Emma Anholt [Fri, 6 Aug 2021 22:39:02 +0000 (15:39 -0700)]
freedreno/a5xx: Reduce packet emits for SSBO state.
This is what I see happening in
dEQP-VK.spirv_assembly.instruction.compute.opatomic_storage_buffer.load on
pixel 2 (also where I found a buffer big enough to show how to encode the
size).
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12258>
Emma Anholt [Fri, 6 Aug 2021 21:05:53 +0000 (14:05 -0700)]
freedreno/a5xx: Use ST4_ constants for SSBO/image state types.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12258>
Eric Anholt [Thu, 11 Mar 2021 19:20:09 +0000 (11:20 -0800)]
freedreno/a6xx: Skip setting up image dims constants.
We just use resinfo anyway. Notably, a6xx was only doing its setup in the
FS case and not CS.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12258>
Eric Anholt [Thu, 11 Mar 2021 19:14:07 +0000 (11:14 -0800)]
freedreno/ir3: Move a6xx's get_ssbo_size shl to NIR.
Just cleaning up a TODO.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12258>
Eric Anholt [Thu, 11 Mar 2021 19:00:24 +0000 (11:00 -0800)]
freedreno/ir3: Use the resinfo path for ssbo sizes on GL, too.
Less state walking at draw time, in exchange for a SHL in the lookup.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12258>
Emma Anholt [Tue, 17 Aug 2021 20:40:57 +0000 (13:40 -0700)]
freedreno/ir3: Only lower cube image sizes once.
shader variants can cause ir3_nir_finalize() to run more than once, which
would make us keep dividing the size by 6.
Fixes:
a48fc88571f9 ("freedreno/a6xx: Apply the cube image size lowering to GL, too.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12258>
Rob Clark [Fri, 23 Jul 2021 16:24:42 +0000 (09:24 -0700)]
freedreno: Add a680 support
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12041>
Jordan Justen [Tue, 17 Aug 2021 09:11:10 +0000 (02:11 -0700)]
meson: Check that bin/meson_get_version.py ran without an error
According to https://mesonbuild.com/Reference-manual.html, the check
parameter is supported since meson 0.47.0.
This could have helped to catch the issue fixed by:
221871fb6da ("meson: Search for python3 before python for bin/meson_get_version.py")
as it would have caused the build to fail immediately.
It's still a good idea to check the result even though that issue is
now fixed.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12404>
Icecream95 [Sun, 1 Aug 2021 11:15:00 +0000 (23:15 +1200)]
panfrost: drm-shim support
Reviewed-and-tested-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12431>
Mike Blumenkrantz [Tue, 17 Aug 2021 21:40:16 +0000 (17:40 -0400)]
zink: ci updates
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12423>
Mike Blumenkrantz [Fri, 12 Feb 2021 15:23:54 +0000 (10:23 -0500)]
zink: enable compat contexts
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12423>
Jason Ekstrand [Thu, 22 Jul 2021 21:24:59 +0000 (16:24 -0500)]
anv,vulkan: Add a vk_image::wsi_legacy_scanout bit
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 20:47:29 +0000 (15:47 -0500)]
anv,vulkan: Move drm_format_mod to vk_image
Even though we can't really do the parsing on behalf of the driver (it's
too complicated), storing it in the vk_image lets us provide a common
implementation of vkGetImageDrmFormatModifierPropertiesEXT(). It'll
also be useful in the next few commits for swapchain images.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 13:15:38 +0000 (08:15 -0500)]
anv,vulkan: Move ANV image layout helpers to common code
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 03:03:12 +0000 (22:03 -0500)]
anv: Make anv_image_view derive from vk_image_view
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 03:02:14 +0000 (22:02 -0500)]
vulkan: Add a vk_image_view struct
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Fri, 30 Jul 2021 17:42:27 +0000 (12:42 -0500)]
vulkan,radv: Move vk_format_depth/stencil_only to common code
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Fri, 30 Jul 2021 17:40:09 +0000 (12:40 -0500)]
radv: Add asserts to vk_format_depth/stencil_only
It doesn't make sense to ask for the depth-only or stencil-only format
if there is no depth or stencil. One bit of radv_image.c did seem to
take advantage of the default case in vk_format_depth_only so throw an
`if (vk_format_has_depth(format))` around it.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Bas Nieuwenhuizen [Tue, 17 Aug 2021 19:56:36 +0000 (21:56 +0200)]
radv: Check format before calling depth_only/stencil_only.
Breaks when we drop the fallback in those functions.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 04:37:44 +0000 (23:37 -0500)]
vulkan: Refactor and better document vk_image_expand_aspect_mask
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 01:48:30 +0000 (20:48 -0500)]
anv,vulkan: Move VkImageSubresource* helpers from ANV
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 04:26:24 +0000 (23:26 -0500)]
anv,vulkan: Move anv_image_expand_aspects to common code
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Thu, 22 Jul 2021 01:42:00 +0000 (20:42 -0500)]
anv: Make anv_image derive from vk_image
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Wed, 21 Jul 2021 23:35:22 +0000 (18:35 -0500)]
vulkan: Add a vk_image struct
Currently, this is optional for drivers to carry around but it scrapes
up most of VkImageCreateInfo for you and parses a couple of extension
structs. We also add a few useful little helpers copied over from ANV.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Wed, 21 Jul 2021 23:11:39 +0000 (18:11 -0500)]
anv: Delete anv_image::format
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Jason Ekstrand [Wed, 21 Jul 2021 23:06:14 +0000 (18:06 -0500)]
anv/image: Use planes[i]->primary_surface.isl.format in check_drm_format_mod
In theory, with linear vs. tiled differences, it could be different
(RGBA vs. RGB etc.) but it won't matter for the two checks we do with
it. Also, we probably want to be checking the real format here anyway.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12023>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:28:32 +0000 (10:28 -0400)]
zink: handle map failures more effectively
the transfer object needs to be destroyed
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12422>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:25:28 +0000 (10:25 -0400)]
zink: break out transfer map destroy
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12422>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:23:48 +0000 (10:23 -0400)]
zink: remove unused variable from image map
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12422>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:22:55 +0000 (10:22 -0400)]
zink: split buffer and image map functions
this is a bit nicer to read
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12422>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:19:00 +0000 (10:19 -0400)]
zink: split out zink_transfer allocation
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12422>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:12:57 +0000 (10:12 -0400)]
zink: remove duplicated zink_resource_object::mem member
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12421>
Mike Blumenkrantz [Fri, 30 Jul 2021 14:09:23 +0000 (10:09 -0400)]
zink: collapse 'dedicated' allocation into zink_bo
this simplifies all the map and memory management code
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12421>
Mike Blumenkrantz [Tue, 17 Aug 2021 19:59:46 +0000 (15:59 -0400)]
zink: flake out some tests for now
I can't repro these on any branch, so I'll just disable them and see if they get fixed
by some upcoming refactors
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12424>
Emma Anholt [Thu, 5 Aug 2021 19:57:13 +0000 (12:57 -0700)]
freedreno/a6xx: Apply the cube image size lowering to GL, too.
Fixes KHR-GLES31.core.texture_cube_map_array.texture_size_compute_sh.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12256>
Emma Anholt [Wed, 28 Jul 2021 23:30:19 +0000 (16:30 -0700)]
ci/freedreno: Move freedreno's deqp testing to suite support.
Using suites makes load-balancing our jobs much easier, keeps the CPU busy
handling the a630_gles_others.sh test sets (and improves the output and
baseline handling for them), and makes it trivial to add in more short
test sets.
a306: still 5 jobs, and we add KHR-GLES2 (KHR-GLES3 is unstable)
a530: still 5 jobs, added KHR-GLES*
a630_gl: 5 jobs becomes 4, and we add KHR-GLES*
a630_vk: still 3 jobs, now 1/3 of all VK instead of 1/4.
a630_vk_full: still 2 jobs, now includes full bypass testing, partial
no-force testing, and testing of pre-merge-skipped tests.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12256>
Emma Anholt [Tue, 17 Aug 2021 19:00:01 +0000 (12:00 -0700)]
ci/turnip: Add a new flake from running more of the CTS.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12256>
Mike Blumenkrantz [Fri, 30 Jul 2021 15:34:40 +0000 (11:34 -0400)]
zink: rejigger PIPE_MAP_ONCE for internal qbo reads
DONTBLOCK is only used internally for this case, and promoting it over
the staging buffer case ensures that it's always reached when it should be
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Mike Blumenkrantz [Fri, 30 Jul 2021 13:22:09 +0000 (09:22 -0400)]
zink: remove PIPE_MAP_ONCE from subdata
subdata calls that happen once for a buffer are likely to happen again,
so just leave them mapped
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Mike Blumenkrantz [Thu, 29 Jul 2021 17:34:43 +0000 (13:34 -0400)]
zink: make map_count useful for dedicated image allocations
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Mike Blumenkrantz [Thu, 29 Jul 2021 17:31:38 +0000 (13:31 -0400)]
zink: split mem unmap logic for images and buffers
buffers should only be unmapped if they're explicitly marked for unmap,
but images should only be unmapped when running in 32bit, in which case
they need to always be unmapped
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Mike Blumenkrantz [Thu, 29 Jul 2021 17:02:22 +0000 (13:02 -0400)]
zink: split transfer_unmap for images and buffers
these should be separate
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Mike Blumenkrantz [Tue, 8 Jun 2021 14:33:35 +0000 (10:33 -0400)]
zink: stop zeroing structs during resource allocation
some of these show up in profiling
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Mike Blumenkrantz [Fri, 11 Jun 2021 13:33:43 +0000 (09:33 -0400)]
zink: repack zink_resource_object struct
this makes better use of cachelines and eliminates holes
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12410>
Danylo Piliaiev [Tue, 17 Aug 2021 16:46:38 +0000 (19:46 +0300)]
turnip: provide dummy CmdSetLogicOpEXT and CmdSetPatchControlPointsEXT
Fixes:
fd62e0b7996afde22447fd929a55b0cc89e9ab36 "tu: enable VK_EXT_extended_dynamic_state2"
Fixes dEQP-VK.api.version_check.entry_points
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5243
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12417>
Connor Abbott [Mon, 16 Aug 2021 10:04:47 +0000 (12:04 +0200)]
tu: Consider depth/stencil for implicit dependencies
Noticed by inspection.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12386>
Connor Abbott [Mon, 16 Aug 2021 09:59:03 +0000 (11:59 +0200)]
tu: Fix logic errors with subpass implicit dependencies
We were adding them if there was an external dep instead of if there
wasn't, and we were skipping the entire subpass which would've resulted
in attachments not getting marked as used.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12386>
Matt Turner [Thu, 5 Aug 2021 01:47:02 +0000 (18:47 -0700)]
util: Replace recursive DFS with iterative implementation
Doesn't fix, but improves the situation in issue #5163. The
VK.spirv_assembly.instruction.graphics.spirv_ids_abuse.lots_ids_* tests
emit about 160k instructions. ir3_sched blows out the stack after
dag_traverse_bottom_up_node reaches a depth of about 130k frames.
This patch replaces the recursively-implemented post-order traversal
with an iterative implementation using a stack, allowing us to process
DAGs as large as memory can hold.
Definitely makes you appreciate the elegance of recursion...
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12232>
Matt Turner [Tue, 10 Aug 2021 19:10:22 +0000 (12:10 -0700)]
util: Add unit tests for dag
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12232>
Connor Abbott [Tue, 3 Aug 2021 13:34:56 +0000 (15:34 +0200)]
freedreno: Decode a650+ CP_START_BIN/CP_END_BIN packets
The blob uses them for GMEM renderpasses.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12184>
Erik Faye-Lund [Thu, 12 Aug 2021 13:59:55 +0000 (15:59 +0200)]
st/mesa: correct point_tri_clip for gles2
The OpenGL ES 2.0 (and later) specifications, section 2.13 (Primitive
Clipping) say the following about point-clipping:
> If the primitive under consideration is a point, then clipping
> discards it if it lies outside the near or far clip plane; otherwise
> it is passed unchanged.
This matches the D3D convention, and we already have a rasterizer-state
bit for it. So let's set that bit when the API is OpenGL ES 2.
We can set this inside the PointSprite-conditional block, because that's
always enabled on GLES 2, and it's undefined to enable this state
without also enabling point_quad_rasterization.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12350>
Erik Faye-Lund [Thu, 12 Aug 2021 18:50:58 +0000 (20:50 +0200)]
llvmpipe: take intersection with bbox for non-legacy points
When I updated this code for multi-sampling, I missed one detail; if we
want to be able to support pipe_rasterizer_state::point_tri_clip, we
need to use the intersection of the bbox (clipped to the viewport
rectangle) and the generated primitive, otherwise we won't end up doing
x/y viewport clipping at all.
Because we've adjusted some of the parts of the bbox when adjusting for
inclusiveness/exclusiveness and fill-rule, we also need to reverse the
adjustment.
Fixes:
f530e72ea05 ("llvmpipe: do not always use pixel-rounded coordinates for points")
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12350>
Jason Ekstrand [Wed, 7 Jul 2021 16:28:05 +0000 (11:28 -0500)]
intel/isl: Use uint64_t for computed byte offsets
This is mostly a bit of future-proofing. We never end up with offsets
that don't fit in 32 bits today because, thanks to driver limitations
caused by relocations, we don't allocate buffers bigger than 2GB today.
However, if we ever did, it's possible to create a surface on modern
platforms that consumes more than 4GB and we would end up with wrapping
in our offset calculations.
Acked-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11765>
Jason Ekstrand [Wed, 7 Jul 2021 16:33:49 +0000 (11:33 -0500)]
intel/isl: Add a missing assert in isl_tiling_get_intratile_offset_sa
Fixes:
a4dafe1faddb "intel/isl: Make the offset helpers four dimensional"
Acked-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11765>
Jason Ekstrand [Wed, 7 Jul 2021 16:06:59 +0000 (11:06 -0500)]
intel/isl: Better document isl_tiling_get_intratile_offset_*
The docs weren't updated when we switched it to 4D. Also, the new docs
are way better. While we're here, use the parameter name offset_B to be
more consistent.
Acked-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11765>
Jason Ekstrand [Mon, 5 Mar 2018 23:06:58 +0000 (15:06 -0800)]
intel/isl: Add units to view dimensions in isl_surf_get_uncompressed_surf
This makes things a bit more clear.
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Acked-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11765>
Jason Ekstrand [Wed, 7 Jul 2021 15:52:09 +0000 (10:52 -0500)]
intel/isl: Explicitly set offset_B = 0 in get_uncomp_surf for arrays
The only user of this case is iris which initializes offset_B to 0 so
there's no real bug here. However, it is unexpected from an API PoV.
Fixes:
9946120d2b4e "intel/isl: Add more cases to isl_surf_get_uncompressed_surf"
Acked-by: Ivan Briano <ivan.briano@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11765>
Mike Blumenkrantz [Thu, 13 May 2021 11:33:44 +0000 (07:33 -0400)]
zink: add a suballocator
this is an aux/pipebuffer implementation borrowing heavily from the
one in radeonsi. it currently has the following limitations, which
will be resolved in a followup series:
* 32bit address space still explodes
* swapchain images still have separate memory handling
performance in games like Tomb Raider has been observed to increase by
over 1000%
SQUASHED: simplify get_memory_type_index()
now that the heaps are enumerated, this can be reduced to a simple
array index with a fallback
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12146>
Andreas Baierl [Tue, 17 Aug 2021 10:47:55 +0000 (12:47 +0200)]
lima: CI: Enable GL_R8 and GL_RG8 texture formats
This is fixed in deqp now. See https://github.com/KhronosGroup/VK-GL-CTS/pull/241
Since CI is using deqp version > vulkan-cts-1.2.6.0, this isn't an issue anymore.
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12409>
Erico Nunes [Wed, 7 Jul 2021 20:54:43 +0000 (22:54 +0200)]
ci: enable CI for lima again
Enable CI for lima again on meson-gxl-s805x-libretech-ac boards
with Mali-450.
These boards are managed by a LAVA instance and so follow the LAVA CI
workflow in Mesa.
The goal is to have coverage for deqp-gles2, as lima is a GLES2-only
driver.
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11789>
Filip Gawin [Mon, 16 Aug 2021 11:02:50 +0000 (13:02 +0200)]
radv: improve rounding of zmin
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12388>
Roman Stratiienko [Fri, 13 Aug 2021 17:04:51 +0000 (20:04 +0300)]
lima: Implement lima_resource_get_param() callback
Currently stride, offset, modifier is obtained by invoking
lima_resource_get_handle() with WINSYS_HANDLE_TYPE_KMS.
Before commit
47f000c170cc this path was working. Obtained handle
was simply ignored by DRI frontend and only requested data used.
After commit
47f000c170cc such requests started to fail when
DRI is initialized using KMSRO and resource has no scanout data.
When lima_resource_get_param() is implemented, it will be used in
a first place to obtain resource data.
Fixes:
47f000c170cc ("lima: fail in get_handle(TYPE_KMS) without a scanout resource")
Signed-off-by: Roman Stratiienko <r.stratiienko@gmail.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12362>
Jordan Justen [Tue, 17 Aug 2021 08:01:40 +0000 (01:01 -0700)]
meson: Search for python3 before python for bin/meson_get_version.py
Most systems have either dropped the python executable, or made it
python3.
But it is still possible to configure a system such that python runs
python2. https://www.python.org/dev/peps/pep-0394/
Or, some developers may still be running older distributions where
python is python2.
Since bin/meson_get_version.py now requires python3, we should search
for python3 before python.
Fixes:
f1eae2f8bbe ("python: drop python2 support")
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12400>
Juan A. Suarez Romero [Tue, 17 Aug 2021 07:50:10 +0000 (09:50 +0200)]
v3d/ci: add piglit flake
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12401>
Marcin Ślusarz [Fri, 13 Aug 2021 14:39:09 +0000 (16:39 +0200)]
glsl/opt_algebraic: disable invalid optimization
When operators other than eq and ne are involved we can't really
move operands around and negate them because such transformation
may change the value of the whole expression.
Some examples:
For unsigned var:
0 >= 1u + var would eventually become 0xffffffff >= var,
which would always evaluate to true, when original expression
was true only for var == 0xffffffff.
For signed var:
0 >= 1 + var would become -1 >= var, which would evaluate to
false for var ==
2147483647, when original expression evaluated
to true (because signed overflow is defined to wrap around in
glsl, 1 +
2147483647 == -
2147483648, so 0 >= -
2147483648).
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5226
Fixes:
34ec1a24d61 ("glsl: Optimize (x + y cmp 0) into (x cmp -y).")
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12359>
Marcin Ślusarz [Mon, 19 Jul 2021 15:53:14 +0000 (17:53 +0200)]
intel/error-decode: printout INSTDONE_GEOM register for Gfx12.5
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12392>
Marcin Ślusarz [Mon, 19 Jul 2021 15:52:24 +0000 (17:52 +0200)]
genxml: add INSTDONE_GEOM register for Gfx12.5
Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12392>
Lionel Landwerlin [Thu, 20 May 2021 06:54:18 +0000 (06:54 +0000)]
intel/error-decode: printout more registers
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/12392>
Lionel Landwerlin [Mon, 17 May 2021 07:20:30 +0000 (10:20 +0300)]
genxml: add more INSTDONE registers for Gfx12.5
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/12392>
Ella-0 [Sun, 15 Aug 2021 14:50:39 +0000 (14:50 +0000)]
v3dv: Implement VK_EXT_provoking_vertex
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12382>
Samuel Pitoiset [Fri, 13 Aug 2021 10:04:42 +0000 (12:04 +0200)]
radv: fix pre-computing viewport xform when setting new viewports
viewportCount is the number of viewports in pViewports while
firstViewport is the index.
Fixes new CTS dEQP-VK.draw.depth_clamp.*_clamp_four_viewports
Fixes:
a2ef92d7a58 ("radv: pre-calculate viewport transforms")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12353>
Timothy Arceri [Fri, 6 Aug 2021 13:39:06 +0000 (23:39 +1000)]
nir: fix GCM when GVN enabled
Enabling GVN uncovered a bug where we would crash if the pass
thinking about pushing something into a loop.
Fixes:
6538b3e56614 ("nir: add heuristic for instructions in loops with GCM")
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12242>
Vinson Lee [Tue, 10 Aug 2021 00:42:47 +0000 (17:42 -0700)]
nv50/ir: Add FlatteningPass constructor.
Fix defect reported by Coverity Scan.
Uninitialized scalar field (UNINIT_CTOR)
member_not_init_in_gen_ctor: The compiler-generated constructor for this
class does not initialize gpr_unit.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12295>
Vinson Lee [Tue, 10 Aug 2021 03:40:31 +0000 (20:40 -0700)]
zink: Remove unnecessary null checks.
Fix defects reported by Coverity Scan.
Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking sa suggests that it may be null,
but it has already been dereferenced on all paths leading to the
check.
check_after_deref: Null-checking sb suggests that it may be null,
but it has already been dereferenced on all paths leading to the
check.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12298>
Filip Gawin [Mon, 16 Aug 2021 11:13:43 +0000 (13:13 +0200)]
radeonsi: improve rounding of zmin
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12389>
Dave Airlie [Mon, 9 Aug 2021 04:06:12 +0000 (14:06 +1000)]
llvmpipe: init renderer string once to avoid races.
In a multithreads clover run the get_name call would race against
itself and sometimes an empty device name would occur.
Just init it once.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12285>
Dave Airlie [Thu, 26 Nov 2020 07:24:15 +0000 (17:24 +1000)]
gallivm: fix non-32 bit popcounts.
Fixes
OpenCL CTS integer_ops popcount
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12285>
Dave Airlie [Wed, 25 Nov 2020 20:16:39 +0000 (06:16 +1000)]
gallivm: fix idiv/irem for 8/16/64-bit and 32-bit INT_MIN/-1
This fixes integer division for non-32bit but also fixes the
32-bit case where INT_MIN/-1 causes an exception.
Fixes CL CTS
./integer_ops/test_integer_ops quick_long_math
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12285>
Dave Airlie [Thu, 5 Nov 2020 03:24:27 +0000 (13:24 +1000)]
llvmpipe/cl: limit kernel input size.
Fixes:
api min_max_parameter_size
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12285>
Dave Airlie [Mon, 9 Nov 2020 06:16:01 +0000 (16:16 +1000)]
gallivm: don't lower local invocation index in frontend
The frontend can't handle variable block sizes properly,
so just lower it here in the backend.
Fixes CTS basic local_linear_id + get_linear_ids
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12285>
Emma Anholt [Mon, 16 Aug 2021 05:03:59 +0000 (22:03 -0700)]
i915g: Implement cube/3d texture_subdata() as a series of per-layer maps.
i915 doesn't lay out the images such that one could use a layer_stride to
step between them, and the individual maps should be just as good at
uploading.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12384>
Emma Anholt [Mon, 16 Aug 2021 17:24:47 +0000 (10:24 -0700)]
i915g: Fix 3D texture layouts for width != height.
Obvious typo here.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12384>
Ella-0 [Sat, 14 Aug 2021 14:09:23 +0000 (14:09 +0000)]
v3dv: Implement VK_EXT_pipeline_creation_cache_control
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12381>
Dave Airlie [Sat, 14 Aug 2021 20:22:34 +0000 (06:22 +1000)]
llvmpipe: enable GL compatibility profiles
The two rasterpos fails looks related to GLSL linking, the vertex
shader is linked with the geometry shader which doesn't use any
of it's outputs so they seem to get removed, which stops the rasterpos
from working.
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12374>
Dave Airlie [Sun, 15 Aug 2021 04:41:06 +0000 (14:41 +1000)]
draw: add vertex color clamping to gs/tes
This refactors out the vertex color clamping from the VS shader,
and adds calls to it for the tes/gs stages. It also conditionalised
they key on having later stages as clamping should only happen in
the last stage.
This is needed for GL compatibility profiles
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12374>
Dave Airlie [Sat, 14 Aug 2021 20:21:51 +0000 (06:21 +1000)]
draw/tess: add clipvertex support for compatibility
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12374>
Dave Airlie [Sat, 14 Aug 2021 20:19:52 +0000 (06:19 +1000)]
draw/gs: add clipvertex support for compatibility
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12374>
Dave Airlie [Fri, 13 Aug 2021 05:00:13 +0000 (15:00 +1000)]
draw: handle primitive ID for quads/quad strips.
In order to enable compat contexts QUADS/QUAD_STRIPS need
to support primitive ID. There are some piglit tests for this.
This adds support to the decomposer to pass quads so the prim
assembler can pick them up and add primitive IDs.
Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12374>
Juan A. Suarez Romero [Mon, 16 Aug 2021 10:48:37 +0000 (12:48 +0200)]
v3dv: initialize CL submission structure
This fixes an issue related with testing this with a kernel with the
performance counters enabled: it introduces a "pad" field that in the CL
submission structure that is not initialized.
Fixes:
ca13868098e ("drm-uapi: add v3d performance counters")
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12390>
Rhys Perry [Mon, 16 Aug 2021 12:24:52 +0000 (13:24 +0100)]
ci: update trace hashes
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Wed, 13 Jan 2021 15:11:57 +0000 (15:11 +0000)]
nir,glsl_to_nir: use nir_fdot()
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Tue, 16 Jun 2020 13:28:18 +0000 (14:28 +0100)]
spirv: create ffma more often
We will not be able to combine instructions into ffma later if they are
exact, so create them from the start. They can be lowered later if they
are unwanted.
fossil-db (GFX10.3):
Totals from 14697 (10.05% of 146267) affected shaders:
VGPRs: 645736 -> 614168 (-4.89%)
CodeSize:
59312768 ->
58735352 (-0.97%); split: -0.97%, +0.00%
MaxWaves: 372900 -> 376666 (+1.01%)
Instrs:
11339280 ->
11120882 (-1.93%); split: -1.93%, +0.00%
Latency:
284874519 ->
285277327 (+0.14%); split: -0.10%, +0.24%
InvThroughput:
68791374 ->
68526739 (-0.38%); split: -0.49%, +0.10%
fossil-db (GFX10):
Totals from 11039 (7.55% of 146267) affected shaders:
CodeSize:
54785444 ->
54785268 (-0.00%); split: -0.00%, +0.00%
Instrs:
10401349 ->
10401396 (+0.00%); split: -0.00%, +0.00%
Latency:
277781803 ->
278572890 (+0.28%); split: -0.00%, +0.29%
InvThroughput:
65035902 ->
65100855 (+0.10%); split: -0.00%, +0.10%
fossil-db (GFX9):
Totals from 24055 (16.43% of 146401) affected shaders:
SGPRs: 1790704 -> 1790640 (-0.00%)
VGPRs: 1105736 -> 1105716 (-0.00%)
CodeSize:
110944732 ->
110948812 (+0.00%); split: -0.00%, +0.01%
Instrs:
21609095 ->
21610227 (+0.01%); split: -0.00%, +0.01%
Latency:
756137596 ->
756145812 (+0.00%); split: -0.02%, +0.02%
InvThroughput:
344103825 ->
344112245 (+0.00%); split: -0.00%, +0.01%
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Fri, 5 Mar 2021 17:28:36 +0000 (17:28 +0000)]
nir: lower fdot to ffma if lower_ffma=false
fossil-db (GFX10.3):
Totals from 57689 (39.44% of 146267) affected shaders:
VGPRs: 2873712 -> 2873432 (-0.01%); split: -0.01%, +0.00%
CodeSize:
227661100 ->
227583572 (-0.03%); split: -0.08%, +0.04%
MaxWaves: 1289562 -> 1289598 (+0.00%); split: +0.01%, -0.00%
Instrs:
43115433 ->
43083308 (-0.07%); split: -0.12%, +0.05%
Latency:
869947191 ->
870279826 (+0.04%); split: -0.06%, +0.10%
InvThroughput:
199425811 ->
199434448 (+0.00%); split: -0.04%, +0.05%
fossil-db (GFX10):
Totals from 2 (0.00% of 146267) affected shaders:
Latency: 8123 -> 8107 (-0.20%)
fossil-db (GFX9):
Totals from 2 (0.00% of 146401) affected shaders:
(no stat changes)
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Thu, 23 Jul 2020 10:40:11 +0000 (11:40 +0100)]
nir: create ffma from builders more often
We will not be able to combine instructions into ffma later if they are
exact, so create them from the start. They can be lowered later if they
are unwanted.
fossil-db (GFX10.3):
Totals from 16589 (11.34% of 146267) affected shaders:
VGPRs: 938872 -> 938704 (-0.02%)
SpillSGPRs: 11334 -> 10785 (-4.84%)
CodeSize:
96551964 ->
96498040 (-0.06%); split: -0.08%, +0.02%
MaxWaves: 338760 -> 338772 (+0.00%)
Instrs:
18356857 ->
18350486 (-0.03%); split: -0.06%, +0.02%
Latency:
561563310 ->
561414360 (-0.03%); split: -0.08%, +0.05%
InvThroughput:
145629673 ->
145594740 (-0.02%); split: -0.04%, +0.01%
fossil-db (GFX10):
Totals from 16252 (11.11% of 146267) affected shaders:
VGPRs: 893820 -> 893744 (-0.01%)
SpillSGPRs: 11334 -> 10785 (-4.84%)
CodeSize:
95890244 ->
95839124 (-0.05%); split: -0.08%, +0.02%
MaxWaves: 367704 -> 367734 (+0.01%)
Instrs:
18199741 ->
18194437 (-0.03%); split: -0.06%, +0.03%
Latency:
560912971 ->
560854179 (-0.01%); split: -0.07%, +0.06%
InvThroughput:
142899814 ->
142877939 (-0.02%); split: -0.03%, +0.02%
fossil-db (GFX9):
Totals from 16287 (11.12% of 146401) affected shaders:
SGPRs: 1312784 -> 1312768 (-0.00%); split: -0.05%, +0.05%
VGPRs: 931440 -> 931444 (+0.00%); split: -0.00%, +0.00%
SpillSGPRs: 14623 -> 14597 (-0.18%)
CodeSize:
94428788 ->
94344404 (-0.09%); split: -0.10%, +0.01%
MaxWaves: 90105 -> 90109 (+0.00%)
Instrs:
18486905 ->
18473434 (-0.07%); split: -0.08%, +0.01%
Latency:
720947295 ->
720818323 (-0.02%); split: -0.07%, +0.05%
InvThroughput:
365240104 ->
365224659 (-0.00%); split: -0.02%, +0.01%
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Wed, 2 Jun 2021 14:14:41 +0000 (15:14 +0100)]
nir: add ffma creation helpers
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Thu, 13 May 2021 11:50:09 +0000 (12:50 +0100)]
nir/algebraic: add is_used_once to dot product reassociation optimization
This improves register usage.
fossil-db (Sienna Cichlid, on top of !9805):
Totals from 4317 (2.88% of 149839) affected shaders:
VGPRs: 352592 -> 351704 (-0.25%); split: -1.48%, +1.23%
SpillSGPRs: 182 -> 248 (+36.26%)
CodeSize:
31601192 ->
31587624 (-0.04%); split: -0.09%, +0.04%
MaxWaves: 56964 -> 57298 (+0.59%); split: +2.48%, -1.90%
Instrs: 5973557 -> 5974122 (+0.01%); split: -0.05%, +0.06%
Latency:
72088175 ->
72253033 (+0.23%); split: -0.36%, +0.59%
InvThroughput:
14978160 ->
14798919 (-1.20%); split: -1.29%, +0.09%
VClause: 100994 -> 98645 (-2.33%); split: -3.05%, +0.73%
SClause: 278206 -> 276820 (-0.50%); split: -0.54%, +0.04%
Copies: 200264 -> 199556 (-0.35%); split: -1.17%, +0.82%
Branches: 86410 -> 85930 (-0.56%); split: -0.56%, +0.01%
PreSGPRs: 207355 -> 207759 (+0.19%); split: -0.00%, +0.20%
PreVGPRs: 314646 -> 310911 (-1.19%); split: -1.35%, +0.17%
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Wed, 17 Mar 2021 19:54:38 +0000 (19:54 +0000)]
nir/algebraic: reassociate add chains for more MAD/FMA-friendly code
fossil-db (GFX10.3):
Totals from 25866 (17.68% of 146267) affected shaders:
VGPRs: 1625456 -> 1644936 (+1.20%); split: -0.05%, +1.24%
SpillSGPRs: 11729 -> 11725 (-0.03%); split: -0.07%, +0.03%
CodeSize:
161604460 ->
161458052 (-0.09%); split: -0.11%, +0.02%
MaxWaves: 454842 -> 452160 (-0.59%); split: +0.04%, -0.63%
Instrs:
30652596 ->
30456446 (-0.64%); split: -0.65%, +0.01%
Latency:
723098749 ->
722084247 (-0.14%); split: -0.21%, +0.07%
InvThroughput:
166023468 ->
165506875 (-0.31%); split: -0.36%, +0.05%
fossil-db (GFX10):
Totals from 25866 (17.68% of 146267) affected shaders:
VGPRs: 1593576 -> 1611976 (+1.15%); split: -0.09%, +1.25%
SpillSGPRs: 11729 -> 11725 (-0.03%); split: -0.07%, +0.03%
CodeSize:
162294468 ->
162154456 (-0.09%); split: -0.11%, +0.02%
MaxWaves: 477448 -> 474166 (-0.69%); split: +0.10%, -0.79%
Instrs:
30820164 ->
30625805 (-0.63%); split: -0.65%, +0.02%
Latency:
723190249 ->
722273445 (-0.13%); split: -0.20%, +0.08%
InvThroughput:
163114872 ->
162582966 (-0.33%); split: -0.37%, +0.04%
fossil-db (GFX9):
Totals from 25866 (17.67% of 146401) affected shaders:
SGPRs: 2167808 -> 2169920 (+0.10%); split: -0.09%, +0.19%
VGPRs: 1649404 -> 1667592 (+1.10%); split: -0.43%, +1.53%
CodeSize:
161273556 ->
161281996 (+0.01%); split: -0.07%, +0.08%
MaxWaves: 114910 -> 113519 (-1.21%); split: +0.10%, -1.31%
Instrs:
31557180 ->
31403708 (-0.49%); split: -0.50%, +0.02%
Latency:
899594793 ->
898786283 (-0.09%); split: -0.19%, +0.10%
InvThroughput:
412265691 ->
411551698 (-0.17%); split: -0.28%, +0.11%
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Fri, 5 Mar 2021 18:41:13 +0000 (18:41 +0000)]
nir/algebraic: add various ffma optimizations
fossil-db (GFX10.3):
Totals from 7532 (5.15% of 146267) affected shaders:
VGPRs: 414696 -> 414304 (-0.09%); split: -0.18%, +0.08%
CodeSize:
33393444 ->
33375908 (-0.05%); split: -0.13%, +0.08%
MaxWaves: 149854 -> 150094 (+0.16%); split: +0.27%, -0.11%
Instrs: 6279823 -> 6271364 (-0.13%); split: -0.18%, +0.05%
Latency:
60308898 ->
60296025 (-0.02%); split: -0.13%, +0.11%
InvThroughput:
13770542 ->
13745192 (-0.18%); split: -0.24%, +0.06%
fossil-db (GFX10):
Totals from 7532 (5.15% of 146267) affected shaders:
VGPRs: 406664 -> 405564 (-0.27%); split: -0.39%, +0.12%
CodeSize:
33544656 ->
33527568 (-0.05%); split: -0.13%, +0.08%
MaxWaves: 158584 -> 158858 (+0.17%); split: +0.30%, -0.13%
Instrs: 6316242 -> 6307913 (-0.13%); split: -0.18%, +0.05%
Latency:
60243290 ->
60232844 (-0.02%); split: -0.13%, +0.11%
InvThroughput:
13643345 ->
13620171 (-0.17%); split: -0.24%, +0.07%
fossil-db (GFX9):
Totals from 7543 (5.15% of 146401) affected shaders:
SGPRs: 546384 -> 547472 (+0.20%); split: -0.08%, +0.28%
VGPRs: 412636 -> 411896 (-0.18%); split: -0.27%, +0.09%
CodeSize:
33216196 ->
33210564 (-0.02%); split: -0.12%, +0.11%
MaxWaves: 38771 -> 38789 (+0.05%); split: +0.17%, -0.12%
Instrs: 6419878 -> 6414891 (-0.08%); split: -0.18%, +0.11%
Latency:
70972327 ->
70922754 (-0.07%); split: -0.15%, +0.08%
InvThroughput:
33949039 ->
33909258 (-0.12%); split: -0.20%, +0.08%
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Wed, 17 Jun 2020 11:07:13 +0000 (12:07 +0100)]
spirv: swap fadd operands in build_asin() and matrix_multiply()
This shouldn't do anything but will make testing a later patch easier.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>
Rhys Perry [Wed, 17 Jun 2020 10:53:13 +0000 (11:53 +0100)]
nir: swap fadd operands in nir_atan()
This shouldn't do anything but will make testing a later patch easier.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8056>