Georg Lehmann [Tue, 3 May 2022 19:57:23 +0000 (21:57 +0200)]
meson: Tell glslang to be quiet.
Without --quiet glslang unconditionally prints the input file name to stdout.
Check if --quiet is supported because some distros only have ancient glslang
versions.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16312>
Rhys Perry [Tue, 3 May 2022 11:13:55 +0000 (12:13 +0100)]
aco/ra: fix vgpr_limit
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Fixes:
b98a4d4dd7e ("aco: refactor GPR limit calculation")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16297>
Georg Lehmann [Thu, 31 Mar 2022 13:40:53 +0000 (15:40 +0200)]
aco: Remove D16 zero components from image stores.
No foz-db changes.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15179>
Georg Lehmann [Mon, 7 Mar 2022 12:10:47 +0000 (13:10 +0100)]
radv: Run copy_prop and dce after folding 16bit sampling/load/store.
Totals from 10 (0.01% of 134913) affected shaders:
CodeSize: 53168 -> 54832 (+3.13%); split: -0.17%, +3.30%
Instrs: 9117 -> 9200 (+0.91%); split: -1.74%, +2.65%
Latency: 41595 -> 41787 (+0.46%); split: -0.95%, +1.41%
InvThroughput: 16412 -> 16424 (+0.07%); split: -1.95%, +2.02%
VClause: 107 -> 112 (+4.67%); split: -0.93%, +5.61%
Copies: 199 -> 535 (+168.84%); split: -3.02%, +171.86%
PreVGPRs: 520 -> 502 (-3.46%)
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15179>
Georg Lehmann [Fri, 25 Feb 2022 17:55:42 +0000 (18:55 +0100)]
radv: Use nir_fold_16bit_image_load_store_conversions.
Totals from 10 (0.01% of 134913) affected shaders:
CodeSize: 53316 -> 53168 (-0.28%)
Instrs: 9219 -> 9117 (-1.11%)
Latency: 41744 -> 41595 (-0.36%)
InvThroughput: 16616 -> 16412 (-1.23%)
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15179>
Georg Lehmann [Fri, 25 Feb 2022 17:52:27 +0000 (18:52 +0100)]
aco: Implement image_load d16.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15179>
Georg Lehmann [Fri, 25 Feb 2022 17:21:00 +0000 (18:21 +0100)]
aco: Implement image_store d16.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15179>
Georg Lehmann [Mon, 7 Feb 2022 20:43:13 +0000 (21:43 +0100)]
nir/lower_mediump: Add a new pass to fold 16bit image load/store.
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15179>
Alejandro Piñeiro [Tue, 3 May 2022 10:48:10 +0000 (12:48 +0200)]
v3dv/pipeline: include pipeline layout on the pipeline sha1
Fixes failures on tests like this when the on-disk-cache is enabled:
dEQP-VK.binding_model.descriptor_copy.compute.uniform_buffer_0
We only found them when running full CTS runs. What happens is that we
got a hit from the on-disk shader cache, for several tests using the
same shaders. But some tests seems to be using a uniform buffer, and
others a inline buffer. Right now inline buffers leads to some changes
on the final nir shader, and generated assembly, compared with uniform
buffers. So we got a wrong shader. Fortunately we only got an assert
instead of weird behaviour.
With this commit we include the pipeline layout on the pipeline sha1,
so those two cases would get different sha1. FWIW, this is what other
drivers are already doing.
Surprisingly that didn't cause a problem before.
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16313>
Alejandro Piñeiro [Tue, 3 May 2022 11:10:30 +0000 (13:10 +0200)]
v3dv/pipeline_cache: add on disk cache hit stats
Useful when debugging/testing on disk cache.
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16313>
Alejandro Piñeiro [Tue, 3 May 2022 11:03:54 +0000 (13:03 +0200)]
v3dv/pipeline_cache: adds check to skip searching for a entry
If we are calling pipeline_cache_upload_shared_data with
from_disk_cache, that means that we had used the disk-cache to found
that entry. And that should only happens if we didn't find the entry
on the cache. So on that case we can skip to search for it.
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16313>
Alejandro Piñeiro [Tue, 3 May 2022 10:57:15 +0000 (12:57 +0200)]
v3dv/pipeline: fix small comment typo
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16313>
Jonathan Gray [Wed, 4 May 2022 06:52:08 +0000 (16:52 +1000)]
intel/dev: sync ADL-S pci ids with linux
sync ADL-S pci ids with linux
c79b846f892d ("drm/i915/adl_s: Update ADL-S PCI IDs")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16317>
Juan A. Suarez Romero [Tue, 19 Apr 2022 14:53:20 +0000 (16:53 +0200)]
v3d: disable early-Z on odd frame dimensions
The early-Z buffer may load incorrect depth values if the frame has an
od width or height. In this case we need to disable early-Z.
v3:
- Set job->early_zs_clear only for V3D_VERSION >= 40 (Iago)
- Check early-Z is disabled if no zsbuf (Iago)
v4:
- Borrow comments from v3dv around v3d_update_job_ez() (Iago)
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3557
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16166>
Juan A. Suarez Romero [Tue, 19 Apr 2022 14:52:44 +0000 (16:52 +0200)]
v3d: enable early Z/S clears
This performance optimization can be enabled if we are clearing Z/S
buffer, and not storing or loading it.
v2:
- Add assertion on depth/stencil job loads (Iago)
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16166>
Mike Blumenkrantz [Tue, 3 May 2022 19:12:45 +0000 (15:12 -0400)]
wgl: don't auto-load zink before software drivers
as in glx/egl, zink+lavapipe should only load if explicitly selected
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16311>
Mike Blumenkrantz [Mon, 2 May 2022 12:58:25 +0000 (08:58 -0400)]
zink: use VK_EXT_primitives_generated_query when available
the old codepath must be maintained, but runtime will be far simpler
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 13:05:29 +0000 (09:05 -0400)]
zink: add a flag to zink_query to trigger rasterizer discard workaround
make this agnostic of query types; no functional changes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 12:53:57 +0000 (08:53 -0400)]
zink: add and use a function to detected emulated primgen queries
no functional changes, just reducing instances of PIPE_QUERY_PRIMITIVES_GENERATED
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 12:53:00 +0000 (08:53 -0400)]
zink: pass screen param to convert_query_type()
no functional changes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 12:36:42 +0000 (08:36 -0400)]
zink: pass query object to get_num_results()
no functional changes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 12:35:07 +0000 (08:35 -0400)]
zink: pass query object to get_num_query_pools()
no functional changes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 12:31:45 +0000 (08:31 -0400)]
zink: pass query object to get_num_queries()
no functional changes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Mike Blumenkrantz [Mon, 2 May 2022 12:24:33 +0000 (08:24 -0400)]
zink: hook up VK_EXT_primitives_generated_query
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16274>
Autumn on Tape [Fri, 1 Apr 2022 20:03:40 +0000 (13:03 -0700)]
gallivm: use VPERMPS (x86/AVX2) for 32-bit 8-element shuffles
Signed-off-by: Autumn on Tape <autumn@cyfox.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13671>
Autumn on Tape [Fri, 1 Apr 2022 20:03:30 +0000 (13:03 -0700)]
gallivm: use shufflevector for shuffles when index is constant data
This checks if index is a ConstantAggregateZero, ConstantDataSequential,
or UndefValue and emits a single shufflevector instead of a loop if so.
Signed-off-by: Autumn on Tape <autumn@cyfox.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13671>
Autumn on Tape [Fri, 1 Apr 2022 20:03:20 +0000 (13:03 -0700)]
lavapipe: enable subgroup shuffle operations
Bits flipped in VkPhysicalDeviceSubgroupProperties.supportedOperations:
* SUBGROUP_FEATURE_SHUFFLE_BIT
* SUBGROUP_FEATURE_SHUFFLE_RELATIVE_BIT
Signed-off-by: Autumn on Tape <autumn@cyfox.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13671>
Autumn on Tape [Fri, 1 Apr 2022 20:02:55 +0000 (13:02 -0700)]
gallivm: add subgroup shuffle support
Signed-off-by: Autumn on Tape <autumn@cyfox.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13671>
Emma Anholt [Sat, 30 Apr 2022 14:40:59 +0000 (07:40 -0700)]
nir: Don't assert on tg4 offset range.
From the GL 4.6 spec: "If the value of any non-ignored component of the
offset vector operand is outside implementation-dependent limits, the
results of the texture lookup are undefined." We shouldn't assertion
fail, then.
GLSL-to-NIR shouldn't be enforcing limits on TG4 offsets, since it doesn't
for non-TG4 tex_src_offset either. Leave it up to the driver to handle
it.
Fixes a crash in a piglit test on nouveau that supplies a negative random
number up to 10,000 as the first coordinate for some reason. Other NIR
drivers lowered TG4 explicit offsets to tex_src_offset, so they weren't
affected.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16261>
Mike Blumenkrantz [Mon, 25 Apr 2022 20:02:14 +0000 (16:02 -0400)]
wgl: always set alpha on kopper windows
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16153>
Erik Faye-Lund [Tue, 3 May 2022 08:41:05 +0000 (10:41 +0200)]
util: fix test on msvc
We also miss this function on MSVC. But let's use the functionality in
meson to check for supported functions instead of hand-rolling the list
here.
Fixes:
067023dce2c ("util: Add some unit tests of the half-float conversions.")
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16290>
Alyssa Rosenzweig [Mon, 2 May 2022 17:59:55 +0000 (13:59 -0400)]
nir: Don't set writes_memory for reading XFB
That's a read, not a write. Fixes optimizations getting disabled for fragment
shaders when linked with a shader producing transform feedback varyings.
Fixes:
85a723975bb ("nir: add and gather shader_info::writes_memory")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16285>
Emma Anholt [Wed, 20 Apr 2022 21:31:50 +0000 (14:31 -0700)]
nir: Eliminate out-of-bounds read/writes in local lowering.
Avoids nir validation assertion failures, and it's not like backend
drivers would want to see definitely-out-of-bounds read/writes either.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16066>
David Heidelberg [Mon, 2 May 2022 07:01:22 +0000 (09:01 +0200)]
ci: traces: re-enable disabled traces which are now trimmed traces
After checksum mismatch, visually verified both re-enabled traces and
updated checksums for virgl-traces job.
Suggested-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Emma Anholt <emma@anholt.net>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16260>
David Heidelberg [Sat, 30 Apr 2022 12:31:01 +0000 (14:31 +0200)]
ci: traces: switch to Valve trimmed traces
Valve has trimmed traces already for some time, switch to
them to save network bandwidth, storage and run-time.
Reviewed-by: Emma Anholt <emma@anholt.net>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16260>
Alyssa Rosenzweig [Tue, 19 Apr 2022 22:58:27 +0000 (18:58 -0400)]
pan/bi: Use a dynarray for predecessors
This is deterministic, unlike a set. Note we need the extra dereferencing to
keep the macro safe, simple, and standards compliant:
1. Nesting two for-loops would cause break/continue to fail.
2. Declaring variables outside the loop would pollute the namespace.
3. Declaring an anonymous struct is not conformant and doesn't compile in clang.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 18:08:44 +0000 (14:08 -0400)]
pan/bi: Use worklist for scoreboard analysis
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 18:06:09 +0000 (14:06 -0400)]
pan/bi: Use bi_worklist for post-RA liveness
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 18:04:20 +0000 (14:04 -0400)]
pan/bi: Use bi_worklist for liveness
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 17:35:19 +0000 (13:35 -0400)]
pan/bi: Use bi_worklist in analyze_helper_requirements
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 18:00:10 +0000 (14:00 -0400)]
pan/bi: Add u_worklist wrapper macros
..expanding to bi_block.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 18:00:01 +0000 (14:00 -0400)]
pan/bi: Count blocks
For u_worklist.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Alyssa Rosenzweig [Tue, 19 Apr 2022 17:59:31 +0000 (13:59 -0400)]
pan/bi: Rename bi_block->name to bi_block->index
This is consistent with nir_block and (IMO) less confusing.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16279>
Lionel Landwerlin [Fri, 29 Apr 2022 21:16:20 +0000 (00:16 +0300)]
docs: explain state emission in Anv
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 07:54:27 +0000 (10:54 +0300)]
anv: remove static_state_mask
This is now unnecessary. Either an instruction is never dynamic and
it's emitted in genX_pipeline.c or it can be and it's emitted in
genX_cmd_buffer.c/gfx8_cmd_buffer/gfx7_cmd_buffer.c
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 08:35:35 +0000 (11:35 +0300)]
anv: don't emit 3DSTATE_VF_TOPOLOGY in pipeline batch
v2: drop primitive_topology = 0xffffffff (Tapani)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 07:42:44 +0000 (10:42 +0300)]
anv: don't emit 3DSTATE_DEPTH_BOUNDS in pipeline batch
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Wed, 27 Apr 2022 21:12:52 +0000 (00:12 +0300)]
anv: don't emit 3DSTATE_BLEND_STATE_POINTERS in pipeline batch
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 06:26:39 +0000 (09:26 +0300)]
anv: don't emit 3DSTATE_WM in pipeline batch
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 06:23:23 +0000 (09:23 +0300)]
anv: don't emit 3DSTATE_STREAMOUT in pipeline batch
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 06:15:02 +0000 (09:15 +0300)]
anv: never emit 3DSTATE_CPS in the pipeline batch
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Wed, 27 Apr 2022 20:44:40 +0000 (23:44 +0300)]
anv: rework sample location
On Gfx7 we can only give the sample location for a given multisample
number. This means everytime the multisampling value changes, we have
to re-emit the locations. It's fine because it's also where
(3DSTATE_MULTISAMPLE) the number of samples is stored.
On Gfx8+ though, 3DSTATE_MULTISAMPLE only holds the number of samples
and all the sample locations for all number of samples are located in
3DSTATE_SAMPLE_PATTERN. So to be more effecient there, we need to
track the locations for all sample numbers and compare new values with
the relevant sample count when touching the dynamic state.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Wed, 27 Apr 2022 19:38:28 +0000 (22:38 +0300)]
Revert "anv: fix dynamic state emission"
This reverts commit
f348103fce4484b891a58bf3147b71e3a3558135. The
change was causing performance regressions.
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Fri, 29 Apr 2022 20:08:42 +0000 (23:08 +0300)]
anv: add missing logic op set in pipeline dyn state
v2: add ANV_CMD_DIRTY_DYNAMIC_LOGIC_OP check (Tapani)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes:
75ad0e4b0874 ("anv: support blending logic op dynamic state")
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Lionel Landwerlin [Thu, 28 Apr 2022 10:15:53 +0000 (13:15 +0300)]
anv: reset all dynamic state after secondary execution
We don't know in what state the secondary buffer will leave the HW
when it ends. It's easier to consider everything needs to be reemitted
for now.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16220>
Fabrice Fontaine [Sat, 21 Aug 2021 15:55:33 +0000 (17:55 +0200)]
src/util/futex.h: fix build on 32-bit architectures using 64-bit time_t
Fix the following build failure on 32-bit architectures using 64-bit
time_t (e.g. riscv32):
../src/util/futex.h: In function 'sys_futex':
../src/util/futex.h:39:19: error: 'SYS_futex' undeclared (first use in this function); did you mean 'sys_futex'?
39 | return syscall(SYS_futex, addr1, op, val1, timeout, addr2, val3);
| ^~~~~~~~~
| sys_futex
Fixes:
- http://autobuild.buildroot.org/results/
692700a5f967760a0b8cd358b1712f1d5a7b681e
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12496>
Marek Olšák [Mon, 25 Apr 2022 07:54:49 +0000 (03:54 -0400)]
radeonsi: remove si_create_surface_custom
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 07:52:59 +0000 (03:52 -0400)]
radeonsi: remove si_create_sampler_view_custom and related code
This was used for compressed and subsampled gfx copies.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 07:45:08 +0000 (03:45 -0400)]
radeonsi: remove compressed and subsampled gfx copy from resource_copy_region
This is unused because the compute copy is always used in these cases.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 10:44:25 +0000 (06:44 -0400)]
radeonsi: improve the subsampled copy_image calculation
This fixes a few cases for AMD_TEST=copyimage, but it's still pretty broken.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 07:31:50 +0000 (03:31 -0400)]
radeonsi: handle compressed formats in si_compute_copy_image
This fixes a lot of AMD_TEST=copyimage cases, but there are still some
failures.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 14:45:44 +0000 (10:45 -0400)]
radeonsi/ci: update sienna_cichlid results
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:37:50 +0000 (07:37 -0400)]
radeonsi/test_image_copy: test all formats
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:37:50 +0000 (07:37 -0400)]
radeonsi/test_image_copy: test mipmaps
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:37:50 +0000 (07:37 -0400)]
radeonsi/test_image_copy: test copying a block-compressed format
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:37:50 +0000 (07:37 -0400)]
radeonsi/test_image_copy: test all interesting texture targets
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:37:50 +0000 (07:37 -0400)]
radeonsi/test_image_copy: separate printing pipe_resource into a function
also print texture targets that are going to be tested
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:37:50 +0000 (07:37 -0400)]
radeonsi/test_image_copy: separate generating a random pipe_resource
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:19:13 +0000 (07:19 -0400)]
radeonsi/test_image_copy: simplify texture size generation
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 11:13:15 +0000 (07:13 -0400)]
radeonsi/test_image_copy: remove special SDMA codepaths
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 10:36:27 +0000 (06:36 -0400)]
radeonsi/test_image_copy: remove linear and power-of-two size testing
it doesn't add any value
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 10:20:14 +0000 (06:20 -0400)]
radeonsi/test_image_copy: remove whole image copies
it doesn't add any value
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 08:25:37 +0000 (04:25 -0400)]
radeonsi: rename si_test_blit -> si_test_image_copy_region
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 09:06:29 +0000 (05:06 -0400)]
radeonsi: don't expose formats that CB supports, but not shaders
e.g. R4G4B4A4_UINT
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 09:05:59 +0000 (05:05 -0400)]
radeonsi: assert format support in si_compute_copy_image
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 07:14:50 +0000 (03:14 -0400)]
radeonsi: cosmetic changes in si_compute_copy_image
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 04:04:41 +0000 (00:04 -0400)]
radeonsi: implement compute_copy_image between 1D_ARRAY and other texture types
And set more optimal compute block sizes.
The compute copy is required to preserve NaNs, so this fixes a lot of
AMD_TEST=copyimage cases.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 03:06:43 +0000 (23:06 -0400)]
radeonsi: add color buffer support for R9G9B9E5_FLOAT on gfx10.3
This was missed when we added gfx10.3.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 03:05:37 +0000 (23:05 -0400)]
radeonsi: add color buffer support for 2_10_10_10 pipe formats
The hw enum is reversed.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 02:19:39 +0000 (22:19 -0400)]
radeonsi: don't allow broken compute image copies into A8R8_UNORM
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 02:18:51 +0000 (22:18 -0400)]
radeonsi: don't report R64_*INT as a sampler format because it doesn't work
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 08:18:01 +0000 (04:18 -0400)]
radeonsi: add si_can_use_compute_blit
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Sun, 24 Apr 2022 08:17:42 +0000 (04:17 -0400)]
radeonsi: code style fix
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Thu, 28 Apr 2022 10:14:57 +0000 (06:14 -0400)]
radeonsi: simplify setting COMPUTE_PGM_RSRC2.SCRATCH_EN
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Thu, 28 Apr 2022 10:01:15 +0000 (06:01 -0400)]
radeonsi: remove unused si_cs_shader_state::uses_scratch
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Wed, 27 Apr 2022 06:30:49 +0000 (02:30 -0400)]
radeonsi: align the tessellation ring address to 2MB
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Wed, 27 Apr 2022 06:30:00 +0000 (02:30 -0400)]
radeonsi: cleanup local includes
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Wed, 27 Apr 2022 02:29:31 +0000 (22:29 -0400)]
radeonsi: always print the lower-case GPU name for radeonsi-run-tests.py
This fixes the GPU name detection in radeonsi-run-tests.py when
the marketing name is unknown.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 01:20:33 +0000 (21:20 -0400)]
ac,radeonsi: report SCALED formats as unsupported by samplers and color buffers
This was never exercised and it doesn't work.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Mon, 25 Apr 2022 02:18:20 +0000 (22:18 -0400)]
ac: support GR channel order in ac_choose_spi_color_formats
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Wed, 27 Apr 2022 03:22:40 +0000 (23:22 -0400)]
ac/llvm: don't set ac_dlc in visit_intrinsic
It's set by get_load_cache_policy later.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Wed, 27 Apr 2022 03:21:08 +0000 (23:21 -0400)]
ac/llvm: set the correct cache policy for sparse buffer loads
The DLC bit was missing on gfx10.
Fixes:
6d5e26752c664c "ac/nir: implement sparse image/texture loads"
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Thu, 28 Apr 2022 05:14:39 +0000 (01:14 -0400)]
ac: remove really_needs_scratch, parameter from ac_parse_shader_binary_config
it's always true
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Marek Olšák [Wed, 27 Apr 2022 00:50:21 +0000 (20:50 -0400)]
ac/gpu_info: rework how num_se is derived
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16215>
Alyssa Rosenzweig [Wed, 27 Apr 2022 18:59:41 +0000 (14:59 -0400)]
panfrost: Unit test "from legacy" helper
So we don't regress the UABI. This doesn't get much CI coverage otherwise.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16201>
Alyssa Rosenzweig [Wed, 27 Apr 2022 18:50:59 +0000 (14:50 -0400)]
panfrost: Unit test stride calculations
These have reasonable interpretations now, and the three row strides have been
deduplicated. So add stride expectations to our ASTC unit tests.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16201>
Alyssa Rosenzweig [Wed, 27 Apr 2022 18:09:01 +0000 (14:09 -0400)]
panfrost: Unify row stride and AFBC row stride
Row stride is defined in terms of header blocks for AFBC. Usually,
afbc.row_stride is used for AFBC images and row_stride for non-AFBC images;
however, the nonsense non-AFBC stride leaked into the UABI. So handle that in
the legacy conversion path and use a unified row stride (equal to
afbc.row_stride for AFBC images and row_stride otherwise).
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16201>
Alyssa Rosenzweig [Wed, 27 Apr 2022 16:12:05 +0000 (12:12 -0400)]
panfrost: Remove line_stride
There are no more users. This eliminates a class of issues.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16201>
Alyssa Rosenzweig [Wed, 27 Apr 2022 16:04:54 +0000 (12:04 -0400)]
lima,panfrost: Use row stride for tiling routines
This makes more sense.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16201>
Alyssa Rosenzweig [Wed, 27 Apr 2022 16:10:37 +0000 (12:10 -0400)]
panfrost: Use row_stride even for linear resources
In that case, they match up.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16201>
projects / platform / upstream / mesa.git / log