Commit graph

9500 commits

Author SHA1 Message Date
Timur Kristóf
649da9e744 radv: Separate option to print shader disassembly.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32079>
2024-11-19 16:15:35 +01:00
Timur Kristóf
3f92c0bbe9 radv: Separate option to dump NIR.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32079>
2024-11-19 16:15:33 +01:00
Timur Kristóf
d94e65574f aco: Separate options for printing IR and recording disassembly.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32079>
2024-11-19 16:15:28 +01:00
Timur Kristóf
ab89fdc3d4 radv: Add ability to dump shaders based on stage.
The new debug flags can make the output less overwhelming
and only dump what we want to see.

The old RADV_DEBUG=shaders will still print all stages.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32079>
2024-11-19 16:15:23 +01:00
Timur Kristóf
57161f516e radv: Mark GS copy shaders as internal.
Instead of passing the meta_shader arg.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32079>
2024-11-19 16:14:45 +01:00
Samuel Pitoiset
2f13723c0a radv: add a new drirc option to disable DCC for mips and enable it for RDR2
The game aliases two images. It binds a memory object to two different
images, the first one being an image with 4 mips and the second with
only one mip but the bind offset is incorrect. It's like it queried
the first image size with different usage flags, so that DCC was
disabled.

Force disabling DCC for mips fixes the incorrect rendering and doesn't
hurt performance.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10200
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32190>
2024-11-19 07:39:13 +00:00
Samuel Pitoiset
2e51d0c724 radv: add a helper to destroy a logical device
It's less error prone than duplicating every cleanups with a bunch of
gotos.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32112>
2024-11-19 07:02:52 +00:00
Samuel Pitoiset
2181ddf383 radv: destroy meta resources properly when creating the device failed
Better to call radv_device_finish_meta().

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32112>
2024-11-19 07:02:52 +00:00
Marek Olšák
a7ba36f589 ac/nir: get pass_tessfactors_by_reg from nir_gather_tcs_info
If nir_tcs_info::all_invocations_define_tess_levels is true, the pass
doesn't have to insert a barrier and use output loads to get tess level
output values. It can just use the SSA defs that are being stored (or phis
thereof) to get the tess level output values.

The remaining tcs_info fields will be used by the HS shader message.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32171>
2024-11-16 21:58:29 -05:00
Marek Olšák
b258a9aa4e aco: remove unused TCS fields from aco_shader_info
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32171>
2024-11-16 21:58:26 -05:00
Samuel Pitoiset
d2960a8430 radv: consider VK_PIPELINE_STAGE_2_NONE like BOTTOM_OF_PIPE
VK_PIPELINE_STAGE_2_NONE from sync2 is similar to BOTTOM_OF_PIPE.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32115>
2024-11-15 08:22:23 +00:00
Samuel Pitoiset
c08d2c40ed radv: fix ignoring src stage mask when dst stage mask is BOTTOM_OF_PIPE
Otherwise the driver doesn't synchronize if there are image layout
transitions.

This fixes rendering issues with displayable DCC (usually black squares
in the bottom of screen). This mostly happens when an application
uses a lower resolution than the screen supports and fshack
(wine/proton) which upscales images uses COMPUTE_SHADER->BOTTOM_OF_PIPE
for the barrier after a dispatch.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11547
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11600
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11789
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8705
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9890
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32115>
2024-11-15 08:22:23 +00:00
Samuel Pitoiset
45c0ef3bb4 radv: dump SPIR-V and NIR for the faulty shader detected with the trap
More logs is always better for debugging.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32116>
2024-11-14 15:57:07 +00:00
Samuel Pitoiset
9149488a9d radv: mark live invocations when dumping VGPRS with the trap handler
Similar to UMR.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32116>
2024-11-14 15:57:07 +00:00
David Rosca
dcfc956521 radv/video: Override pic_init_qp_minus26 in PPS
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31418>
2024-11-14 07:52:56 +00:00
David Rosca
d166bb5dd1 radv/video: Use 64x16 alignment for HEVC encode
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31418>
2024-11-14 07:52:56 +00:00
David Rosca
d1c1a33b35 radv/video: Avoid selecting rc layer over maximum
Vulkan spec doesn't say if this is allowed or not, but trying
to do this will hang.

Fixes: 4a19047d32 ("radv/video: Select temporal layer when encoding each frame")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31418>
2024-11-14 07:52:56 +00:00
David Rosca
e941acfb9d radv/video: Report correct encodeInputPictureGranularity
Only aligned size can be encoded.

Fixes: 54d499818c ("radv/video: add initial support for encoding with h264.")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31418>
2024-11-14 07:52:56 +00:00
David Rosca
e4ec135d8b radv/video: Fix HEVC slice control
This needs to use aligned size, otherwise it will output two
slices when the size is not 64 aligned.

Fixes: 967e4e09de ("radv/video: add h265 encode support")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31418>
2024-11-14 07:52:56 +00:00
David Rosca
6a121f1507 radv/video: Fix H264 slice control
This needs to use aligned size, otherwise it will output two
slices when the size is not 16 aligned.

Fixes: 54d499818c ("radv/video: add initial support for encoding with h264.")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31418>
2024-11-14 07:52:56 +00:00
Samuel Pitoiset
b4b5f9eeb0 radv,aco: dump VGPRS from the trap handler shader
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32090>
2024-11-13 15:27:54 +00:00
Rhys Perry
7d4cc04156 radv,ac/nir: split global access using nir_lower_mem_access_bit_sizes
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31904>
2024-11-13 12:59:26 +00:00
Rhys Perry
8fdc5d7f9f radv,ac/nir: lower sub-dword loads using nir_lower_mem_access_bit_sizes
fossil-db (navi21):
Totals from 427 (0.54% of 79395) affected shaders:
Instrs: 2939637 -> 2937224 (-0.08%); split: -0.08%, +0.00%
CodeSize: 15982272 -> 15969880 (-0.08%); split: -0.08%, +0.00%
Latency: 21128645 -> 21125738 (-0.01%); split: -0.04%, +0.03%
InvThroughput: 5626811 -> 5626220 (-0.01%); split: -0.03%, +0.02%
SClause: 65771 -> 65731 (-0.06%); split: -0.07%, +0.00%
Copies: 243247 -> 242917 (-0.14%); split: -0.14%, +0.01%
Branches: 100089 -> 100085 (-0.00%)
PreSGPRs: 17879 -> 18118 (+1.34%)
VALU: 1899641 -> 1899278 (-0.02%)
SALU: 468508 -> 466469 (-0.44%)
SMEM: 84305 -> 84291 (-0.02%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31904>
2024-11-13 12:59:26 +00:00
Samuel Pitoiset
44fa24580f radv: optimize the pipe misaligned L2 cache invalidation on GFX11
When using the subresource range, it's possible to reduce the number
of L2 cache invalidations.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31921>
2024-11-12 17:27:39 +00:00
Samuel Pitoiset
7a3a65c0c4 radv: pass the image subresource range to radv_{src,dst}_access_flush()
This will allow us to optimize the pipe misaligned special case for
GFX11 because only the first mip in the mip-tail needs the L2 cache
invalidation.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31921>
2024-11-12 17:27:39 +00:00
Samuel Pitoiset
f7a39fac10 radv: use vk_image_view_subresource_range() when possible
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31921>
2024-11-12 17:27:39 +00:00
Samuel Pitoiset
7a8b725d03 radv: determine the first mip that is pipe misaligned on GFX10+
This will allow us to optimize the GFX11 case where not all mips are
affected by the L2 invalidation.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31921>
2024-11-12 17:27:39 +00:00
Samuel Pitoiset
c5d5f2fbef radv: move the GFX11 special case for mips to radv_image_is_pipe_misaligned()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31921>
2024-11-12 17:27:39 +00:00
Samuel Pitoiset
65bb39bf96 radv: do not always invalidate L2 for GPUs with non-coherent RBs on GFX10+
According to PAL, L2 should be invalidated only for images with
DCC/HTILE even on GPUs with non-coherent RBs. In practice, most of
the images have either DCC/HTILE but this can reduce the number of L2
flushes for images without any compression.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31921>
2024-11-12 17:27:39 +00:00
Samuel Pitoiset
5e0b81413d radv: emit nir_debug_break instructions when the trap handler is enabled
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32061>
2024-11-12 16:05:17 +00:00
Samuel Pitoiset
5f79b8ea2d radv,aco: save/restore overwritten VGPRs in the trap handler shader
The trap currently doesn't return to the shader but it will be needed
for example for the debug mode.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
ccde8ecd64 radv: compute the TMA BO size instead of using a constant
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
3e88f996a5 radv: fix the TMA descriptor size
The TMA BO contains the descriptor first.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
6ec0c85908 radv,aco: use the trap handler layout struct while compiling the shader
It's less error prone to rely on the layout for offsets.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
6bfd92123f aco: simplify postprocessing the trap handler shader
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
44dfeb4479 radv,aco: add a separate function to compile the trap handler shader
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
62e335c779 radv,aco: dump more SQ_WAVE regs from the trap handler
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Samuel Pitoiset
0cc21d0601 radv: cleanup printing SGPRS dumped from the trap handler
It's more readable like that.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32056>
2024-11-12 11:16:13 +00:00
Georg Lehmann
ece1ab3b87 radv: run copy prop before vectorizing
Otherwise there are a lot of scalar movs between texture instructions
and alu. With those removed, the top down vectorizer has more starting
points.

Totals from 296 (0.37% of 79206) affected shaders:
MaxWaves: 5710 -> 5754 (+0.77%)
Instrs: 388051 -> 386630 (-0.37%); split: -0.46%, +0.09%
CodeSize: 2120800 -> 2117144 (-0.17%); split: -0.30%, +0.13%
VGPRs: 17496 -> 17344 (-0.87%)
Latency: 8893751 -> 8901364 (+0.09%); split: -0.10%, +0.18%
InvThroughput: 1740411 -> 1731710 (-0.50%); split: -0.57%, +0.07%
VClause: 6573 -> 6576 (+0.05%); split: -0.21%, +0.26%
SClause: 11233 -> 11209 (-0.21%); split: -0.28%, +0.07%
Copies: 31582 -> 31635 (+0.17%); split: -1.49%, +1.66%
PreSGPRs: 15878 -> 15876 (-0.01%)
PreVGPRs: 15380 -> 15274 (-0.69%)
VALU: 278528 -> 277036 (-0.54%); split: -0.65%, +0.11%
SALU: 49062 -> 49054 (-0.02%); split: -0.03%, +0.02%

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32060>
2024-11-11 18:33:48 +00:00
Samuel Pitoiset
30d9166d80 radv: dump the trap handler shader with RADV_DEBUG=dump_trap_handler
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32031>
2024-11-11 09:34:05 +00:00
Samuel Pitoiset
4d50691ae9 radv: remove unused parameter to radv_fill_nir_compiler_options()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32031>
2024-11-11 09:34:05 +00:00
Konstantin Seurer
e3cf6290e0 radv: Add RADV_DEBUG=nirdebuginfo
Annotates the shader with source locations into the nir shader.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29298>
2024-11-11 08:39:14 +00:00
Konstantin Seurer
736c8c6f23 radv: Dump nir shaders before compiling
It will allow adding source locations that point to the nir_string to
the shader.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29298>
2024-11-11 08:39:14 +00:00
Konstantin Seurer
aaf65d6219 radv: Store debug info inside radv_shader
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29298>
2024-11-11 08:39:14 +00:00
Konstantin Seurer
54c22656b8 radv: Add a helper for accessing the shader binary
Use pointers into the blob instead of hardcoding the layout everywhere.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29298>
2024-11-11 08:39:13 +00:00
Konstantin Seurer
69ebba82d4 aco: Pass debug information to the driver
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29298>
2024-11-11 08:39:13 +00:00
Konstantin Seurer
c5e40a60f8 radv: Lower non-uniform access after vectorization
Scalar access can make nir_lower_non_uniform_access emit a lot of
waterfall loops.

Totals from 83 (0.10% of 84770) affected shaders:
Instrs: 2747926 -> 2745959 (-0.07%); split: -0.07%, +0.00%
CodeSize: 15022460 -> 14998240 (-0.16%); split: -0.16%, +0.00%
Latency: 18602932 -> 18404976 (-1.06%); split: -1.18%, +0.12%
InvThroughput: 4500730 -> 4450364 (-1.12%); split: -1.18%, +0.06%
VClause: 93651 -> 91848 (-1.93%); split: -1.93%, +0.00%
SClause: 63672 -> 63595 (-0.12%); split: -0.13%, +0.00%
Copies: 229377 -> 229896 (+0.23%); split: -0.04%, +0.27%
Branches: 107630 -> 107627 (-0.00%); split: -0.01%, +0.00%
PreSGPRs: 5247 -> 5253 (+0.11%)
PreVGPRs: 5911 -> 5903 (-0.14%); split: -0.29%, +0.15%
VALU: 1761158 -> 1761540 (+0.02%); split: -0.01%, +0.03%
SALU: 419743 -> 419783 (+0.01%); split: -0.01%, +0.02%
VMEM: 152142 -> 150208 (-1.27%)
SMEM: 80251 -> 80244 (-0.01%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30509>
2024-11-11 07:53:13 +00:00
Samuel Pitoiset
437bd63265 radv,aco: dump m0 and exec from the trap handler
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32026>
2024-11-08 14:00:15 +00:00
Pierre-Eric Pelloux-Prayer
acc32cadf5 radv: set info->family_overridden when RADV_FORCE_FAMILY is used
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31841>
2024-11-08 13:31:02 +00:00
Samuel Pitoiset
9cc07bbd09 radv: mark some GFX6-7 GPUs as Vulkan 1.3 conformant
It's the first time RADV is Vulkan conformant on GFX6-7! Some chips
are missing because we don't have access but most of the GFX6-7 GPUs
are covered.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32022>
2024-11-07 11:50:10 +00:00