I added this FIXME 2 years ago because it was unclear if it was
broken or not. Since, CTS coverage improved and the number of tests
with depth/stencil on the compute queue increased a lot. vkd3d-proton
also widely uses depth/stencil with GENERAL on GFX10+ and likely with
async compute as well. No issues so far.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4048
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20785>
The problematic sequence of draws is pretty rare. But there are a small
handful of games which do not exhibit the problematic sequence and for
which invalidating LRZ on draws with blend plus depthwrite enabled hurts
performance slightly. This driconf option enables opting in to the
previous behavior.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20708>
Writing depth with blend enabled means we need to invalidate LRZ,
because the written depth value could mean that a later draw with
depth enabled (where we would otherwise write LRZ) could have
fragments which don't pass the depth test due to this draw. For
example, consider this sequence of draws, with depth mode GREATER:
draw A:
z=0.1, fragments pass
draw B:
z=0.4, fragments pass
blend enabled (LRZ write disabled)
depth write enabled
draw C:
z=0.2, fragments don't pass
blend disabled
depth write enabled
Normally looking at the state in draw C, we'd assume we could
enable LRZ write. But this would cause early-z/lrz to discard
fragments from draw A which should be visible due to draw B.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8065
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20708>
Don't emit the NGG query user SGPR if its state doesn't change.
Based on original work by Mike Blumenkrantz.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18819>
Field must be disabled if any render targets have integer
format, additionally for Gfx12+ field must be disabled when
num multisamples > 1 or forced multisample count > 1.
Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7892
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20671>
Field must be disabled if any render targets have integer format.
Cc: mesa-stable
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20671>
Field must be disabled if any render targets have integer
format, additionally for Gfx12+ field must be disabled when
num multisamples > 1 or forced multisample count > 1.
Cc: mesa-stable
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20671>
Make sure we comply with BSpec and ensure that certain flush flags
are not set for compute batches
Signed-off-by: Rohan Garg's avatarRohan Garg <rohan.garg@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15664>
offset field in kgsl_command_object is NOT used by KGSL, so
we should offset directly to iova.
Fixes weird hangs on KGSL. E.g. fixes the hang in:
dEQP-VK.memory.pipeline_barrier.transfer_dst_storage_texel_buffer.1024
cc: mesa-stable
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20795>
There was a VUID-VkImageViewCreateInfo-image-04739 in the Vulkan 1.3
spec that said:
If image was created with the
VK_IMAGE_CREATE_BLOCK_TEXEL_VIEW_COMPATIBLE_BIT flag and format is a
non-compressed format, viewType must not be VK_IMAGE_VIEW_TYPE_3D
That VUID has since been removed, and when a view of a 3D image is
created, with put the depth into the array_len, so it won't be always 1.
Reviewed-by: Mark Janes <markjanes@swizzler.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20803>
For a depth-only-with-discard pipeline, spi_shader_col_format needs to be
fixed up to a single channel export, or otherwise discard will not work.
Since col_format can change depending on the dynamic state, precompute the
need for this workaround on pipeline creation and apply it when emitting
prolog states.
Fixes: eb07a11b8f ("radv: add support for compiling PS epilogs on-demand")
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20704>
Some D3D12 drivers, like my PC's AMD driver, don't like using a
dynamic index to load from a constant buffer that's bound via
root constants. Instead, just go ahead and load the full set of
vertex data and just bcsel which one to use.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20778>
Better error handling is more reliable.
Options:
-L, follow location
--retry, number of retries
--retry-all-errors, does not fail on ALL errors, that's why there is -f
-f, fail fast with no output at all on server errors
--retry-delay, make curl sleep this amount of time before each retry
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20788>
This is a weird way to do queries, but in multiview, each query
takes up N slots, where N is the number of views. D3D doesn't do
it that way, and only has one result, which fortunately is a valid
way to do Vulkan queries. We just need to take care to zero out
the other view results, and make sure they get "signaled" when
the cmdbuf is submitted.
Note that it is invalid in D3D to use ResolveQueryData on query
slots that have never actually been begun/ended, so we zero out
the data by copying zeroes into the buffer. This probably could
be optimized but oh well.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20650>
For draws, when we're emulating multiview, we need to loop them
and set up the right sysval. For clears, we always need to loop.
When not emulating, we also need to set up the right view instance
mask.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20650>
D3D's view instancing is an optional feature, and even when it's
supported, it only goes up to 4 views, where Vulkan requires a
minimum of 6 supported views. So, we need to have a path for handling
the cases where we can't use the native feature.
In this mode, pass the view ID as a runtime var. The caller is then
responsible for looping the draw calls and filling out the constant
buffer value correctly for each draw. When we get to the last pre-rast
stage, we'll additionally want to write out gl_Layer to select the
right RTV array slice. Lastly, for the fragment shader, if there's
any input attachments, those get loaded using the RTV slice instead
of the view ID. RTV slice input into the PS is done with a signature
entry (which must be output from the previous stage) rather than a
system value.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20650>
This adds support for D3D12-native view instancing to the compiler.
Essentially, it's just the ability to load SV_ViewID (dx.op.viewID),
set the right capability, and fill out some more PSV data. Note that
the PSV data is currently garbage. Ideally, we'd fill out a proper
input -> output and viewID -> output dependency table, but AFAIK
this is only used to enforce D3D API validation, and drivers ignore
it, so it's less critical to get it right.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20650>
