It was based on reverse engineering and inaccurate.
This also get ride of fp64 usage and will allow us to use it in OpenCL C
shaders for indirect/indexed draw.
On the full 32-bit range, this fixes ~10 millions inaccurate results.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Fixes: 467ae0d39d ("panfrost: Factor out panfrost_compute_magic_divisor")
Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35490>
Our existing descriptor implementation already supports
update-after-bind on both bifrost and valhall, but on bifrost the some
of the descriptor limits are lower than the minimum required by the
extension.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
Previously, our descriptor set limits were set based on the hardware
limits on bifrost. On valhall, the hardware descriptor table limits are
much higher. We need to advertise higher update-after-bind descriptor
limits for descriptor indexing, so might as well increase these also.
The 2^20 limit is the same as nvk and honeykrisp.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
We need to lower everything on bifrost, and everything except nonuniform
texture offsets on valhall. With texture offsets, the hardware supports
nonuniform offsets passed in a staging register, but does not support
passing nonuniform offsets as src1 in the narrow layout. We need to make
sure we check for this if we implement narrow offsetms later.
Input attachments are skipped in this commit because they need
additional changes.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
Runtime descriptor array doesn't require any changes, variable descriptor
count is already mostly supported and just needs handling for
maxVariableDescriptorCount.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
We already do this for push constants, just extend the existing
implementation. We need to update the push_consts sysval inside the
cmd_prepare_push_uniforms loop now because indirect draws patch the
sysvals. Previously this value was only used to read push constant FAUs,
so we didn't care.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
These don't require any changes for support.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
This simplifies get_physical_device_properties and allows us to use the
limits elsewhere.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
We were not able to use these previously because the headers require
PAN_ARCH.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
This makes things a little more flexible for groups of fields that are
arch-dependent, and allows us to use existing per-arch constant macros
instead of open-coding their values.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
The name of the struct that we're currently using is panvk_ssbo_addr.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
This is needed for the panvk_descriptor_state struct. All of the current
sites already included panvk_cmd_desc_state.h, so it wasn't a visible
issue.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35408>
PresentWait2 should be possible on any physical device, as it adds a
surface capability query that depends on common wsi code.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35414>
vulkanMemoryModel was the last required feature we needed to expose
Vulkan 1.3, so let's enable it!
Reviewed-by: Caterina Shablia <caterina.shablia@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35502>
We have all the pieces lined up now, so let's enable the memory model
extension.
Reviewed-by: Caterina Shablia <caterina.shablia@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35502>
It seems like a good idea to optimize barriers, so let's do the same
thing as most other drivers do.
Reviewed-by: Caterina Shablia <caterina.shablia@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35502>
NONE-barriers also needs scheduling barriers, because otherwise we can
end up with memory loads and stores being reordered during scheduling.
Reviewed-by: Caterina Shablia <caterina.shablia@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35502>
The MEMORY_BARRIER instruction has some issues, where we end up
dead-code eliminating it before it gets to do what it's supposed to do.
But even if we fix that, we have issues where we can end up inserting
flow control into it, which isn't going to work because we have nothing
to emit here either.
So let's rework this to a special-cased NOP instruction, which is marked
as a scheduling barrier. The beneft here is that NOPs are already properly
handled when it comes to flow control.
Note that this isn't perfect either; this only prevents memory operations
from crossing the scheduling barrier. We should really prevent any
operation with observable side effects from crossing the barrier. This
includes things like reading clocks etc.
But that's a larger change, and it's a step in the right direction to get
this to no longer be dead-code eliminated. So let's put this band-aid on
for now.
Fixes: f77a50e45e ("pan/bi: add a MEMORY_BARRIER pseudo-instruction")
Reviewed-by: Caterina Shablia <caterina.shablia@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35502>
Currently a cs_maybe block affects the ls_tracker of a command stream
even if the cs_maybe block is later replaced by NOP instructions. This
can lead to missing WAIT instructions in the command stream.
* Add a WAIT at the end of the cs_maybe block if the block has changed
the ls_tracker state
* At the end of a cs_maybe block, restore the ls_tracker state to what
it was before the block
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13259
Fixes: 885805560f ("panvk/csf: fix case where vk_meta is used before PROVOKING_VERTEX_MODE_LAST")
Tested-By: Alexandre ARNOUD <aarnoud@me.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Christoph Pillmayer <christoph.pillmayer@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35460>
This matches our existing behavior on both JM and CSF, where when
depthClamp is disabled, we clamp to [0,1].
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Reviewed-by: Eric R. Smith <eric.smith@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35224>
Do not perform a queue family ownership transfer during memory barriers
if srcQueueFamilyIndex equals dstQueueFamilyIndex, as the Vulkan spec
mandates that this should only happen if the two values are unequal.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Fixes: 715d6e740a ("panvk: improve VK_QUEUE_FAMILY_EXTERNAL support")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35458>
LOAD_MULTIPLE can be emitted inside of a loop body. We need to WAIT if
that loads targets a register for which a load was not already in-flight
at the start of the loop body.
Technically we only have to emit the wait if the dst reg of a new load
is actually used inside the loop, but that would require separate
tracking of source regs used in the loop and is probably not worth
the effort for now.
Fixes: f75569734e (panvk: Remove explicit LS waits)
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35436>
8x and 16x MSAA should work now, so we no longer need to deal with those
separately. We have more accurate checks for when it's supported or not
in place already.
2x MSAA isn't supported at all until v12, which we haven't finished
wiring up yet. So let's leave that one for now... We'll need some more
fixes for that anyway.
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Reviewed-by: Eric R. Smith <eric.smith@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34455>
iter_sb is now unused on v11+, let's not store it on this codepath and
only define it for v10.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35307>
Now that all paths support indirect wait for iter scoreboards, we can
remove the previous limit.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35307>
Now that cs_next_iter_sb match NEXT_SB_ENTRY behavior, we can just
implement it with it.
We also setup the scoreboard wait mask when selecting the next
scoreboard entry to prepare for defer indirect wait and signal usage
around the codebase.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35307>
This closely match what NEXT_SB_ENTRY gives us on v11+ and should also
reduce the number of instruction by low margin.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35307>
This patch changes iter_sb to contains the actual scoreboard info
instead of the index in the iter range.
This is required to handle NEXT_SB_ENTRY on v11+.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Lars-Ivar Hesselberg Simonsen <lars-ivar.simonsen@arm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35307>
The ls_tracker will re-emit the WAIT for cs_load32_to(..attrib_offset..)
since it is only used in one branch of an if-else below. Thus, it is
better to manually flush it before.
Fixes: f75569734e (panvk: Remove explicit LS waits)
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35425>
For cs_if_else the ls_tracker needs to be setup such that the loads and
stores of both branches are accounted for separately.
Fixes: f75569734e (panvk: Remove explicit LS waits)
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35425>
The AFBC path is behind a driconf option, and so was not tested by any
existing CI jobs. We had a regression with this that went unnoticed for
several months. To avoid similar situations in the future, add AFBC
smoke tests to CI, similar to the existing spilling smoke tests..
Some tests on g52 fail instead of crashing when AFBC is enabled, but
otherwise the CI expectations are identical.
Signed-off-by: Olivia Lee <olivia.lee@collabora.com>
Acked-by: Eric R. Smith <eric.smith@collabora.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35193>
We recently found a weird case where an image was re-interpreted with
a format that had a bigger size than the image format, which led the
resulting texture descriptor to have a size that was bigger than what
the plane descriptors were covering.
Let's make sure this sort of unexpected use don't slip through in the
future by introducing basic image view consistency checks.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35317>
Adds support for reading PrimitiveID in a fragment shader assuming the
previous shader was vertex.
If support for shaders that can modify PrimitiveID is added, additional
changes will be required.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35367>
For formats that do not support LogicOps, NOOP should be treated as
COPY. This was implemented in nir/lower_blend, but an optimization in
PanVK would trigger for all NOOP that stopped this from happening.
This commit limits the NOOP optimization to supported formats.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35367>
Now that we use a precomp shader for indirect dispatch,
we can safely remove this.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32839>
