Lionel Landwerlin [Thu, 18 Mar 2021 09:04:49 +0000 (11:04 +0200)]
intel/mi_builder: use device info to use the right CS prefetch size
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9679>
Lionel Landwerlin [Thu, 18 Mar 2021 09:03:35 +0000 (11:03 +0200)]
intel/dev: store size of CS prefetch
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9679>
Dave Airlie [Thu, 18 Mar 2021 04:09:50 +0000 (14:09 +1000)]
lavapipe: fix templated descriptor updates
The template path was buggy but CTS only tested it with Vulkan 1.1 enabled.
It was just missing the dstArrayElement offset.
Fixes:
41f7fa273d21 ("lavapipe: add support for VK_KHR_descriptor_update_template")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9675>
Dave Airlie [Sun, 14 Mar 2021 20:14:14 +0000 (06:14 +1000)]
lavapipe: drop unused vk_format in image struct
This wasn't used at all.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9675>
Dave Airlie [Tue, 9 Mar 2021 05:25:48 +0000 (15:25 +1000)]
lavapipe: add missing device group api
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9675>
Dave Airlie [Wed, 17 Mar 2021 05:31:33 +0000 (15:31 +1000)]
lavapipe: add missing sampler minmax properties
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9675>
Dave Airlie [Tue, 22 Dec 2020 07:24:19 +0000 (17:24 +1000)]
lavapipe: fix image format properties
Fix the sample counts for compressed + non-renderable + cube compatible
Fix the mip levels max calcs
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9675>
Eric Anholt [Tue, 16 Mar 2021 19:00:01 +0000 (12:00 -0700)]
freedreno/ir3: Demote centroid usage to pixel on non-msaa.
Like with the sample qualifier on all GPUs, use pixel on older HW when
MSAA rasterization is disabled to get reliable results. Since I ran many
CI jobs on this, this updates the A530 TF flakes list, though I don't
think that this MR necessarily made it flakier (we were already struggling
on a5xx TF, which was what was motivating me to look at this!)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9641>
Timur Kristóf [Thu, 18 Mar 2021 15:53:26 +0000 (16:53 +0100)]
ac/llvm: Fix alignment of shared load intrinsics.
After some research, I found out that the alignment needs to be set
on the return value of the load, rather than on the pointer.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Michel Dänzer <mdaenzer@redhat.com>
Fixes:
14ad82b4e98345214973d76b7395498b4a65292e "ac/llvm: Emit more efficient code for load_shared."
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9688>
Daniel Schürmann [Thu, 11 Mar 2021 12:45:45 +0000 (13:45 +0100)]
aco/ra: allow VCC on SMEM sbase operand on GFX10+
Totals from 26 (0.02% of 146267) affected shaders (Navi10):
CodeSize: 324764 -> 324560 (-0.06%)
Instrs: 61090 -> 61039 (-0.08%)
Cycles: 2681952 -> 2668872 (-0.49%)
VMEM: 5773 -> 5769 (-0.07%)
SMEM: 1621 -> 1611 (-0.62%)
SClause: 2546 -> 2543 (-0.12%)
Copies: 4671 -> 4620 (-1.09%)
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9531>
Daniel Schürmann [Thu, 11 Mar 2021 12:42:50 +0000 (13:42 +0100)]
aco/optimizer: set VCC hint on new v_cmp_* definitions
Totals from 11692 (7.99% of 146267) affected shaders (Navi10):
CodeSize:
97419384 ->
97352560 (-0.07%); split: -0.07%, +0.00%
Instrs:
18571138 ->
18570969 (-0.00%); split: -0.00%, +0.00%
Cycles:
1431348400 ->
1431346296 (-0.00%); split: -0.00%, +0.00%
SMEM: 696646 -> 696650 (+0.00%)
SClause: 668511 -> 668490 (-0.00%); split: -0.00%, +0.00%
Copies: 1279475 -> 1279474 (-0.00%)
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9531>
Juan A. Suarez Romero [Wed, 17 Mar 2021 12:29:37 +0000 (13:29 +0100)]
ci/broadcom: use new piglit runner
Switch from the old piglit to the new piglit-runner executor.
Acked-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9680>
Vasily Khoruzhick [Sun, 14 Mar 2021 17:10:54 +0000 (10:10 -0700)]
lima: update dEQP fails and skips lists
Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Andreas Baierl <ichgeh@imkreisrum.de>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9586>
Eric Anholt [Tue, 16 Feb 2021 22:57:40 +0000 (14:57 -0800)]
freedreno/a5xx: Fix stream-output binning handling.
This makes it match (to the best I was able) a6xx's behavior, with the
exception of the XXX note in fd5_gmem.c highlighting what I think is an
issue on both a5xx and a6xx.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9295>
Jesse Natalie [Wed, 17 Mar 2021 22:05:35 +0000 (15:05 -0700)]
u_format: Add restrict to fn pointer and manual format pack/unpack/fetch
MSVC warns (loudly) about a mismatch between the generated functions in
u_format_table.c and the definition of util_format_[un]pack_description,
specifically having 'restrict' in the function but not in the pointer
type in the struct.
So, add 'restrict' everywhere - to the function types, and to the rest
of the implementations that are assigned to those structs.
v2: util_format_unpack_description is used in gallium/auxiliary/translate
Fixes:
5785fdac ("u_format: Mark the generated pack/unpack src/dst args as restrict.")
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9670>
Mike Blumenkrantz [Thu, 18 Mar 2021 13:49:13 +0000 (09:49 -0400)]
zink: handle nir_intrinsic_image_deref_samples
same as the texop but for images
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9682>
Mike Blumenkrantz [Thu, 18 Mar 2021 13:47:54 +0000 (09:47 -0400)]
compiler/spirv: fix image sample queries
this was only implemented for textures (I assume because drivers which implement
the corresponding intrinsic don't support multisampled images), but it's also
used for shader images
Fixes:
22fdb2f8551 ("nir/spirv: Update to the latest revision")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9682>
Mike Blumenkrantz [Thu, 18 Mar 2021 14:00:30 +0000 (10:00 -0400)]
zink: enable spirv extension for post depth coverage
Fixes:
3c72c867422 ("zink: Wire up ARB_post_depth_coverage")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9683>
Timur Kristóf [Thu, 18 Mar 2021 09:15:14 +0000 (10:15 +0100)]
aco: Fix constant address offset calculation for ds_read2 instructions.
Cc: mesa-stable
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9678>
Erik Faye-Lund [Tue, 16 Mar 2021 17:40:01 +0000 (18:40 +0100)]
gallium/st: reserve space in default uniform block for lowered constants
If we don't reserve these, we risk these lowering passes eating up more
uniforms than we have available.
This fixes a crash due to an assert in Zink, because we end up trying to
use a too large UBO after lowering.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9631>
Erik Faye-Lund [Tue, 16 Mar 2021 18:04:08 +0000 (19:04 +0100)]
gallium/st: fix shader_has_one_variant
I think we need to care about this here as well, otherwise the variant
logic might be short-circuited.
Fixes:
7eb5fd98fd1 ("mesa/st: handle running nir lower passes for ucp and psiz in tess stage")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9631>
Erik Faye-Lund [Tue, 16 Mar 2021 17:43:37 +0000 (18:43 +0100)]
gallium/st: fix shader_has_one_variant
I think we need to care about this here as well, otherwise the variant
logic might be short-circuited.
Fixes:
ad0037fcfe1 ("mesa/st: set lower_point_size for tes/gs during program update")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9631>
Neha Bhende [Tue, 16 Mar 2021 22:45:26 +0000 (15:45 -0700)]
nir_to_tgsi: Fix indices for CMP in nir_to_tgsi for nir_op_fcsel
nir_to_tgsi utiliy was using wrong src for nir_op_fcsel.
Fixes regression with solidworks2012_viewport.trace and mudbox-2010sp1-basic-head.trace
v2: as per suggested by Brian, instead of relying assumption that
value will be 0.0 or 1.0 all the time, use -abs(src) to make any non-zero
value as negative.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9658>
Eric Anholt [Wed, 17 Mar 2021 17:39:35 +0000 (10:39 -0700)]
nir_to_tgsi: Respect PIPE_SHADER_CAP_TGSI_SQRT_SUPPORTED.
If the driver can't do it, we shouldn't be putting it in the TGSI,
regardless of what the NIR compiler above us decided.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Neha Bhende <bhenden@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9662>
Mike Blumenkrantz [Wed, 17 Mar 2021 18:40:40 +0000 (14:40 -0400)]
zink: clear framebuffer state on context destroy
ensure we don't leak surface references
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Tue, 16 Mar 2021 18:38:27 +0000 (14:38 -0400)]
zink: move fence reset to zink_fence_init()
fence stuff goes in fence file
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Tue, 16 Mar 2021 18:34:46 +0000 (14:34 -0400)]
zink: remove query batch-tracking init from begin_query()
this is always created on startup now
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Tue, 16 Mar 2021 18:29:13 +0000 (14:29 -0400)]
zink: use macro to streamline batch struct member init
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Tue, 16 Mar 2021 17:56:46 +0000 (13:56 -0400)]
zink: move other batch-tracking implementations to unified codepath
prep for monotonic batch ids
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Thu, 5 Nov 2020 17:32:32 +0000 (12:32 -0500)]
zink: also move batch destructor into zink_batch.c
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Thu, 5 Nov 2020 17:30:07 +0000 (12:30 -0500)]
zink: move batch init into zink_batch.c
this consolidates more batch code into the batch file
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Thu, 5 Nov 2020 17:10:57 +0000 (12:10 -0500)]
zink: move active query pruning to batch reset
this didn't ever really need to be on the fence in the first place
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Thu, 5 Nov 2020 17:05:15 +0000 (12:05 -0500)]
zink: split out batch resource-set clearing into separate function
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Wed, 28 Oct 2020 18:15:28 +0000 (14:15 -0400)]
zink: return enum zink_queue from zink_batch_reference_resource_rw()
this is a little cleaner
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Wed, 28 Oct 2020 17:21:34 +0000 (13:21 -0400)]
zink: abstract zink_get_resource_usage() and move it to be internal
I'll be rewriting how resource tracking works, so abstracting it and removing
direct uses is going to reduce the chances of breaking things as well as code churn
plus it's a bit easier to use
downside is that until that rewrite happens, this will be a (very small) perf hit and
there's some kinda gross macros involved to consolidate code
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Wed, 28 Oct 2020 17:19:53 +0000 (13:19 -0400)]
zink: convert ZINK_RESOURCE_ACCESS defines to enum
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Wed, 28 Oct 2020 17:18:55 +0000 (13:18 -0400)]
zink: refactor resource_sync_writes_from_batch_usage() to manage batch id internally
this is always used the same way, so we can simplify the code a bit for future use
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Wed, 28 Oct 2020 17:17:52 +0000 (13:17 -0400)]
zink: add enum for different queues
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Tue, 27 Oct 2020 19:01:17 +0000 (15:01 -0400)]
zink: add a pipe_context::fence_server_sync hook
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9626>
Mike Blumenkrantz [Fri, 30 Oct 2020 12:18:31 +0000 (08:18 -0400)]
zink: more consolidation for null sampler/image view hashing
by using the new helper functions, this is even more consistent
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9625>
Mike Blumenkrantz [Thu, 18 Mar 2021 00:52:36 +0000 (20:52 -0400)]
zink: ci updates
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9673>
Mike Blumenkrantz [Thu, 18 Mar 2021 00:37:44 +0000 (20:37 -0400)]
zink: always use requested format for sampler view creation
this should be fine and good
Fixes:
c768c5297a8 ("zink: force stencil format for stencil-only samplers and swizzle the right component")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9673>
Neha Bhende [Tue, 16 Mar 2021 22:20:28 +0000 (15:20 -0700)]
gallium/u_vbuf: use updated pipe_draw_start_count while using draw_vbo
new_draw has updated draw count. Stale info in draw caused regression
with piglit gl-3.0-multidrawarrays-vertexid -indirect
fixes piglit test ./gl-3.0-multidrawarrays-vertexid -indirect
Fixes:
1cd455b17b7a ("gallium: extend draw_vbo to support multi draws")
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9656>
Mauro Rossi [Mon, 15 Mar 2021 22:39:26 +0000 (23:39 +0100)]
android: util: create some standalone compression helpers
Fixes the following building errors:
FAILED: out/target/product/x86_64/obj/SHARED_LIBRARIES/vulkan.android-x86_intermediates/LINKED/vulkan.radv.so
...
ld.lld: error: undefined symbol: util_compress_inflate
>>> referenced by disk_cache_os.c:459 (external/mesa/src/util/disk_cache_os.c:459)
...
ld.lld: error: undefined symbol: util_compress_max_compressed_len
>>> referenced by disk_cache_os.c:614 (external/mesa/src/util/disk_cache_os.c:614)
...
ld.lld: error: undefined symbol: util_compress_deflate
>>> referenced by disk_cache_os.c:622 (external/mesa/src/util/disk_cache_os.c:622)
Fixes:
d7ecbd5bf837 ("util: create some standalone compression helpers")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9613>
Neha Bhende [Tue, 16 Mar 2021 22:48:31 +0000 (15:48 -0700)]
mesa: set states in fast path for restoring light attributes
Since states were not updated in fast path for restoring light attributes,
seen darker images in solidworks2012_viewprt.trace
Fixes regression seen with solidworks2012_viewport.trace
Fixes:
7fa9d9d06c4 ("mesa: add a fast path for restoring light attributes")
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9657>
Jordan Justen [Wed, 17 Mar 2021 17:08:56 +0000 (10:08 -0700)]
anv: Use fallback paths if DRM_I915_QUERY_ENGINE_INFO fails
Anvil can handle if this call fails, but not if we assert. :)
Reported-by: Brian Paul <brianp@vmware.com>
Fixes:
5d84c764fde ("anv: Gather engine info from i915 if available")
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9664>
Jason Ekstrand [Tue, 12 Jan 2021 04:18:11 +0000 (22:18 -0600)]
intel/fs: Add support for 16-bit A64 float and integer atomics
The messages for those 16-bit operations still use 32-bit sources and
destinations, so expand them accordingly when building the payload.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8750>
Jason Ekstrand [Thu, 27 Aug 2020 06:11:25 +0000 (01:11 -0500)]
spirv: Add support for SPV_EXT_shader_atomic_float_min_max
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8750>
Jason Ekstrand [Thu, 27 Aug 2020 18:11:54 +0000 (13:11 -0500)]
nir: Add image atomic_fmin/fmax intrinsics
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8750>
Caio Marcelo de Oliveira Filho [Wed, 17 Mar 2021 21:20:14 +0000 (14:20 -0700)]
nir: Handle deref_atomic_fadd in a couple of passes
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8750>
Caio Marcelo de Oliveira Filho [Wed, 17 Mar 2021 04:20:12 +0000 (21:20 -0700)]
spirv: Update headers and metadata from latest Khronos commit
This corresponds to
bcf55210f13a4fa3c3d0963b509ff1070e434c79
("Merge pull request #178 from orbea/datadir") in
https://github.com/KhronosGroup/SPIRV-Headers.
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8750>
Dave Airlie [Wed, 17 Mar 2021 03:33:14 +0000 (13:33 +1000)]
lavapipe: bump maxMemoryAllocationCount
not sure why this was so low
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9644>
Marek Olšák [Wed, 24 Feb 2021 18:11:07 +0000 (13:11 -0500)]
st/mesa: add a driconf option to transcode ETC2 to DXTC
for performance analysis
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9357>
Lionel Landwerlin [Wed, 17 Mar 2021 19:30:52 +0000 (21:30 +0200)]
intel/fs/vec4: add missing dependency in write-on-write fixed GRFs
If we load constant data using pull constant SENDS, and we later load
that register with some other data, we can end up in a situation where
we don't track the initial fixed register write and therefore end up
using uninitialized registers.
This tracks write-on-write of fixed GRFs like we do for normal virtual
GRFs.
v2: Fix post_alloc_reg case (Jason)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: <mesa-stable@lists.freedesktop.org>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9667>
Adam Jackson [Fri, 12 Mar 2021 19:44:29 +0000 (14:44 -0500)]
zink: Wire up ARB_post_depth_coverage
Just a matter of passing the bits through in the right place.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9637>
Marek Olšák [Mon, 15 Mar 2021 13:58:19 +0000 (09:58 -0400)]
mesa: fix parameter reservation size
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9360>
Marek Olšák [Tue, 2 Mar 2021 06:07:10 +0000 (01:07 -0500)]
mesa: clear reserved parameter storage because it's stored in the shader cache
The elements might not be initialized and we don't want random bytes
in the shader cache.
Discovered by valgrind.
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9360>
Marek Olšák [Tue, 2 Mar 2021 09:19:53 +0000 (04:19 -0500)]
mesa: don't overallocate ParameterValues 4 times (v2)
The additional memory was never used.
v2: rework
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9360>
Marek Olšák [Tue, 2 Mar 2021 09:17:46 +0000 (04:17 -0500)]
mesa: fix a oldNum typo in reallocation in _mesa_reserve_parameter_storage
oldNum was incorrect. oldValNum is the correct number of elements
to copy inside realloc. (oldNum is for Parameters, not ParameterValues)
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9360>
Marek Olšák [Tue, 2 Mar 2021 05:56:29 +0000 (00:56 -0500)]
mesa: add assertions for buffer reference counts
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9360>
Marek Olšák [Tue, 2 Mar 2021 06:38:26 +0000 (01:38 -0500)]
mesa: fix Blender crash due to optimizations in buffer reference counting
The problem was that I assumed that deleted zombie buffers can't have any
references in the context, so buffers were released sooner than they should
have been.
The fix is to count the non-atomic references in the new field
gl_buffer_object::CtxRefCount. When we detach the context from the buffer,
we can just add CtxRefCount to RefCount to re-enable atomic reference
counting. This also allows removing code that was doing a similar thing.
Fixes:
e014e3b6be63 "mesa: don't count buffer references for the context that created them"
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4259
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9360>
Marek Olšák [Sat, 13 Mar 2021 09:11:36 +0000 (04:11 -0500)]
radeonsi: use pipe_sampler_state::border_color_is_integer to simplify stuff
We don't need the separate integer sampler state if we know the border
color type.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9577>
Michel Dänzer [Wed, 17 Mar 2021 10:41:31 +0000 (11:41 +0100)]
ci: Don't run meson tests in strace for meson-mingw32-x86_64 job
There have been repeated timeouts:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/3437#note_842273
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9648>
Axel Davy [Sun, 14 Mar 2021 18:31:48 +0000 (19:31 +0100)]
radeonsi: fix leak when the in-memory cache is full
When the hw_binary is not put in the in-memory
cache it must be freed.
Fixes:
8283ed65cfd ("radeonsi: Limit the size of the in-memory shader cache")
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9587>
Simon Ser [Wed, 17 Mar 2021 08:09:48 +0000 (09:09 +0100)]
gbm: remove fprintf calls in gbm_dri_bo_create
These errors can be handled by the caller. The caller can't guess
whether the GBM implementation supports modifiers, for instance.
Signed-off-by: Simon Ser <contact@emersion.fr>
References: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7601#note_778845
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8715>
Simon Ser [Tue, 26 Jan 2021 12:46:25 +0000 (13:46 +0100)]
gbm: fail early when modifier list only contains INVALID
The current check only accomodates for a list with a single INVALID
item. However the driver won't be able to pick any modifier if the
list only contains INVALID. This includes the following cases:
- The modifier list is empty (count == 0)
- The modifier list contains more than a single item, but all items
are INVALID
In these cases, also fail early.
Signed-off-by: Simon Ser <contact@emersion.fr>
References: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7601#note_778845
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8715>
Mike Blumenkrantz [Tue, 16 Mar 2021 18:10:49 +0000 (14:10 -0400)]
zink: switch to deqp-runner for piglit jobs
There's a few changes in the expected results. First of all, there's a
few failures that are now interpreted as crashes. These test are:
- glx@glx-visuals-depth
- glx@glx-visuals-depth -pixmap
- glx@glx-visuals-stencil
- glx@glx-visuals-stencil -pixmap
Secondly, and more surprisingly, there's three tests that were
previously passing, but are now failing. These are all EGL-related, so
it's likely that there's some EGL interaction that is different with the
new runner. These tests are:
- spec@egl 1.4@eglterminate then unbind context
- spec@egl_khr_surfaceless_context@viewport
- spec@egl_mesa_configless_context@basic
commit log and skiplist by Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9630>
Eric Anholt [Wed, 17 Mar 2021 17:10:17 +0000 (10:10 -0700)]
ci/freedreno: Mark the rest of the glx_arb_sync_control@timing as flakes.
IIRC I've seen -msc-delta 2 flake set, so just complete the set.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9660>
Eric Anholt [Wed, 17 Mar 2021 17:09:06 +0000 (10:09 -0700)]
ci/freedreno: Mark an a630 piglit flake from async shader compiling.
It seems that right around when we enabled piglit, some timing also
changed so this one started flaking.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9660>
Jason Ekstrand [Tue, 16 Mar 2021 15:08:21 +0000 (10:08 -0500)]
anv/apply_pipeline_layout: Add support for A64 descriptor access
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 22:44:44 +0000 (16:44 -0600)]
anv: Do UBO loads with global addresses for bindless
This makes UBO loads in the variable pointers or bindless case work just
like SSBO loads in the sense that they use A64 messages and 64-bit
global addresses. The primary difference is that we have an
optimization in anv_nir_lower_ubo_loads which uses a (possibly
predicated) block load message when the offset is constant so we get
roughly the same performance as we would from plumbing load_ubo all the
way to the back-end.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 20:59:42 +0000 (14:59 -0600)]
anv: Add a pass for lowering A64 UBO access
Instead of load_global_constant_offset/bounded, we want to use the
Intel-specific block load intrinsic whenever we can. This way we get
the same wide block loads that we usually use for constant offset UBO
pulls with a binding table.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 06:35:19 +0000 (00:35 -0600)]
nir/lower_io: Support global addresses for UBOs in nir_lower_explicit_io
For nir_address_format_64bit_global_32bit_offset and
nir_address_format_64bit_bounded_global, we use a new intrinsics which
take the base address and offset as separate parameters. For bounds-
checked access, the bound is also included in the intrinsic. This gives
the drive more control over the bounds checking so that UBOs don't
suddenly become massively more expensive.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Sun, 14 Mar 2021 05:58:13 +0000 (23:58 -0600)]
anv/apply_pipeline_layout: Use the new helpers for images
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 12 Mar 2021 23:35:32 +0000 (17:35 -0600)]
anv/apply_pipeline_layout: Use the new helpers for early lowering
This also means that some of the newly added helpers need to grow a bit
to support VK_DESCRIPTOR_TYPE_INLINE_UNIFORM_DATA_EXT.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 12 Mar 2021 23:49:40 +0000 (17:49 -0600)]
anv/apply_pipeline_layout: Rework the desc_addr_format helper
We're about to add a new helper which is more detailed.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 12 Mar 2021 21:25:13 +0000 (15:25 -0600)]
anv/apply_pipeline_layout: Refactor all our descriptor address builders
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Wed, 20 Jan 2021 21:59:23 +0000 (15:59 -0600)]
anv/apply_pipeline_layout: Apply dynamic offsets in load_ssbo_descriptor
This function has exactly two call sites. The first is where we had
these calculations before. The second only cares about the size of the
SSBO so all the extra code we emit will be dead. However, NIR should
easily clean that up and this lets us consolidate things a bit better.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 23:20:22 +0000 (17:20 -0600)]
anv: Zero out the last dword of UBO/SSBO descriptors in the shader
This way, NIR can constant fold it.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Sat, 13 Mar 2021 00:27:28 +0000 (18:27 -0600)]
anv: Rework the 64bit_bounded_global resource index format
Instead of packing the descriptor offset into the packed portion, use
that unused channel we have lying around. This potentially allows for
larger descriptor sets. We also re-arrange the components a bit to make
it more like the 64bit_bounded_global memory address format.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Wed, 20 Jan 2021 21:07:23 +0000 (15:07 -0600)]
anv: Use 64bit_global_32bit_offset for SSBOs
This has the advantage of giving us cheaper address calculations because
we can calculate in 32 bits first and then do a single 64x32 add. It
also lets us delete a bunch of code for dealing with descriptor
dereferences (vulkan_resource_reindex, and friends) because our bindless
SSBO pointers are now vec4s regardless of whether or not we're doing
bounds checking. This also unifies UBOs and SSBOs. The one down-side
is that, in certain variable pointers cases, it may end up burning more
memory and/or increasing register pressure. This seems like a worth-
while trade-off.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Wed, 20 Jan 2021 20:27:32 +0000 (14:27 -0600)]
nir: Add a new 64+32-bit address format
This is a global address format where you have a 64-bit base pointer and
a 32-bit offset. It's intentionally identical to 64bit_bounded_global
except nir_lower_explicit_io does no bounds checking with it.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Thu, 21 Jan 2021 23:31:50 +0000 (17:31 -0600)]
anv/apply_pipeline_layout: Add some switch statements
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 22:40:41 +0000 (16:40 -0600)]
anv/apply_pipeline_layout: Plumb through a UBO address format
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Tue, 9 Mar 2021 23:32:14 +0000 (17:32 -0600)]
anv/apply_pipeline_layout: Move bounds checking later for index/offset
Instead of doing the array check at the load_vulkan_resource_index
intrinsic, stuff it in the vec2 and handle it at load_vulkan_descriptor
time. This allows the bounds check to take any re-index intrinsics into
account. This only affects variablePointers + SSBOs + Gen7.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 12 Mar 2021 20:04:38 +0000 (14:04 -0600)]
anv/apply_pipeline_layout: Run DCE between the early and late passes
This allows us to ignore UBOs in the late code going forward.
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 06:18:37 +0000 (00:18 -0600)]
anv/apply_pipeline_layout: Lower UBO loads in the early pass
We're about to enable bindless UBOs via A64 memory access like we do for
SSBOs. In order to prevent 100% of UBOs from hitting that path, we
enable them in the early lowering. This way we'll still get binding
table-based UBO access for any non-bindless ones. In particular, we
need this for UBO pushing to work.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Thu, 21 Jan 2021 01:42:16 +0000 (19:42 -0600)]
anv/apply_pipeline_layout: Rework the early pass index/offset helpers
Rewrite them all to work on an index/offset vec2 instead of some only
returning the index. This means SSBO size handling is a tiny bit more
complicated but it will also mean we can use them for descriptor buffers
properly.
This also fixes a bug where we weren't bounds-checking re-index
intrinsics because we applied the bounds check at the tail of the
recursion and not at the beginning.
Fixes:
3cf78ec2bd "anv: Lower some SSBO operations in..."
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Thu, 21 Jan 2021 00:49:26 +0000 (18:49 -0600)]
anv/apply_pipeline_layout: Refactor descriptor chasing code
This makes things a bit more generic for use in the next commit.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 12 Mar 2021 19:00:47 +0000 (13:00 -0600)]
anv: Use nir_shader_instructions_pass in apply_pipeline_layout
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Wed, 20 Jan 2021 22:21:38 +0000 (16:21 -0600)]
anv: Use load_global_constant for shader constants
NIR can do a bit better job optimizing this version.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Jason Ekstrand [Fri, 15 Jan 2021 00:00:00 +0000 (18:00 -0600)]
intel/fs,rt: Add a predicate to load_global_const_block
This allows us to do bounds checked A64 block load without the it being
counted as control-flow by NIR. This means that NIR optimizations like
CSE will be able to work on these the same as a regular load.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8635>
Eric Anholt [Wed, 17 Mar 2021 16:19:18 +0000 (09:19 -0700)]
ci/bare-metal: Restart a run on intermittent kernel lockups.
Since enabling SMP on db820c and cranking up how many tests we run, we've
been seeing lockups like this a couple of times a week.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9655>
Rob Clark [Tue, 16 Mar 2021 22:45:38 +0000 (15:45 -0700)]
freedreno/drm: Avoid unitialized timestamp in submit fail
Saw a flood of "waiting on invalid fence" with a completely bogus
looking fence # in a log of a rather strange low-memory crash. Not
sure if it is coming from memory corruption in userspace, but if a
submit ioctl is failing due to failed allocation (or other reason)
we would get left with random stack garbage as the fence #. Let's
not have that as a potential problem.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9638>
Rhys Perry [Mon, 15 Mar 2021 13:35:54 +0000 (13:35 +0000)]
aco: use a single instruction for uadd32_sat() on GFX8
fossil-db (GFX8):
Totals from 8 (0.01% of 147787) affected shaders:
SGPRs: 352 -> 368 (+4.55%)
CodeSize: 49576 -> 48788 (-1.59%)
Instrs: 9487 -> 9318 (-1.78%)
Latency: 49935 -> 49607 (-0.66%)
InvThroughput: 138493 -> 137443 (-0.76%)
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9598>
Rhys Perry [Mon, 15 Mar 2021 13:33:29 +0000 (13:33 +0000)]
aco: use uadd32_sat() helper for nir_op_uadd_sat
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9598>
Rhys Perry [Mon, 15 Mar 2021 13:28:17 +0000 (13:28 +0000)]
aco: implement 64-bit VGPR {u,i}find_msb
This can be created by subgroupBallotFindMSB().
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4458
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9598>
Marek Olšák [Fri, 12 Mar 2021 15:26:54 +0000 (10:26 -0500)]
ac/gpu_info: fix more non-coherent RB and GL2 combinations
It ignored non-harvested chips with a non-power-of-two memory bus.
Fixes:
abed921ce71 - amd: add support for Navy Flounder
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9568>
Erik Faye-Lund [Wed, 17 Mar 2021 09:32:15 +0000 (10:32 +0100)]
zink: fix free of ralloced pointer
When we alloc with ralloc, we also need to free with it.
But let's take a step back here; we don't just need to use ralloc, we
also need to destroy all other screen-resources. So let's call the
destructor here instead.
Fixes:
2643f9ed284 ("zink: ralloc screen objects")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9647>
Erik Faye-Lund [Wed, 17 Mar 2021 10:36:59 +0000 (11:36 +0100)]
zink: fix emulation of no mipfilter
This approach is taken from the Vulkan spec[1], where a robust maxLod
of 0.25 is proposed instead of 0.0.
This has the effect of allowing room for both minification and
magnification filters, yet still rounding down to the right miplevel in
the end.
[1]: https://www.khronos.org/registry/vulkan/specs/1.2-extensions/man/html/VkSamplerCreateInfo.html#_description
Fixes:
8d46e35d16e ("zink: introduce opengl over vulkan")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9649>
projects / platform / upstream / mesa.git / log