AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
169,013 Commits 1 Branch 0 Tags
22903bcdedb29a23586a73ce6a52e9e2eb3cc149
Commit Graph

156372 Commits

Author SHA1 Message Date
Georg Lehmann 22903bcded aco: add tests for swap operand with opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:35 +00:00
Georg Lehmann 0b29dc5c06 aco: add tests for dpp with opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann f7bb794dda aco: add tests for fma with opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann bb7c2b70c1 aco/optimizer: remove to_SDWA 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann e699a4181c aco: keep label_mul/usedef/minmax in apply_extract 
16bit int mad/fma/minmax combining can work with opsel set.

All other optimizations should already check if the instruction uses sdwa,
because we don't check this when applying the label initially.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 7014145ab2 aco/optimizer: use opsel for VOP12C 
Foz-DB GFX1100:
Totals from 11759 (8.72% of 134864) affected shaders:
VGPRs: 848288 -> 844556 (-0.44%); split: -0.44%, +0.00%
SpillSGPRs: 8527 -> 8543 (+0.19%)
SpillVGPRs: 1411 -> 1423 (+0.85%); split: -0.21%, +1.06%
CodeSize: 114337120 -> 113882472 (-0.40%); split: -0.40%, +0.01%
Scratch: 128768 -> 129024 (+0.20%); split: -0.20%, +0.40%
MaxWaves: 250962 -> 252014 (+0.42%)
Instrs: 22187426 -> 22062378 (-0.56%); split: -0.57%, +0.00%
Latency: 232655375 -> 232376977 (-0.12%); split: -0.20%, +0.08%
InvThroughput: 28292530 -> 28217699 (-0.26%); split: -0.45%, +0.18%
VClause: 352463 -> 352364 (-0.03%); split: -0.12%, +0.10%
SClause: 659282 -> 659354 (+0.01%); split: -0.02%, +0.04%
Copies: 1371369 -> 1342340 (-2.12%); split: -2.30%, +0.19%
Branches: 495903 -> 495941 (+0.01%); split: -0.00%, +0.01%
PreSGPRs: 867295 -> 863664 (-0.42%)
PreVGPRs: 793480 -> 790549 (-0.37%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 9650724370 aco/gfx11: allow opsel for VOP12C 
Foz-DB GFX1100:
Totals from 515 (0.38% of 134864) affected shaders:
CodeSize: 2768228 -> 2761076 (-0.26%)
Instrs: 520301 -> 518523 (-0.34%)
Latency: 5190860 -> 5187254 (-0.07%)
InvThroughput: 2120844 -> 2119447 (-0.07%)
Copies: 57238 -> 56101 (-1.99%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 3907c54443 aco: don't label mul with opsel as abs/neg 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann ace017bba8 aco/ir: copy opsel when converting to DPP 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann a60b9313d3 aco: swap opsel when swapping VOP2/C operands 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann fc1bf9c3b4 aco: return true in usesModifiers for VOP12C with opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 82f7b3acfa aco: support neg(mul)/abs(mul) optimization in more cases 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 9d841507e1 aco: support v_cvt_f32_f16 with opsel in combine_mad_mix 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 9d6e223a7a aco: update match_op3_for_vop3 for VOP12C opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 0896ecec9a aco: handle opsel in combine_constant_comparison_ordering 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann d8f07a0ddc aco: handle opsel in combine_ordering_test 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 4db43415e5 aco: handle opsel in combine_comparison_ordering 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 8e6d79d10d aco/optimizer: preserve opsel when fusing fma 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 32d7a11acf aco/ra: prepare for VOP12C opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 8ee1519cee aco/to_hw_instr: use VOP1 opsel for v_mov_b16 
Foz-DB GFX1100:
Totals from 4661 (3.46% of 134864) affected shaders:
CodeSize: 36500568 -> 36391704 (-0.30%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 066cee0896 aco: validate VOP12C opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 2c49b7babf aco/assembler: support VOP12C opsel 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann 9b4ea9ff90 aco/vn: hash opsel for VOP12C 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Georg Lehmann c62e5ef82e aco/ra: don't reallocate VOP3 instruction for non-vcc lane mask 
This would need to copy opsel soon but we can just reuse the old instruction.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22069>
 2023-03-30 03:34:34 +00:00
Friedrich Vock 424825c6e5 aco: Un-swap addressable VGPRs/SGPRs in RT prolog 
Fixes: 6446b79168 ("aco: implement select_rt_prolog()")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22188>
 2023-03-30 02:55:54 +00:00
Friedrich Vock 83a38987b1 radv/rt: Also adjust the SGPR count in postprocess_rt_config 
Fixes: bea022d1f6 ("radv/rt: Add shader config combination/postprocessing utils")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22188>
 2023-03-30 02:55:53 +00:00
Emma Anholt c443503e07 perfetto: Move intel's cmdbuf/queue annotation code to the shared util. 
This will let other drivers use the same way of presenting annotations
without duplicating the whole hash table thing.

Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22157>
 2023-03-30 02:19:35 +00:00
Emma Anholt 9c364a346f perfetto: Deduplicate clock sync packet emit from renderstage sources. 
This is way more horrifying than I hoped -- I can't figure out a way to
have the method be on TraceContext, so it's a static method of the
datasource, but then you have to name the templated types over and over.
You have to pass in a TraceContext because intel emits the clock sync
packet within a Trace(), and perfetto just silently corrupts the trace if
you Trace() in a Trace().

Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22157>
 2023-03-30 02:19:35 +00:00
Emma Anholt 3fd825d3e2 perfetto: Make a MesaRenderpassDataSource with common setup/start/stop. 
Deduplicates some code from intel/tu/freedreno, and will be a common place
to put other shared code.

The downside I can see is this logging:

[013.129]      tu_perfetto.cc:122 Tracing started
[013.129]  intel_driver_ds.cc:133 Tracing started

("oh, huh, apparently data sources for both drivers are registered?  wild")

becomes:

[142.906] erfetto_renderpass.h:50 Tracing started
[142.907] erfetto_renderpass.h:50 Tracing started

("huh, why is my driver's data source being started twice?").
Unfortunately we can't easily get a string for the data source type due to
not having rtti.

Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22157>
 2023-03-30 02:19:34 +00:00
Emma Anholt 42f1df1ec5 intel/perfetto: Drop unused "pipelined" field. 
Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22157>
 2023-03-30 02:19:34 +00:00
Emma Anholt da78d5d729 perfetto: Add a .clang-format for the directory. 
I was frustrated trying to write code and not be able to just mash ^K^F to
format what I'd written.  This .clang-format is just cargo-cult of turnip
with a few tweaks to reduce the diff to the current directory contents.
The remaining deltas in the reformat look decent to me, and mostly bring
things closer to mesa-vague-consensus style.

Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22157>
 2023-03-30 02:19:34 +00:00
José Roberto de Souza 66a8f7dcdb anv: Add assert in functions not supported by Xe kmd 
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22171>
 2023-03-30 01:57:44 +00:00
José Roberto de Souza 74ff665bd0 anv: Disable anv_bo_sync_type for Xe kmd 
anv_bo_sync_type is only supported with i915, if necessary it will be
implemented for Xe kmd.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22171>
 2023-03-30 01:57:44 +00:00
José Roberto de Souza fdea48df5e anv: Implement Xe version of anv_queue_exec_locked() and queue_exec_trace() 
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22171>
 2023-03-30 01:57:44 +00:00
José Roberto de Souza 91ee00a741 anv: Move to a function code to clflush batch buffers 
This also need to be executed in Xe kmd, so moving it to a function.
No changes in behavior intended here.

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22171>
 2023-03-30 01:57:44 +00:00
Juston Li 88d074cb8f util/disk_cache: use posix_fallocate() for index files 
ftruncate() allocates disk space lazily. If the disk is full and it is
unable to allocate disk space when accesed via mmap(), it will crash
with a SIGBUS.

Switch to posix_fallocate(), which ensures disk space is allocated
otherwise it fails if there isn't enough disk space. The disk cache
won't be enabled in this case.

For normal cases, a small increase in disk usage as the 1.3MB index
file will be fully allocated when initialized now.

fallback to ftruncate() if posix_fallocate() isn't found.

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22097>
 2023-03-30 01:09:10 +00:00
Lionel Landwerlin ba670f0cdf anv: hash immutable sampler conversion data not pointers 
The conversion pointer has a vk_object_base containing pointers.

Previous commits had a similar issue, storing an internal pointer to
the array of formats in anv_format.c

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22160>
 2023-03-30 00:31:20 +00:00
Ian Romanick 71e5530c07 nir/algebraic: Undistribute fsat from fmax 
To be helpful, the thing inside the fsat has to be used with and without
the fsat. Otherwise it just moves a saturate destination modifier
around. To not be harmful, the fsat has to only be used by the bcsel.

All Broadwell and newer Intel platforms had similar results. (Ice Lake shown)
total instructions in shared programs: 20174475 -> 20174449 (<.01%)
instructions in affected programs: 3913 -> 3887 (-0.66%)
helped: 13 / HURT: 0

total cycles in shared programs: 866844832 -> 866844719 (<.01%)
cycles in affected programs: 46037 -> 45924 (-0.25%)
helped: 10 / HURT: 1

All Intel platforms had similar results. (Ice Lake shown)
Instructions in all programs: 161491468 -> 161491372 (-0.0%)
helped: 31 / HURT: 8

Cycles in all programs: 10933090736 -> 10933024716 (-0.0%)
helped: 32 / HURT: 18

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22169>
 2023-03-29 23:48:19 +00:00
Ian Romanick 782de1932c intel/fs: Don't copy propagate from saturate to sel 
There are already NIR algebraic optimizations (see also ac6646129f
("nir: Move fsat outside of fmin/fmax if second arg is 0 to 1.") that
will try to remove the saturate from things like

    fmax(0.5, fsat(x))

This basically reverts 40aeb558ce ("i965/fs: Allow propagation of
instructions with saturate flag to sel"). That commit message had no
shader-db information, so it's unclear whether this actually helped
anything ever.

No shader-db changes on any Intel platform.

One shader in Far Cry New Dawn was affected.

Cycles in all programs: 10933090738 -> 10933090736 (-0.0%)
Cycles helped: 1

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22169>
 2023-03-29 23:48:19 +00:00
Rhys Perry 125ecd7e2d radv: fix setting radv_shader_info::user_data_0 with rt 
Fixes raytracing pipelines.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Fixes: 0e81ec98ce ("radv: move user_data_0 to the shader info pass")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22187>
 2023-03-29 22:18:33 +00:00
Guilherme Gallo 47774d2414 ci/zink: Reduce zink-tu-a618-traces parallelism 
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22155>
 2023-03-29 21:30:19 +00:00
Guilherme Gallo a5de0174af ci/zink: Fix zink-tu-a618-traces perf job rules 
The pipeline should create this job only when it is under performance
enabled workflow, represented by the presence of
MESA_CI_PERFORMANCE_ENABLED variable.

For example:
The following pipeline
https://gitlab.freedesktop.org/gallo/mesa/-/pipelines/840525 has
`zink-tu-a618-traces-performance` job, even if it was not triggered by
marge-bot, which means that performance jobs should be disabled.

This job should appear in pipelines with performance jobs enabled
https://gitlab.freedesktop.org/gallo/mesa/-/pipelines/840529

Fixes: 93e3d37b47
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22155>
 2023-03-29 21:30:19 +00:00
Marek Olšák 43dc19f44d radeonsi: simplify binning settings to work around GPU hangs 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22084>
 2023-03-29 20:36:10 +00:00
Marek Olšák 281126f1d6 radeonsi/gfx11: reduce MSAA samples to 8 for no-attachment framebuffer 
EQAA is unsupported

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22084>
 2023-03-29 20:36:09 +00:00
Marek Olšák e0d449dd40 amd: set the correct LLVM processor name for gfx1036 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22084>
 2023-03-29 20:36:09 +00:00
Marek Olšák 0b6a7cba0b amd: rename GFX1036 -> RAPHAEL_MENDOCINO 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22084>
 2023-03-29 20:36:09 +00:00
Marek Olšák 52b6886992 amd: update addrlib 
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22084>
 2023-03-29 20:36:09 +00:00
antonino 5fb8ba0eb8 zink/ci: remove primitive-id-no-gs-quads from radv-vangogh-fails 
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21238>
 2023-03-29 19:18:40 +00:00
antonino 1c3a4ab896 zink: unbind generated gs in bind_last_vertex_stage 
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21238>
 2023-03-29 19:18:40 +00:00
antonino 303e06b19d zink: improve generated gs unbinding 
Avoid looping by using the new `parent` field to check if a generaetd gs
is bound and use `bind_gs_state` insted of `bind_gfx_stage` so that
`bind_last_vertex_stage` is automatically called

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21238>
 2023-03-29 19:18:40 +00:00