Mike Blumenkrantz [Fri, 8 Apr 2022 16:07:26 +0000 (12:07 -0400)]
zink: handle multisampled fbfetch
this uses a shader key to flag the fbfetch as per-sample
it doesn't work, but at least it doesn't error anymore?
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15823>
Mike Blumenkrantz [Fri, 8 Apr 2022 15:06:31 +0000 (11:06 -0400)]
zink: handle SUBPASS_MS in ntv
for completeness but also maybe fbfetch
Acked-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15823>
Jordan Justen [Fri, 8 Apr 2022 06:35:06 +0000 (23:35 -0700)]
intel/dev: Add ATS-M pci-ids
Ref: Bspec 44477
Ref: https://patchwork.freedesktop.org/series/101907/
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Acked-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15814>
Emma Anholt [Tue, 5 Apr 2022 17:05:37 +0000 (10:05 -0700)]
nir_to_tgsi: Fix emitting the sample number for non-array MSAA image access.
It's always in .w, rather than being the next component after the
x/y/array index.
Fixes:
c6d3fd8c21e0 ("gallium/ntt: Emit sample index when necessary for image load/store.")
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15825>
Emma Anholt [Mon, 28 Mar 2022 23:48:31 +0000 (16:48 -0700)]
virgl: Disable nir_op_ffloor to avoid sending DFLR to virglrenderer.
This means that we send ffract+fsub in place of a normal FLR, but
hopefully virglrenderer can be fixed (or doubles support removed).
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15825>
Samuel Pitoiset [Tue, 5 Apr 2022 11:09:35 +0000 (13:09 +0200)]
radv: introduce new radv_pipeline_stage structure
This is used to store everything for a pipeline stage like the module,
the NIR, the shader arguments etc. This is inspired from ANV.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15794>
Samuel Pitoiset [Tue, 5 Apr 2022 12:04:54 +0000 (14:04 +0200)]
radv: add missing multi inclusion define to radv_shader_args.h
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15794>
Samuel Pitoiset [Fri, 8 Apr 2022 06:38:05 +0000 (08:38 +0200)]
radv: rework pipeline and shaders creation feedback
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15794>
Samuel Pitoiset [Thu, 7 Apr 2022 14:01:10 +0000 (16:01 +0200)]
radv: re-order shader stages directly in radv_create_shaders()
We will have to access pStages::pNext for modules and this will also
allow to rework feedback creation.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15794>
Rajnesh Kanwal [Mon, 4 Apr 2022 13:01:56 +0000 (14:01 +0100)]
pvr: Remove logic to set vk_device::alloc.
The field is being already set by vk_device_init().
Signed-off-by: Rajnesh Kanwal <rajnesh.kanwal@imgtec.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15768>
Iago Toral Quiroga [Fri, 8 Apr 2022 11:48:54 +0000 (13:48 +0200)]
v3dv: fix bogus VkDrmFormatModifierProperties2EXT usage
The array is allocated for VkDrmFormatModifierPropertiesEXT, so
writring entried with type VkDrmFormatModifierProperties2EXT is
bogus.
It seems this was a mistake added with a change intended to get
rid of VK_OUTARRAY_MAKE, that changed the type of the write by
mistake.
Fixes:
56a2ccf05864 ('v3dv: Stop using VK_OUTARRAY_MAKE()')
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15819>
Tomeu Vizoso [Tue, 5 Apr 2022 10:20:11 +0000 (12:20 +0200)]
ci: Disable Link Power Management with RTL8153
There are reliability problems with the RTL8153 ethernet driver under
certain network loads, related to incompatibility of the device with
Link Power Management.
Add usbcore.quirks=0bda:8153:k to the kernel command line to enable the
USB_QUIRK_NO_LPM option.
Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15791>
Tomeu Vizoso [Tue, 5 Apr 2022 10:16:27 +0000 (12:16 +0200)]
Revert "ci/freedreno: Disable a618 jobs"
This reverts commit
96e17287b4abe6ac1e0245473ed9ccfe40d0d54a.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15791>
Juan A. Suarez Romero [Thu, 7 Apr 2022 16:05:36 +0000 (18:05 +0200)]
v3d: do not leak BO on query begin
This happens when we have a sequence of multiple beginQuery / endQuery
with the same target and query identifier.
As a BO is created on beginQuery but not free on endQuery (we need to
wait until either explicitly deleting the query or querying the
results), in the above sequence we are basically leaking the BO.
This ensures that any BO created before is unreference before creating
the new one.
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/15796>
Erik Faye-Lund [Mon, 21 Mar 2022 10:49:27 +0000 (11:49 +0100)]
wgl: do not disable error-dialogs by default
We don't know if an application developer wants error-dialog by default
or not, even when using a Mesa OpenGL driver.
So let's not assume this is a good thing, and instead make an option for
this behavior that we can enable for CI testing.
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15485>
Erik Faye-Lund [Mon, 21 Mar 2022 10:43:54 +0000 (11:43 +0100)]
util: limit error-dialogs to win32
These dialogs only exist on Windows, so let's not even expose a util
function for this on other platforms.
The code is only ever called from Windows specific code anyway.
While we're at it, clean up the name a bit as well.
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15485>
Erik Faye-Lund [Mon, 21 Mar 2022 10:09:51 +0000 (11:09 +0100)]
wgl: rename force-msaa env-var
This env-var is not specific to the SVGA driver, so we should rename it
for consistency.
While we're at it, let's document the feature.
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15485>
Timur Kristóf [Wed, 6 Apr 2022 12:15:50 +0000 (14:15 +0200)]
radv: Disable NGG for GS with suboptimal output vertex count.
When GS has a lot of output vertices: In this case,
the occupancy of NGG GS is very low because API GS invocations
can't even occupy a single Wave32 wave.
Therefore the legacy pipeline performs better here.
Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6260
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15767>
Boris Brezillon [Mon, 4 Apr 2022 11:55:00 +0000 (13:55 +0200)]
dzn: Support independent depth/stencil access
Needed for VK_KHR_maintenance2. While at it, fix various places where
we were only issuing resource state transition on the first sub-resource
instead of doing it per-layer/aspect.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Mon, 4 Apr 2022 11:50:02 +0000 (13:50 +0200)]
dzn: Fix 2D <-> 3D blits
layer_count == 1 doesn't imply image_is_3d. So let's add explicit
is_3d checks where appropriate.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Mon, 4 Apr 2022 11:45:06 +0000 (13:45 +0200)]
dzn: Lower partial copy of multisample resources to blits
Unfortunately that won't work on transfer queues, but we don't have a
better option right now.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Fri, 1 Apr 2022 14:16:56 +0000 (07:16 -0700)]
vulkan/util: Make STACK_ARRAY() C++-friendly
C++ doesn't like the {0} initializer pattern when the first
field is not an integer, let's use the {} when __cplusplus is
defined.
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Mon, 4 Apr 2022 11:41:30 +0000 (13:41 +0200)]
dzn: Drop extra blank line in dzn_CmdCopyImage2()
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Mon, 4 Apr 2022 11:21:46 +0000 (13:21 +0200)]
dzn: Check image view usage instead of image usage when creating an image view
So we take VkImageViewUsageCreateInfo extension instead of ignoring it.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 16:04:24 +0000 (18:04 +0200)]
dzn: Force sampleCounts to 1 for bgra4 images
Those are not expected to be used as render-target, and Vulkan mandates
that such formats get their sampleCounts set to VK_SAMPLE_COUNT_1_BIT.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Mon, 4 Apr 2022 11:25:48 +0000 (13:25 +0200)]
dzn: Get rid of dzn_GetPhysicalDeviceProperties()
Rely on the vk_common_ wrapper.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:59:22 +0000 (17:59 +0200)]
dzn: Get rid of dzn_GetPhysicalDeviceFeatures()
And rely on the vk_common_ wrapper to get it implemented.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:57:29 +0000 (17:57 +0200)]
dzn: 3D array images don't exist
Let's force maxArrayLayers to one in that case.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Tue, 5 Apr 2022 10:01:05 +0000 (03:01 -0700)]
dzn: Set bufferFeatures to zero on depth/stencil formats
Those are not supposed to advertise buffer features.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:56:37 +0000 (17:56 +0200)]
dzn: Make sure the properties are all zero when the format is not supported
Move one of the is_supported() check before we start filling the
structure so we don't end up with a partially filled object when
we return VK_ERROR_FORMAT_NOT_SUPPORTED (which deqp doesn't seem to like,
so it's probably coming from a spec requirement).
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:52:45 +0000 (17:52 +0200)]
dzn: Return a valid imageFormatProperties.maxMipLevels
maxMipLevels is encoding the maximum number of MIP levels, but
dzn_physical_device_get_max_mip_levels() return the maximum MIP level.
Let's rename the function and add one to the returned value to fix
the problem.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:49:49 +0000 (17:49 +0200)]
dzn: Fix 3D <-> 2D image copies
We just need to treat layers as slices when manipulating 3D resources
whose slices are coming from/going to 2D array layers.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:47:21 +0000 (17:47 +0200)]
dzn: Support 2Darray views on 3D images for color attachments
Those are declared as 3D RTVs in D3D12, and layers are treated as
slices.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:44:19 +0000 (17:44 +0200)]
dzn: Fix dzn_image_get_rtv_desc() for 3D views
VK_REMAINING_ARRAY_LAYERS maps to -1 in the D3D12 world. Let's make sure
we set WSize to -1 in that case, because the layer_count calculated by
dzn_get_layer_count() won't work for 3D images which never have more
than one layer (in case of 3D images, we treat slices as layers).
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:40:55 +0000 (17:40 +0200)]
dzn: Replace C++ references by pointers
Let's keep as much as we can in plain C.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:39:14 +0000 (17:39 +0200)]
dzn: Align the default case in dzn_image_view_prepare_dsv_desc()
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 15:36:52 +0000 (17:36 +0200)]
dzn: Don't crash when EndCommandBuffer() returns an error
Leftover from a debug session.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Thu, 31 Mar 2022 14:19:56 +0000 (07:19 -0700)]
dzn: Fix dzn_translate_viewport() when height < 0
Since negative height is not a thing in D3D12, we need to adjust the
TopLeftY accordingly.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Boris Brezillon [Fri, 1 Apr 2022 08:04:12 +0000 (01:04 -0700)]
dzn: Fix pipeline creation when rasterization is disabled
We use some of the VkGraphicsPipelineCreateInfo fields that should
be ignored when rasterization in disabled, assuming those who be set
to NULL by the caller in that case, which is not mandated by the
Vulkan specification.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15698>
Corentin Noël [Wed, 6 Apr 2022 13:58:46 +0000 (15:58 +0200)]
virgl/ci: Uprev virglrenderer and crosvm
Use a patch file for crosvm instead of relying on private repositories for faster
uprevs.
Use latest virglrenderer to keep the tests in-sync.
Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15772>
Corentin Noël [Wed, 6 Apr 2022 13:57:40 +0000 (15:57 +0200)]
ci: Only apply patches with the build-skqp prefix
Allows to ship patches for other components too.
Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15772>
Samuel Pitoiset [Wed, 6 Apr 2022 10:43:12 +0000 (12:43 +0200)]
radv: stop relying on shader modules after SPIRV->NIR
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/15766>
Samuel Pitoiset [Wed, 6 Apr 2022 10:37:47 +0000 (12:37 +0200)]
radv: stop passing the module to the compiler debug callback
After SPIRV->NIR, the driver shouldn't rely on the module. This will
still report messages via VK_EXT_debug_report but the object will be
NULL.
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/15766>
Samuel Pitoiset [Wed, 6 Apr 2022 10:25:33 +0000 (12:25 +0200)]
radv: drop the module reference for enable_mrt_output_nan_fixup
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/15766>
Samuel Pitoiset [Wed, 6 Apr 2022 10:54:37 +0000 (12:54 +0200)]
radv: drop the module reference in radv_can_dump_shader_stats()
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/15766>
Samuel Pitoiset [Wed, 6 Apr 2022 10:23:40 +0000 (12:23 +0200)]
radv: drop the module reference in radv_can_dump_shader()
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/15766>
Samuel Pitoiset [Wed, 6 Apr 2022 10:34:46 +0000 (12:34 +0200)]
radv: copy the spirv module for debugging after compilation
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/15766>
Samuel Pitoiset [Tue, 5 Apr 2022 08:27:28 +0000 (10:27 +0200)]
radv: remove more references to the pipeline layout during compilation
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15749>
Samuel Pitoiset [Tue, 5 Apr 2022 08:33:46 +0000 (10:33 +0200)]
radv: lower ycbcr textures just before applying the pipeline layout
This shouldn't change anything.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15749>
Samuel Pitoiset [Thu, 7 Apr 2022 09:47:45 +0000 (11:47 +0200)]
radv: assert that the arg is declared when used in get_scalar_arg()
Help debugging.
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/15789>
Samuel Pitoiset [Thu, 7 Apr 2022 12:21:31 +0000 (14:21 +0200)]
radv: add radv_is_vrs_enabled()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15792>
Samuel Pitoiset [Mon, 4 Apr 2022 10:28:49 +0000 (12:28 +0200)]
radv: add a new helper to initialize various type of pipelines
This is common to graphics, compute and library pipelines.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15792>
Samuel Pitoiset [Thu, 7 Apr 2022 11:54:12 +0000 (13:54 +0200)]
radv: add radv_generate_pipeline_key() for common graphics/compute keys
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15792>
Samuel Pitoiset [Thu, 7 Apr 2022 08:38:43 +0000 (10:38 +0200)]
radv: remove unused parameters in radv_get_{wave,ballot_bit}_size()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15792>
Samuel Pitoiset [Wed, 6 Apr 2022 12:32:30 +0000 (14:32 +0200)]
radv: use radv_pipeline_has_ds_attachments() more
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15792>
Samuel Pitoiset [Wed, 6 Apr 2022 12:29:12 +0000 (14:29 +0200)]
radv: do not check if VkPipelineRenderingCreateInfo is NULL
The driver converts legacy render pass to dynamic rendering, so this
structure should always be in pNext.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15792>
Iago Toral Quiroga [Fri, 1 Apr 2022 08:51:50 +0000 (10:51 +0200)]
broadcom/compiler: prefer reconstruction over TMU spills when possible
We have been reconstructing/rematerializing uniforms for a while, but we
can do this in more scenarios, namely instructions which result is
immutable along the execution of a shader across all channels.
By doing this we gain the capacity to eliminate TMU spills which not
only are slower, but can also make us drop to a fallback compilation
strategy.
Shader-db results show a small increase in instruction counts caused
by us now being able to choose preferential compiler strategies that
are intended to reduce TMU latency. In some cases, we are now also
able to avoid dropping thread counts:
total instructions in shared programs:
12658092 ->
12659245 (<.01%)
instructions in affected programs: 75812 -> 76965 (1.52%)
helped: 55
HURT: 107
total threads in shared programs: 416286 -> 416412 (0.03%)
threads in affected programs: 126 -> 252 (100.00%)
helped: 63
HURT: 0
total uniforms in shared programs: 3716916 -> 3716396 (-0.01%)
uniforms in affected programs: 19327 -> 18807 (-2.69%)
helped: 94
HURT: 50
total max-temps in shared programs: 2161796 -> 2161578 (-0.01%)
max-temps in affected programs: 3961 -> 3743 (-5.50%)
helped: 80
HURT: 24
total spills in shared programs: 3274 -> 3266 (-0.24%)
spills in affected programs: 98 -> 90 (-8.16%)
helped: 6
HURT: 0
total fills in shared programs: 4657 -> 4642 (-0.32%)
fills in affected programs: 130 -> 115 (-11.54%)
helped: 6
HURT: 0
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15710>
Connor Abbott [Thu, 7 Apr 2022 17:10:40 +0000 (19:10 +0200)]
freedreno/a6xx: Fix SP_DS_CTRL_REG0 definition
Bit 20 isn't actually MERGEDREGS, the mode for the entire geometry
pipeline is controlled by SP_VS_CTRL_REG0::MERGEDREGS and it appears to
be something preamble-related instead since writing any register in the
preamble hangs if it's set. This fixes those hangs on freedreno and
turnip since we no longer set it.
Fixes:
fccc35c2def ("ir3: Add preamble optimization pass")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15801>
Mike Blumenkrantz [Mon, 4 Apr 2022 23:08:13 +0000 (19:08 -0400)]
mesa/st: simplify st_can_add_pointsize_to_program iterator
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Thu, 31 Mar 2022 16:36:47 +0000 (12:36 -0400)]
mesa/st: don't precompile the pointsize upload variant anymore
this is no longer likely to be used, so precompile the base variant now
also delete some now-unused code
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Thu, 31 Mar 2022 16:31:29 +0000 (12:31 -0400)]
mesa/st: only flag pointsize constant uploads if they're needed
now that shaders are guaranteed to have a pointsize export, the only time
the variant using the uploaded constant is needed is when pointsize != 1.0
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Thu, 31 Mar 2022 16:40:28 +0000 (12:40 -0400)]
mesa/st: only use constant upload pointsize variants if pointsize != 1.0
it's not that common for apps to need varying pointsize, so now that shaders
are guaranteed to have the export in the shader, the constant version is almost
never used
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Mon, 4 Apr 2022 15:35:21 +0000 (11:35 -0400)]
mesa/st: also add pointsize to fixedfunction vertex shaders as needed
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Wed, 30 Mar 2022 13:24:03 +0000 (09:24 -0400)]
mesa/st: always inject a 1.0 pointsize for vertex stages
since 1.0 is used in nearly every case, drivers requiring this exporting
can avoid potential shader variants by adding a 1.0 export to the base
shader variant and the only using the ubo upload when pointsize is explicitly
set for wide point functionality
drivers can then be responsible for removing unused pointsize exports
as needed (or desired)
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Wed, 30 Mar 2022 16:46:21 +0000 (12:46 -0400)]
mesa/st: declare added pointsize var as hidden
ensure this isn't counted as part of the shader and ignored for e.g.,
glGetProgramInterfaceiv
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Wed, 30 Mar 2022 13:23:03 +0000 (09:23 -0400)]
nir/lower_point_size_mov: fix check for overwriting existing pointsize
this should match the comment and allow overwriting injected pointsize
variables regardless of whether xfb is flagged
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Mike Blumenkrantz [Wed, 30 Mar 2022 13:19:50 +0000 (09:19 -0400)]
mesa: add a bool indicating when pointsize == 1.0
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15699>
Emma Anholt [Thu, 7 Apr 2022 20:03:21 +0000 (13:03 -0700)]
Revert "ci/freedreno: Reduce concurrency when replaying traces on a630"
This reverts commit
d948f32365f53aa0677fef833ebaa2cc09d9ab67.
I think that fixing the timeout will have resolved this problem.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15805>
Emma Anholt [Thu, 7 Apr 2022 19:55:37 +0000 (12:55 -0700)]
freedreno: Fix the cpu-prep wait to be "infinite".
We don't need to restrict our timeout to 5 seconds, because the kernel's
hangcheck will ensure that the wait completes in finite time if the GPU
gets wedged. If the GPU is making progress, we don't want to time out
early and have pipe_transfer_map() return an error, causing glReadPixels()
to throw a confusing GL_OOM even though we're not out memory.
The INFINITE arg to this function isn't actually infinite, it's limited to
an hour. But an hour of GPU processing to wait on is probably plenty.
This 5s timeout has caused problems with the CTS on freedreno at high
parallelism, and I suspect is the cause of recent issues in the closed
traces replay jobs.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15805>
Timothy Arceri [Wed, 6 Apr 2022 02:47:52 +0000 (12:47 +1000)]
glsl: fix needs_lowering() call in varying packing pass
Here we remove the outer arrays on geom and tess shaders where
needed. Without this the pass can sometimes attempt to pack a
varying on only one side of the shader interface where it is not
actually needed. The result can be mismatching varying types.
Fixes:
d6b9202873f0 ("glsl: disable varying packing when its not safe")
Tested-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15761>
Mike Blumenkrantz [Thu, 7 Apr 2022 21:57:15 +0000 (17:57 -0400)]
docs: update features for lavapipe
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15808>
Mike Blumenkrantz [Thu, 7 Apr 2022 21:56:13 +0000 (17:56 -0400)]
zink: ci updates
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15808>
Mike Blumenkrantz [Thu, 7 Apr 2022 21:55:09 +0000 (17:55 -0400)]
zink: add error logging for SRGB framebuffer without KHR_swapchain_mutable_format
this is going to explode, so at least print an error explaining why
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15808>
Mike Blumenkrantz [Thu, 7 Apr 2022 21:19:52 +0000 (17:19 -0400)]
lavapipe: KHR_swapchain_mutable_format
it Just Works(tm)
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15808>
Mike Blumenkrantz [Thu, 7 Apr 2022 21:01:00 +0000 (17:01 -0400)]
zink: adds refs to user index buffers when tc is not active
there are no ref tricks to abuse in this case, so add our own ref
fixes #6273
cc: mesa-stable
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15807>
Mike Blumenkrantz [Thu, 7 Apr 2022 18:49:59 +0000 (14:49 -0400)]
zink: allow lod for RECT sampler types
the RECT is a lie, so this is fine
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15804>
Mike Blumenkrantz [Thu, 7 Apr 2022 18:49:07 +0000 (14:49 -0400)]
nir/lower_tex: avoid adding invalid LOD to RECT textures
this is illegal
Fixes:
74ec2b12be1 ("nir/lower_tex: Rework invalid implicit LOD lowering")
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15804>
Mike Blumenkrantz [Thu, 7 Apr 2022 16:04:57 +0000 (12:04 -0400)]
zink: set nir_shader_compiler_options::has_txs
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15804>
Jason Ekstrand [Tue, 5 Apr 2022 20:59:38 +0000 (15:59 -0500)]
nir: Dont set coord_components on txs
Fixes:
e1fc23265f9f ("nir: Add a pass for lowering CL-style image ops to texture ops")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15758>
Mike Blumenkrantz [Thu, 7 Apr 2022 18:28:55 +0000 (14:28 -0400)]
kopper: add a dmabuf-free image interface for use with sw drivers
sw drivers don't support modifiers or dmabufs or any of that, so separate
interfaces are needed to avoid advertising extensions that will only lead
to crashes
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15802>
Erik Faye-Lund [Wed, 6 Apr 2022 12:35:00 +0000 (14:35 +0200)]
dzn: add missing space
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Thu, 7 Apr 2022 11:43:27 +0000 (13:43 +0200)]
dzn: drop unused include
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Thu, 7 Apr 2022 12:02:57 +0000 (14:02 +0200)]
dzn: drop incorrect return statement
We're returning nothing here. Let's not do that.
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Fri, 25 Mar 2022 15:52:42 +0000 (16:52 +0100)]
dzn: drop unused header
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Fri, 25 Mar 2022 14:43:06 +0000 (15:43 +0100)]
dzn: remove unused variable
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Fri, 25 Mar 2022 14:09:00 +0000 (15:09 +0100)]
dzn: fixup indent
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Fri, 25 Mar 2022 13:11:44 +0000 (14:11 +0100)]
dzn: add D3D12_IGNORE_SDK_LAYERS define
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Fri, 25 Mar 2022 13:07:12 +0000 (14:07 +0100)]
dzn: drop needless includes
These include either dzn_nir.h or dzn_internal.h which already includes
this header.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Tue, 5 Apr 2022 15:31:45 +0000 (17:31 +0200)]
dzn: remove unused struct
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Erik Faye-Lund [Fri, 25 Mar 2022 12:16:26 +0000 (13:16 +0100)]
dzn: remove needless using
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15790>
Mike Blumenkrantz [Thu, 7 Apr 2022 13:16:55 +0000 (09:16 -0400)]
zink: handle swapchain readbacks when a present is pending
if a present semaphore is pending, don't try to create another one
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15784>
Mike Blumenkrantz [Wed, 6 Apr 2022 20:50:37 +0000 (16:50 -0400)]
zink: only apply swapchain behavior in flush_resource for swapchain images
these could otherwise be pixmaps or dmabufs
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15784>
Mike Blumenkrantz [Wed, 6 Apr 2022 20:49:03 +0000 (16:49 -0400)]
zink: handle deferred swapchain resource flushing
if the swapchain image hasn't been acquired yet, flag it as a deferred
present so it can be picked up on flush
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15784>
Mike Blumenkrantz [Wed, 6 Apr 2022 20:47:52 +0000 (16:47 -0400)]
egl: implement more hooks for swrast
these just need to use swapbuffers instead of flush
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15784>
Benjamin Cheng [Thu, 7 Apr 2022 16:56:49 +0000 (12:56 -0400)]
anv: drop from_wsi bit from anv_image
It was originally introduced in
ca791f5c but it was never actually set
anywhere. It doesn't serve any purpose other than some sanity checking
so let's clean it up for now.
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15799>
Ian Romanick [Mon, 14 Feb 2022 22:07:18 +0000 (14:07 -0800)]
intel/fs: Better handle constant sources of FS_OPCODE_PACK_HALF_2x16_SPLIT
I noticed that a *LOT* of fragment shaders in Shadow of the Tomb Raider,
for instance, end up with a sequence of NIR like:
vec1 32 ssa_2 = load_const (0x00000000 = 0.000000)
...
vec1 32 ssa_191 = pack_half_2x16_split ssa_188, ssa_2
vec1 32 ssa_192 = pack_half_2x16_split ssa_189, ssa_2
vec1 32 ssa_193 = pack_half_2x16_split ssa_190, ssa_2
This results in an assembly sequence like:
mov(8) g28<1>UD 0x00000000UD
mov(8) g21<2>HF g28<8,8,1>F
shl(8) g21<1>UD g21<8,8,1>UD 0x00000010UD
mov(8) g21<2>HF g25<8,8,1>F
mov(8) g19<2>HF g28<8,8,1>F
shl(8) g19<1>UD g19<8,8,1>UD 0x00000010UD
mov(8) g19<2>HF g23<8,8,1>F
mov(8) g20<2>HF g28<8,8,1>F
shl(8) g20<1>UD g20<8,8,1>UD 0x00000010UD
mov(8) g20<2>HF g24<8,8,1>F
After this commit, the generated assembly is:
mov(8) g21<1>UD 0x00000000UD
mov(8) g21<2>HF g23<8,8,1>F
mov(8) g19<1>UD 0x00000000UD
mov(8) g19<2>HF g17<8,8,1>F
mov(8) g20<1>UD 0x00000000UD
mov(8) g20<2>HF g18<8,8,1>F
Tiger Lake, Ice Lake, Skylake, and Haswell had similar results. (Ice Lake shown)
total instructions in shared programs:
20119086 ->
20119034 (<.01%)
instructions in affected programs: 9056 -> 9004 (-0.57%)
helped: 8
HURT: 0
helped stats (abs) min: 2 max: 16 x̄: 6.50 x̃: 4
helped stats (rel) min: 0.29% max: 1.75% x̄: 1.00% x̃: 0.98%
95% mean confidence interval for instructions value: -11.01 -1.99
95% mean confidence interval for instructions %-change: -1.56% -0.44%
Instructions are helped.
total cycles in shared programs:
861019414 ->
861021044 (<.01%)
cycles in affected programs: 279862 -> 281492 (0.58%)
helped: 4
HURT: 2
helped stats (abs) min: 6 max: 936 x̄: 239.00 x̃: 7
helped stats (rel) min: 0.03% max: 8.13% x̄: 2.09% x̃: 0.09%
HURT stats (abs) min: 18 max: 2568 x̄: 1293.00 x̃: 1293
HURT stats (rel) min: 0.36% max: 1.14% x̄: 0.75% x̃: 0.75%
95% mean confidence interval for cycles value: -972.56 1515.89
95% mean confidence interval for cycles %-change: -4.77% 2.49%
Inconclusive result (value mean confidence interval includes 0).
Broadwell
total instructions in shared programs:
17812327 ->
17812263 (<.01%)
instructions in affected programs: 9867 -> 9803 (-0.65%)
helped: 8
HURT: 0
helped stats (abs) min: 2 max: 28 x̄: 8.00 x̃: 4
helped stats (rel) min: 0.32% max: 1.80% x̄: 1.00% x̃: 0.95%
95% mean confidence interval for instructions value: -15.46 -0.54
95% mean confidence interval for instructions %-change: -1.54% -0.47%
Instructions are helped.
total cycles in shared programs:
904768620 ->
904773291 (<.01%)
cycles in affected programs: 454799 -> 459470 (1.03%)
helped: 4
HURT: 4
helped stats (abs) min: 36 max: 586 x̄: 344.50 x̃: 378
helped stats (rel) min: 0.47% max: 4.04% x̄: 2.01% x̃: 1.77%
HURT stats (abs) min: 1 max: 5572 x̄: 1512.25 x̃: 238
HURT stats (rel) min: <.01% max: 2.77% x̄: 1.46% x̃: 1.53%
95% mean confidence interval for cycles value: -1122.40 2290.15
95% mean confidence interval for cycles %-change: -2.26% 1.71%
Inconclusive result (value mean confidence interval includes 0).
total spills in shared programs: 18581 -> 18579 (-0.01%)
spills in affected programs: 323 -> 321 (-0.62%)
helped: 1
HURT: 0
total fills in shared programs: 24985 -> 24981 (-0.02%)
fills in affected programs: 1348 -> 1344 (-0.30%)
helped: 1
HURT: 0
Tiger Lake, Ice Lake, and Skylake had similar results. (Ice Lake shown)
Instructions in all programs:
143585431 ->
143513657 (-0.0%)
Instructions helped: 14403
Cycles in all programs:
8439312778 ->
8439371578 (+0.0%)
Cycles helped: 10570
Cycles hurt: 3290
Gained: 146
Lost: 74
All of the lost and gained fossil-db shaders are SIMD32 fragment
shaders. 14,247 of the affected shaders are from Shadow of the Tomb
Raider. 154 are from Batman Arkham Origins, and the remaining two are
from Octopath Traveler.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15089>
Alyssa Rosenzweig [Sun, 27 Feb 2022 20:46:17 +0000 (15:46 -0500)]
pan/bi: Imply round mode most of the time
Much less noisy, and provides a path to further improvements. There is a slight
behaviour change: int-to-float conversions now use RTE instead of RTZ. For
32-bit opcodes, this affects conversions of integers with magnitude greater than
2^23 by at most 1 ulp. As this behaviour is unspecified in GLSL, this change is
believed to be acceptable.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15187>
Alyssa Rosenzweig [Sun, 27 Feb 2022 20:22:03 +0000 (15:22 -0500)]
pan/bi: Use should_skip in bi_builder generation
To avoid further code duplication.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15187>
Alyssa Rosenzweig [Sun, 27 Feb 2022 20:46:03 +0000 (15:46 -0500)]
pan/bi: Mark some opcodes as default round-to-zero
Conversions to integer have different rounding rules.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15187>
Alyssa Rosenzweig [Sun, 27 Feb 2022 19:46:15 +0000 (14:46 -0500)]
pan/bi: Don't use funny round modes in tests
To prepare for defeaturing round modes, replace uses of round-to-positive with
round-to-even in our unit tests. This doesn't meaningfully impact test coverage;
there is no way to generate that round mode.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15187>