Stops gamescope from recompiling pipelines on every start.
Cc: mesa-stable
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15454>
In the future, whether we need to use fma will depend on which
multiplication is chosen.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14769>
The v_fma_mix optimizations assume v_cvt_f16_f32 and v_mul_f16 use a v2b
definition.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14769>
It seems the default D3D behavior and it's complicated to emulate this
in DXVK/vkd3d. Enable it by default to prevent rendering issues in
other games not listed here.
Cc: 22.0 mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15368>
If the LLVM version (even minor) isn't the same on the OS that
precompiles shaders vs the OS that runs them, the cache UUID would
be different, even if only ACO is used.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15406>
The notifier state must be destroyed when a device is destroyed. Oops.
This fixes crashes at launch with The Witcher 3.
Fixes: c50557d961 ("radv: allow applications to dynamically change RADV_FORCE_VRS")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15405>
dEQP-VK.api.object_management.multithreaded_shared_resources.image_2d
is flakey crashing on RENOIR and VEGA10.
Thanks Rhys Perry for pointing out the flake.
Acked-by: Martin Roukala <martin.roukala@mupuf.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15386>
We don't have to clear other internal BOs when it's set for eg. vkd3d.
This should reduce the number of SDMA clears.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15156>
From the Vulkan spec:
"drmFormatModifierTilingFeatures is a bitmask of
VkFormatFeatureFlagBits that are supported by any image created
with format and drmFormatModifier. The returned
drmFormatModifierTilingFeatures must contain at least one bit."
This fixes recent CTS dEQP-VK.drm_format_modifiers.*.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15383>
This extension is quite old and useless now.
VK_KHR_pipeline_executable_properties should be used instead.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15299>
Since neither PKT3_LOAD_SH_REG_INDEX nor PKT3_COPY_DATA work with compute
queues on GFX7-10, we have to load the dispatch size from memory in the
shader.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15064>
An example is the num_work_groups argument. Fixes invalid assembly with
func.compute.num-workgroups.basic.q0
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15064>
This is useless because all fields should be correctly filled if the
pNext struct is found.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15164>
cmd_buffer->pool should never be NULL. Even if AllocateCommandBuffers()
fails, the successfully created cmdbuffers would have it set correctly.
From the Vulkan spec:
"VUID-vkFreeCommandBuffers-pCommandBuffers-parent
Each element of pCommandBuffers that is a valid handle must have
been created, allocated, or retrieved from commandPool."
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15164>
This function seems rarely used or maybe never but I noticed this.
From the Vulkan spec:
"VUID-vkTrimCommandPool-commandPool-parameter
commandPool must be a valid VkCommandPool handle".
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15164>
Otherwise inputs_read doesn't contain the information. This shouldn't
fix anything in practice because radv_shader_info gathers this from
the variable.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15337>
For weird reasons, the COPY_DATA packet doesn't seem to copy anything
while on the compute queue. Instead, use PKT3_LOAD_SH_REG_INDEX which
seems to work as expected.
Note that LOAD_SH_REG_INDEX on the compute queue is only supported by
the CP on GFX10.3, so we need to implement a different solution (load
from the indirect BO in the shader) for older generations.
This should fix the Control RT GPU hang.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15053>
Move it to the pipeline creation to reduce computations in the hot path.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15162>
RadeonSI uses a different terminology and info->blocks is actually the
number of threads, not the number of blocks (ie. info->grid).
Found by inspection.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15162>
The WSI code wants to remain generic and try and use vulkan APIs,
even though these queues aren't exposed through the API.
Add the transfer queue to the end of the queues.
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Mike Lothian <mike@fireburn.co.uk>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15357>
If we introduce another queue type (video decode) we can have a
disconnect between the RADV_QUEUE_ enum and the API queue_family_index.
currently the driver has
GENERAL, COMPUTE, TRANSFER which would end up at QFI 0, 1, <nothing>
since we don't create transfer.
Now if I add VDEC we get
GENERAL, COMPUTE, TRANSFER, VDEC at QFI 0, 1, <nothing>, 2
or if you do nocompute
GENERAL, COMPUTE, TRANSFER, VDEC at QFI 0, <nothing>, <nothing>, 1
This means we have to add a remapping table between the API qfi
and the internal qf.
This patches tries to do that, in theory right now it just adds
overhead, but I'd like to exercise these paths.
v2: add radv_queue_ring abstraction, and pass physical device in,
as it makes adding uvd later easier.
v3: rename, and drop one direction as unneeded now, drop queue_family_index
from cmd_buffers.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13687>
