENTS/VERTEX_BUFFER_ELEMENT data as possible.

The compiler ignores unused bits and driver avoids emitting the
corresponding elements in 3DSTATE_VERTEX_ELEMENTS.

For device generated commands, we want an 3DSTATE_VERTEX_ELEMENTS
programming that is independent from the shader so that we can
implement indirect pipeline binding without complicating the
generation shader as well as emitting fewer generated commands.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
f19c5f4fcc brw: use meaningful io locations for system values
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
6b99bf76ca anv: ensure Wa_16012775297 interacts correctly with Wa_18020335297
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: dddd765553 ("anv: implement VF_STATISTICS emit for Wa_16012775297")
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
a85717f313 anv: enable vertex fetching component packing
DG2 a/b testing:
   Borderlands3	 -0.55%
   Cyberpunk     +0.38%
   Superposition -0.67%

The shader stats mostly don't look like an improvement :

DG2 shader stats:

  Blackops 3:
  Totals from 265 (16.44% of 1612) affected shaders:
  Instrs: 109055 -> 109080 (+0.02%); split: -0.01%, +0.04%
  Cycle count: 6166549 -> 6021371 (-2.35%); split: -2.53%, +0.17%

  Cyberpunk 2077:
  Totals from 297 (23.50% of 1264) affected shaders:
  Instrs: 197305 -> 197297 (-0.00%); split: -0.03%, +0.02%
  Cycle count: 3374325 -> 3356562 (-0.53%); split: -1.23%, +0.70%

  Fortnite:
  Totals from 2090 (27.97% of 7471) affected shaders:
  Instrs: 1777944 -> 1781070 (+0.18%); split: -0.01%, +0.18%
  Cycle count: 25188758 -> 25162910 (-0.10%); split: -0.86%, +0.76%
  Spill count: 1439 -> 1729 (+20.15%); split: -0.69%, +20.85%
  Fill count: 1226 -> 1395 (+13.78%); split: -0.82%, +14.60%
  Scratch Memory Size: 122880 -> 138240 (+12.50%); split: -1.67%, +14.17%

  Hitman 3:
  Totals from 490 (9.09% of 5392) affected shaders:
  Instrs: 407489 -> 407486 (-0.00%); split: -0.00%, +0.00%
  Cycle count: 1831149 -> 1831890 (+0.04%); split: -0.33%, +0.38%

  Metro Exodus:
  Totals from 4169 (9.68% of 43076) affected shaders:
  Instrs: 817730 -> 817726 (-0.00%); split: -0.00%, +0.00%
  Cycle count: 4646954 -> 4641559 (-0.12%); split: -0.61%, +0.50%

Xe2 shader stats :

  Blackops 3:
  Totals from 283 (19.46% of 1454) affected shaders:
  Cycle count: 7662980 -> 7916316 (+3.31%); split: -0.38%, +3.69%

  Cyberpunk 2077:
  Totals from 329 (26.79% of 1228) affected shaders:
  Instrs: 203312 -> 203327 (+0.01%); split: -0.01%, +0.02%
  Cycle count: 4415812 -> 4434906 (+0.43%); split: -0.69%, +1.12%

  Fortnite:
  Totals from 1981 (30.18% of 6565) affected shaders:
  Instrs: 1709583 -> 1711379 (+0.11%); split: -0.00%, +0.11%
  Cycle count: 26882682 -> 26914014 (+0.12%); split: -0.66%, +0.78%
  Spill count: 863 -> 1020 (+18.19%)
  Fill count: 1195 -> 1271 (+6.36%)
  Scratch Memory Size: 116736 -> 122880 (+5.26%)

  Hitman 3:
  Totals from 540 (10.56% of 5115) affected shaders:
  Instrs: 478993 -> 478994 (+0.00%)
  Cycle count: 3198740 -> 3198416 (-0.01%); split: -0.27%, +0.26%

  Metro Exodus:
  Totals from 4554 (12.28% of 37071) affected shaders:
  Cycle count: 6460340 -> 6475666 (+0.24%); split: -0.38%, +0.62%

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
462d8e3fab anv: disable VF statistics for memcpy
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
ca66f22e90 blorp: emit 3DSTATE_VF
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
4f892ae4f7 brw: enable vertex fetching component packing
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
9b8d75c95c brw: add a max HW vertices attribute limit
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
fae8d325a7 brw: update vulkan max attribute limit
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
bae9344baf brw: port vs input to lower_64bit_to_32_new
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
e9e4aa0f29 brw: remove nr_attribute_slots from vs_prog_data
It's not used outside of the compiler.

We add a new nr_attribute_regs which now seems useless but will be
useful in a later change.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
c00830083e brw: fix indentation
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
2a8dddb519 genxml: add convenience dwords for packing components
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Lionel Landwerlin
e40f47abd3 genxml: make component packing an array
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32418>
2025-02-13 14:36:15 +00:00
Eric R. Smith
c7fed8b053 panfrost: fix YUV center information for 422
It turns out that the change from CENTER_Y to CENTER_X for
422 YUV didn't actually happen until generation 14 of the
hardware, not generation 10 as some documents claimed. This
fixes the failing piglit tests ext_image_dma_buf_import-sample_yuv
associated with 422 formats (which apparently we aren't running on CI).

Fixes: 23aa784c
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33516>
2025-02-13 14:06:59 +00:00
Valentine Burley
2e48bcf064 ci/angle: Uprev ANGLE
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33513>
2025-02-13 13:21:10 +00:00
Valentine Burley
5eea8f6fe8 intel/ci: Fix manual rules for ANGLE jobs
Disable auto-retry for .intel-manual-rules to prevent unnecessary reruns
and switch ANGLE jobs from this rule to .anv-manual-rules, as there’s no
point in running anv-on-angle jobs on iris changes.

Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33513>
2025-02-13 13:21:09 +00:00
Valentine Burley
93569f3a8f ci/angle: Use lld-19 for linking ANGLE
This fixes the linking issues on newer ANGLE versions.

Fixes: 9707746b2d ("ci/lava: Build ANGLE for arm64")
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33513>
2025-02-13 13:21:09 +00:00
Samuel Pitoiset
e977c6968f radv/meta: add radv_{copy,fill,update}_memory() helpers
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33511>
2025-02-13 13:48:09 +01:00
Samuel Pitoiset
7aa4c81744 radv: rename fill/copy memory helpers
memory seems more appropriate than buffer because they operate on VAs.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33511>
2025-02-13 13:47:14 +01:00
Samuel Pitoiset
69bf2a13f8 radv/meta: inline copy_buffer()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33511>
2025-02-13 13:43:30 +01:00
Samuel Pitoiset
5684c1687c radv/meta: disable conditional rendering for fill/update buffer operations
These commands shouldn't be affected by conditional rendering, similar
to the copy buffer operation.

Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33511>
2025-02-13 13:43:30 +01:00
Samuel Pitoiset
5d6e4d3b25 radv/meta: remove the heuristic that prefers CP DMA for GTT BOs
This is actually slower on my side. Tested the copy_buffer/fill_buffer
benchmarks from crucible on NAVI31.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33511>
2025-02-13 13:43:30 +01:00
Mary Guillemard
7881e48321 panvk: Switch JM clear queries to CLC
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32721>
2025-02-13 11:50:06 +00:00
Mary Guillemard
3ed5557baf panvk: Switch JM copy queries to CLC
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32721>
2025-02-13 11:50:06 +00:00
Alyssa Rosenzweig
c51a2e85d8 libcl/vk: add common query copy write routine
every VK driver ends up wanting this.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32721>
2025-02-13 11:50:06 +00:00
Maíra Canal
db4d7a3032 v3dv: VK_EXT_acquire_drm_display doesn't require a DRM master fd
When using VK_EXT_acquire_drm_display, the Vulkan API user must provide
the DRM master fd that will be used. This operation is performed through
`vkAcquireDrmDisplayEXT()` in which `drmFd` will be assigned to
`wsi->fd` and will be used for privileged operations.

This means that, when we are creating the physical device, we need to
open a DRM primary node (as the specification states that "The provided
drmFd must correspond to the one owned by the physicalDevice."), but it
doesn't need to be the DRM master.

Therefore, when using VK_EXT_acquire_drm_display, keep the primary fd
open and don't check if the fd is the DRM master.

Signed-off-by: Maíra Canal <mcanal@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33096>
2025-02-13 11:28:42 +00:00
Daniel Schürmann
6395c6d6d1 nir/validate: validate divergence metadata
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30814>
2025-02-13 10:08:44 +00:00
Daniel Schürmann
bc810e98d4 nir/serialize: don't serialize divergence information
We don't serialize metadata.

Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30814>
2025-02-13 10:08:44 +00:00
Daniel Schürmann
63a656797b nir: only print divergence information if metadata is valid
This disables printing of vertex divergence as well as outdated
divergence information. Also remove info::divergence_analysis_run
and use nir_metadata_divergence instead.

Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30814>
2025-02-13 10:08:44 +00:00
Daniel Schürmann
1d6082bf56 nouveau: switch to nir_metadata_divergence
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30814>
2025-02-13 10:08:43 +00:00
Daniel Schürmann
175c06e5cd intel: switch to nir_metadata_divergence
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30814>
2025-02-13 10:08:43 +00:00
Daniel Schürmann
067478358f amd: switch to nir_metadata_divergence
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30814>
2025-02-13 10:08:43 +00:00