Commit graph

1700 commits

Author SHA1 Message Date
Samuel Pitoiset
a9f4931a91 ac,radv,radeonsi: a function that sets mutable CB surface fields
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29378>
2024-05-28 08:49:53 +00:00
Konstantin Seurer
9b932aadf4 radv: Remove radv_cmd_dirty_dynamic_bits
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29345>
2024-05-27 20:34:36 +00:00
Samuel Pitoiset
07c6fdb878 radv: use the common helper for initializing CB surfaces
This adds GFX12 support implicitly.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29342>
2024-05-27 17:04:26 +02:00
Samuel Pitoiset
ed30b320c8 radv: use the common helper for initializing DS surfaces
This adds GFX12 support implicitly.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29329>
2024-05-23 15:47:14 +00:00
Samuel Pitoiset
e9a390cb94 radv: replace db_{z,stencil}_{read,write}_base by db_{depth,stencil}_base
Both read/write register values are similar.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29329>
2024-05-23 15:47:14 +00:00
Samuel Pitoiset
074f3cfe73 radv: use new common helpers for building buffer descriptor
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29268>
2024-05-22 08:31:39 +00:00
Samuel Pitoiset
caca5e0de6 radv: add more helpers to emit viewports
Viewports are emitted differently on GFX12 and adding helpers would
help.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29215>
2024-05-17 08:04:02 +00:00
Samuel Pitoiset
73e1ff6b54 radv: use float instead of double for viewport zscale/ztranslate
The float precision should be largely enough.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29215>
2024-05-17 08:04:02 +00:00
Samuel Pitoiset
50ef8600bc radv: pass radv_physical_device to radv_emit_default_sample_locations()
To emit PA_SC_CENTROID_0 which changed on GFX12.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29215>
2024-05-17 08:04:02 +00:00
Samuel Pitoiset
c8852719d0 radv: rename radeon perfctr uconfig helpers
To match other helpers.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29192>
2024-05-15 11:34:35 +00:00
Samuel Pitoiset
2957cedad7 radv: remove redundant radeon_set_perfctr_reg() helper
It's exactly the same as radeon_set_uconfig_reg_perfctr().

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29192>
2024-05-15 11:34:35 +00:00
Samuel Pitoiset
f0d4212847 radv: stop using radv_physical_device for radeon helpers
It will be easier to share helpers between RadeonSI and RADV.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29192>
2024-05-15 11:34:35 +00:00
Samuel Pitoiset
6c8224c693 radv: simplify radv_emit_ps_inputs() slightly
Also remove useless check about ps_offset being zero because memcmp
return false in this case.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28983>
2024-05-14 15:32:08 +00:00
Samuel Pitoiset
c63ac0521a radv: track all graphics shaders context registers
This should be similar to the previous ctx_cs logic that was
implemented in order to reduce context rolls but this also applies
to shader objects.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28983>
2024-05-14 15:32:08 +00:00
Samuel Pitoiset
9532b0f1b2 radv: emit graphics pipelines directly from the cmdbuf
This allows us to unify emitting monolithic graphics pipelines and
shader objects.

Though, this temporarily reduces performance in some games due to more
context rolls. But this will be fixed in the following commits by
using the recent mechanism to track context register writes.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28983>
2024-05-14 15:32:07 +00:00
Samuel Pitoiset
2097bec163 radv: fix the late scissor workaround for GFX9 since a recent refactoring
This was a typo.

Fixes: 92337aff03 ("radv: split cmdbuf dirty flags into dirty/dirty_dynamic")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11142
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29156>
2024-05-14 08:36:46 +00:00
Rhys Perry
d48c8905f1 radv: keep track of unaligned dynamic vertex access
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/29071>
2024-05-13 17:22:26 +00:00
Samuel Pitoiset
dd3ae170a7 radv: do not emit VGT_GS_OUT_PRIM_TYPE to ctx_cs on GFX11
Missed that one but it doesn't cause any context rolls.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29132>
2024-05-13 16:10:20 +00:00
Marek Olšák
58a5de5c34 amd: add gfx12 register definitions into the register header generator
The generator renamed some definitions to resolve conflicts.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29007>
2024-05-11 22:14:05 -04:00
Samuel Pitoiset
43fbbc0732 radv: track and bind more VRS states from the graphics pipeline
This doesn't change anything but this will allow us to emit all
graphics shaders from the cmdbuf.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29103>
2024-05-09 08:15:56 +00:00
Samuel Pitoiset
24814be08a radv: stop recomputing the last VGT API stage when emitting graphics shaders
The last VGT shader is already set correctly.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29103>
2024-05-09 08:15:56 +00:00
Samuel Pitoiset
c6a22dd05c radv: precompute NGG register values
To make emission faster.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29031>
2024-05-09 06:29:29 +00:00
Samuel Pitoiset
1173058002 radv: add a new mechanism for tracking registers per cmdbuf
We already track a couple of registers per cmdbuf and this introduces
a generic mechanism, instead of having a bunch of last_xxx fields.

Loosely based on RadeonSI.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28644>
2024-05-08 11:45:52 +00:00
Samuel Pitoiset
c533a79878 radv: implement VK_KHR_dynamic_rendering_local_read
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27263>
2024-05-07 10:35:04 +00:00
Samuel Pitoiset
c9162034bc radv: precompute DB_SHADER_CONTROL for fragment shaders later
To regroup all precomputed register values.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29022>
2024-05-06 18:00:02 +00:00
Samuel Pitoiset
e5bc4d85bb radv: precompute existing legacy GS register values later
To precompute all registers at the same place.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29022>
2024-05-06 18:00:02 +00:00
Samuel Pitoiset
92337aff03 radv: split cmdbuf dirty flags into dirty/dirty_dynamic
We are out of bits.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29039>
2024-05-06 08:33:37 +02:00
Samuel Pitoiset
d71d189790 radv: add a new dirty state for emitting the color output state
SPI_SHADER_COL_FORMAT/CB_SHADER_MASK are used slightly differently
for PS epilogs, shader objects and monolithic graphics pipelines.

This introduces a new state that will allow us to emit these two
registers in only place. The main motivation is for depth-only RB+
support and for tracking context registers in the cmdbuf.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28976>
2024-05-03 06:29:05 +00:00
Samuel Pitoiset
66d4188ec5 radv: store cb_shader_mask for fragment shaders and epilogs
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28976>
2024-05-03 06:29:05 +00:00
Samuel Pitoiset
0ce1bfc040 radv: rename col_format_non_compacted to spi_shader_col_format
This is always the non-compacted format because it's compacted right
before it's emitted. This looks much cleaner to me.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28976>
2024-05-03 06:29:05 +00:00
Samuel Pitoiset
199f521804 radv: compact SPI_SHADER_COL_FORMAT as late as possible
This will allow us to do more cleanups because this thing is a complete
mess.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28976>
2024-05-03 06:29:05 +00:00
Samuel Pitoiset
3b41fbd4b8 radv: precompute compute/task shader register values
To make emission faster.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29014>
2024-05-03 06:07:46 +00:00
Samuel Pitoiset
8c4d0b287f radv: emit compute pipelines directly from the cmdbuf
Using this intermediate CS isn't really useful and it prevents us to
optimize register writes in the near future. This will also be removed
for graphics pipelines.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28977>
2024-05-02 10:39:03 +00:00
Samuel Pitoiset
86281ef15f radv: add shaders BO to the cmdbuf BO list at bind time
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
42554e81b9 radv: add RT prolog BO to the cmdbuf BO list at bind time
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
42dc4b463b radv: add GS copy shader BO to the cmdbuf BO list at bind time
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
2664e058de radv: use the bound GS copy shader when emitting shader objects
Similar but doesn't rely on shader_objs.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
be98fe2724 radv: pre-compute VGT_TF_PARAM.DISTRIBUTION_MODE
For less CPU overhead.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
d7679c0370 radv: remove useless DB_Z_INFO.NUM_SAMPLES when emitting the MSAA state
DB_Z_INFO.NUM_SAMPLES is now correctly set when a null framebuffer is
emitted and this is redundant.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
4dd682e227 radv: inline radv_get_pa_su_sc_mode_cntl() in radv_emit_culling()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
e651a2c856 radv: simplify radv_emit_primitive_restart_enable()
Move emitting VGT_MULTI_PRIM_IB_RESET_INDX into the GFX6-8 branch.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28965>
2024-04-30 07:18:08 +00:00
Samuel Pitoiset
85deb9f706 radv: simplify DB_Z_INFO.NUM_SAMPLES with null ds target on GFX11
According to PAL, the hw uses the smaller value of
DB_Z_INFO.NUM_SAMPLES and PA_SC_AA_CONFIG.MSAA_EXPOSED_SAMPLES when
there is no bound depth/stencil buffer, and it uses 8x to make sure
the used value is MSAA_EXPOSED_SAMPLES.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28952>
2024-04-29 11:02:02 +00:00
Konstantin Seurer
d6c9b1d03f radv: Handle all dependencies of CmdWaitEvents2
The spec describes pDependencyInfos as an array with eventCount elements.

cc: mesa-stable

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10579
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28896>
2024-04-26 08:09:22 +00:00
Konstantin
46598758e7 radv: Trace indirect dispatch sizes
For figuring out hanging indirect dispatches.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28838>
2024-04-25 10:20:03 +00:00
Konstantin
2b2f67aa2b radv: Use a struct for the trace_bo layout
Now we can use the members on the CPU side and offsetof on the GPU side
instead of magic offsets.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28838>
2024-04-25 10:20:03 +00:00
Samuel Pitoiset
fb323ae46b radv: rework the number of tess patches computation
This uses the same helper as RadeonSI which seems more robust and more
optimal (eg. it reduces the number of patches to increase occupancy).

fossils-db (NAVI21):
Totals from 638 (0.80% of 79395) affected shaders:
MaxWaves: 13182 -> 13142 (-0.30%)
Instrs: 419446 -> 419322 (-0.03%); split: -0.08%, +0.05%
CodeSize: 2261408 -> 2261200 (-0.01%); split: -0.06%, +0.05%
VGPRs: 32560 -> 32600 (+0.12%)
LDS: 4648960 -> 5343232 (+14.93%); split: -1.67%, +16.61%
Latency: 4812105 -> 4811141 (-0.02%); split: -0.04%, +0.02%
InvThroughput: 1159924 -> 1153998 (-0.51%); split: -0.60%, +0.09%
VClause: 7837 -> 7871 (+0.43%); split: -0.36%, +0.79%
SClause: 9378 -> 9381 (+0.03%); split: -0.21%, +0.25%
Copies: 28451 -> 28211 (-0.84%); split: -0.97%, +0.13%
PreVGPRs: 25404 -> 25411 (+0.03%); split: -0.06%, +0.09%
VALU: 278086 -> 277975 (-0.04%); split: -0.11%, +0.07%
SALU: 43657 -> 43617 (-0.09%)

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28015>
2024-04-23 17:20:40 +00:00
Samuel Pitoiset
7de95e7742 radv: track if nested command buffers uses indirect draws
IB2 packets should be avoided when a cmdbuf executes nested cmdbufs.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28826>
2024-04-23 16:41:57 +00:00
Samuel Pitoiset
e4f945cd4a vulkan: pass cmdbuf level to vk_command_buffer_ops::create()
RADV needs to know the command buffer level in the create() helper.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28861>
2024-04-23 06:33:31 +00:00
Timur Kristóf
2d9e38dbe5 radv: Calculate VRAM tess patch size independently of LDS size.
We recently made some effort to reduce the LDS use of TCS:
The lowering no longer uses the same output location mapping when
storing TCS outputs to LDS and VRAM. This means that the same
patch will use a different amount of LDS and VRAM.

Therefore, we need to properly calculate the patch size in VRAM
when determining the number of output patches.

Fixes: 0e481a4adc
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28739>
2024-04-19 08:54:23 +00:00
Samuel Pitoiset
74615bb704 radv: clear color attachments without exports before compaction
For PS epilogs, this isn't necessary because spi_shader_col_format is
already cleared. This will help for implementing color attachment
remapping because colors_written is always the original mapping.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28781>
2024-04-17 13:58:37 +00:00