This is required for a bunch of stuff that can occur in any Vulkan
shader stage, not just these few. So let's always emit this cap.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10371>
internal_usage is the memory allocated by the current process (intent)
while system_usage is the memory allocated globally (actual).
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/9774>
The pointers won't be used at that point. Just decrement the refcounts.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.eric.pelloux-prayer@amd.com>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10606>
A lot of space was wasted due to 16-byte alignment for slots. This new
layout tries to match glthread. Highlights:
- the slot size changed to 8 bytes (was 16), so less padding
- the header size changed to 4 bytes (was 8), so some calls can use
the remaining 4 bytes in the slot for parameters
- draw merging merges up to 307 draws (was 256) due to space savings
- parameters in structures are sorted based on implicit type alignment
(uint8_t first, pointers last) to make it easier to utilize the 4 bytes
after the header and to remove holes
- some structures use smaller types for parameters than pipe_context where
it's safe (e.g. clear uses float instead of double for depth)
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.eric.pelloux-prayer@amd.com>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10606>
This reverts commit 3b1ce49bc1.
It will be completely rewritten, but let's revert this first.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.eric.pelloux-prayer@amd.com>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10606>
Not that Z32F is especially common, but we shouldn't consider it to
imply that the color buffers are also float, which is what floatMode
is meant to mean.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10491>
The blitter commands don't show up in INTEL_DEBUG=bat, but on
gen4/5 they are emitted on the render engine ring so just change
the XML to reflect that.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10636>
MI_LOAD_REGISTER_MEM is only available on gen7+, so avoid build errors
on earlier generations.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10636>
sparse buffers are not cpu-readable, so any mapping requires that they
use a staging buffer, either the stream uploader for writes or a manual
copy for readback
future work here should attempt to resolve two perf issues:
* sparse allocations should be allocated dynamically using a suballocator
or some other, more useful strategy
* readback shouldn't allocate a huge staging buffer
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10509>
On Wayland, if the wl_drm interface is not available, for example if the
compositor is using the proprietary NVIDIA driver along with their egl-wayland
library, the device_select layer will fail to initialize. However, the failure
path will unconditionally call wl_drm_destroy even though info.wl_drm would be
NULL in that case. This can cause a segfault in libwayland-client.so.
To fix this, check if info.wl_drm is NULL before calling wl_drm_destroy. This
way, initialization will fail gracefully even if that interface is not present.
Signed-off-by: Erik Kurzinger <ekurzinger@nvidia.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10598>
Move it out of the "cs" sub-struct, since the bit can be used for
other shader stages in the future.
This also removes a subtle issue in spirv_to_nir:
info.cs.shared_memory_explicit_layout was used without checking for
the CS shader stage. It ended up being "harmless" since the effects
also depended on presence of shared variables.
Fixes: 5de6c5973a ("spirv: Implement SPV_KHR_workgroup_memory_explicit_layout")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10529>
We were implictly including it in exposing VK 1.2, but we weren't making
use of the supplied struct. Actually enabling it gives us a chance to do
slightly better at Z/S UBWC, and means we won't lose the separate usage
test coverage when switching back to exposing VK 1.1.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10594>
This makes us handle all non-aggregate types before we handle aggregate
types. This is going to matter in the next commit.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10372>
