This allows to set VK_PRIMITIVE_TOPOLOGY_PATCH_LIST dynamically when
tessellation used.
If other values are set via vkCmdSetPrimitiveTopologyEXT for the case,
the validation layer can detect the issue.
Fixes dEQP-VK.pipeline.extended_dynamic_state.*.topology_patch*
Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Reviewed-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12299>
It can happen that the user reads an input attachment as the first use
of that attachment. In that case there are no subpass dependencies
required at all, because there could be a pipeline barrier before the
renderpass instead, and in any case we assume that dependencies with the
first subpass as a destination can be executed only once outside the
renderpass. The result is that we only do a CACHE_INVALIDATE once
before the entire renderpass, but it's actually required after each GMEM
load, because input attachments read GMEM through UCHE and those writes
to GMEM invalidate UCHE.
While we could add the missing CACHE_INVALIDATE "by hand" somehow, it
turns out it's actually just as easy to do an optimization the blob
does, where it simply doesn't patch those input attachments and reads
them directly instead. This means we can skip allocating memory in GMEM
for them entirely in some circumstances.
This fixes e.g.
dEQP-VK.api.copy_and_blit.core.resolve_image.whole_array_image.4_bit
with TU_DEBUG=forcebin.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12213>
Move away from using gpu_id as the primary means to identify which
adreno we are running on, as future GPUs (starting with 7c3) stop
providing a gpu_id as a new naming scheme is introduced.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12159>
... and reduce maxDescriptorSetUpdateAfterBindStorageBuffersDynamic from 12 to
8.
MAX_DYNAMIC_BUFFERS is MAX_DYNAMIC_UNIFORM_BUFFERS +
MAX_DYNAMIC_STORAGE_BUFFERS. We set
maxDescriptorSetUniformBuffersDynamic = MAX_DYNAMIC_UNIFORM_BUFFERS
maxDescriptorSetStorageBuffersDynamic = MAX_DYNAMIC_STORAGE_BUFFERS
maxDescriptorSetUpdateAfterBindUniformBuffersDynamic = MAX_DYNAMIC_BUFFERS / 2
maxDescriptorSetUpdateAfterBindStorageBuffersDynamic = MAX_DYNAMIC_BUFFERS / 2
The CTS test checks that
maxDescriptorSetUpdateAfterBindUniformBuffersDynamic
- is at least 8; and
- is at least maxDescriptorSetUniformBuffersDynamic
maxDescriptorSetUpdateAfterBindStorageBuffersDynamic
- is at least 4; and
- and is at least maxDescriptorSetStorageBuffersDynamic
Prior to this patch maxDescriptorSetUpdateAfterBindUniformBuffersDynamic was 12
but maxDescriptorSetUniformBuffersDynamic was 16, thus causing the CTS failure
in
dEQP-VK.api.info.vulkan1p2_limits_validation.ext_descriptor_indexing
By raising maxDescriptorSetUpdateAfterBindUniformBuffersDynamic to the same
value as maxDescriptorSetUniformBuffersDynamic, we bring the limits into the
appropriate ranges. We do the same thing for
maxDescriptorSetUpdateAfterBindStorageBuffersDynamic by assigning it the same
value as maxDescriptorSetStorageBuffersDynamic.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12193>
This is supposed to be set when that stage needs the PrimID sysval
preloaded, except for the VS which doesn't have this bit and instead
infers it from the HS or GS bit (depending on whether tess/GS is
enabled). Therefore for HS, GS, and DS we should set it whenever the
corresponding sysval is there. This includes adding a missing
PC_HS_OUT_CNTL, which I confirmed is set when the HS reads PrimID from
the VS. Note that the DS sysval is currently always enabled whenever
there's a GS, if we were to fix that then we should also change the
logic here.
This doesn't fix anything that I know of, but aligns us more with what
the blob does.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12166>
The previous handling conflated RelPatchID and PrimID, which would
result in incorrect gl_PrimitiveID when doing draw splitting and didn't
work with PrimID passthrough which fills the VPC slot with the "correct"
PrimID value from the tess factor BO which we left 0. Replace PrimID in
the tess lowering pass with a new RelPatchID sysval, and relace PrimID
with RelPatchID in the VS input code in turnip/freedreno at the same
time so that there is no net change in the tess lowering code. However,
now we have to add new mechanisms for getting the user-level PrimID:
- In the TCS it comes from the VS, just like gl_PrimitiveIDIn in the GS.
This means we have to add another register to our VS->TCS ABI. I
decided to put PrimID in r0.z, after the TCS header and RelPatchID,
because it might not be read in the TCS.
- If any stage after the TCS uses PrimID, the TCS stores it in the first
dword of the tess factor BO, and it is read by the fixed-function
tessellator and accessed in the TES via the newly-uncovered DSPRIMID
field. If we have tess and GS, the TES passes this value through to
the GS in the same way as the VS does. PrimID passthrough for reading
it in the FS when there's tess but no GS also "just works" once we
start storing it in the TCS. In particular this fixes
dEQP-VK.pipeline.misc.primitive_id_from_tess which tests exactly that.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12166>
DSPATCHID and HSPATCHID, which we mapped gl_PrimitiveID to, are actually
relative to the current subdraw. Subdraws aren't supported yet by turnip
but they are by freedreno for indirect draws.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12166>
If secondary command buffer is emitted within a subpass it may have
barriers which forces us to disable gmem for current renderpass.
Fixes: 20547a110e "tu: delay decision of forcing sysmem due to subpass self-dependencies"
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12219>
We would determine that it was unsupported, then ask for the size and
triggered the assertion checking that we never ask for the size of a
combined sampler.
Fixes: ee3495e465 ("turnip: Add support for VK_VALVE_mutable_descriptor_type")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12148>
Blob advertises { 1024, 1024, 64 }, but from tests they all
could be 1024.
Fixes tests:
dEQP-VK.compute.basic.max_local_size_x
dEQP-VK.compute.basic.max_local_size_y
dEQP-VK.compute.basic.max_local_size_z
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9409>
We were trying to calculate how much space they need, That was already
difficult and one of the most opaque and hard-to-verify uses of sub_cs,
but it will become even more difficult with the 3D path. What's worse is
that sometimes we have to touch that path when we start touching
registers that would affect rasterization, and there's no indication
that you have to then recalculate the size etc. Just rip this out and
start keeping a separate CS for it instead. Note that this adds a small
amount of memory wastage and extra buffers (at worst one buffer per
command buffer).
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12102>
This is much more maintainable, extensible, and easy to read than
hand-rolled structs approximating assembly. This also removes the last
use of the old hand-written packing structs. There are a few minor
differences:
- The shaders are larger because ir3 currently doesn't support (rpt),
which means that some shaders are larger than one instrlen and the
current logic has to be extended to allow for that. This seems a small
price to pay, ir3 will gain support for (rpt) eventually, and we
shouldn't have limitations like this baked in anyway. For example some
GL blob r8g8 <-> r16 copy shaders are apparently quite large.
- Due to the inability to switch inputs/outputs on the fly, we need to
split the VS into two variants. I made the layer-writing variant also
used for other clears, because the old method of overloading c0.z/c1.z
to mean both "src x coordinate" and "z clear value" in the same shader
seemed too clever and I didn't want to add yet another variant. This
means that non-layered clears will also write the layer (to 0), but
that shouldn't be a big deal performance-wise.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12079>
All vulkan drivers have been copying anv's code to convert
VkSpecializationInfo into nir_spirv_specialization.
Recently there was a Vulkan spec change on allowed values for
VkSpecializationInfo, and all drivers got affected.
This commits creates a new helper, and uses it on all Vulkan Mesa
drivers.
v2: use (uint8_t*) castings, instead of void*, to avoid C2036 with
MSVC (detected by the CI, inspired on what radv was doing)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12047>
DXVK always inserts vertex stage subpass self-dependency for every
subpass regardless of whether there actually would be a barrier.
This effectively disabled gmem rendering with DXVK.
Thus we delay the decision to disable gmem rendering until we
see a barrier with vertex stages.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12038>
v1. Hyunjun Ko <zzoon@igalia.com>
- Add to hanlde VK_DESCRIPTOR_POOL_CREATE_HOST_ONLY_BIT_VALVE
- Don't support VK_DESCRIPTOR_TYPE_INPUT_ATTACHMENT and
VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER
v2. Hyunjun Ko <zzoon@igalia.com>
- Fix some indentations and nitpicks.
- Add the extension to features.txt
v3. Hyunjun Ko <zzoon@igalia.com>
- Remove unnecessary asserts.
Signed-off-by: Eduardo Lima Mitev <elima@igalia.com>
Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Reviewed-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9322>
Newer a6xx devices drop this packet from the sqe firmware, and use
direct (pkt4) register writes instead for the few cases that previously
used CP_REG_WRITE.
The turnip part was adapted from Jonathans patch on !10892
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
This way we can make the tables const. At the same time, for a6xx, this
introduces a "sub-generation template" to reduce the copy/paste for
parameters which are keyed to the sub-generation. It also explicitly
lists every supported GPU, to get rid of duplicate lists of supported
gpus between the device-info and drivers.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
Replace it with a calculation which works for all current GPUs.
Duplicated the calculation in both drivers because freedreno_dev_info isn't
meant for derived parameters (and drivers might want to just calculate on
the fly instead).
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
- It hurts users with newer firmware who don't need the workaround
- Kernel now rejects older firmware due to security issues, so this will
prevent users from using older firmware anyway.
- Only whitelisting 650 enables the workaround by default for any new GPUs
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11790>
Emit all the state layout config (such as push-const CONSTLEN) first,
before emitting anything that depends on that state. This fixes an
issue that was showing up when FLUT is enabled in ir3 (which results
in higher probability of not having any immediats lowered to push-
consts).
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8705>
It's only used in assert. Avoids compiler warning/error with assertions disabled:
../src/freedreno/vulkan/tu_cs.h: In function 'tu_cs_reserve':
../src/freedreno/vulkan/tu_cs.h:208:13: error: unused variable 'result' [-Werror=unused-variable]
208 | VkResult result = tu_cs_reserve_space(cs, reserved_size);
| ^~~~~~
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11412>
We say that they're for debug only but we don't really have a good
policy around when to set them and when not to. In particular,
nir_lower_system_values and nir_lower_vars_to_ssa which are the chief
producers of SSA values which might reasonably have a name do not bother
to set one. We have some names set from things like BLORP and RADV's
meta shaders but AFAICT, they're setting a name more because it's there
than because they actually care.
Also, most things other than nir_clone and nir_serialize don't bother to
try and preserve them. You can see in the diffstat of this commit
exactly what passes attempt to preserve names. Notably missing from the
list is opt_algebraic which is the single largest source of SSA def
churn and it happily throws names away.
These observations lead me to question whether or not names are actually
useful at all or if they're just taking up space (8B per instruction)
and wasting CPU cycles (to ralloc_strdup on the off chance we do have
one). I don't think I can think of a single time in recent history
where I've been debugging a shader issue and a SSA value name has been
there and been useful. If anything, the few times they are there, they
just throw me off because they mess up the indentation in nir_print.
iris shader-db on my system gets runtime -2.07734% +/- 1.26933% (n=5)
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5439>
The way that the blob obtains the subgroup id on compute shaders is by
just and'ing gl_LocalInvocationIndex with 63, since it advertizes a
subgroupSize of 64. In order to support VK_EXT_subgroup_size_control and
expose a subgroupSize of 128, we'll have to do something a little more
flexible. Sometimes we have to fall back to a subgroup size of 64 due to
various constraints, and in that case we have to fake a subgroup size of
128 while actually using 64 under the hood, by just pretending that the
upper 64 invocations are all disabled. However when computing the
subgroup id we need to use the "real" subgroup size. For this purpose we
plumb through a driver param which exposes the real subgroup size. If
the user forces a particular subgroup size then we lower
load_subgroup_size in nir_lower_subgroups, otherwise we let it through,
and we assume when translating to ir3 that load_subgroup_size means
"give me the *actual* subgroup size that you decided in RA" and give you
the driver param.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6752>
There are some cases using kgsl backend on linux that is still not usual
setup though, we need to consider too.
Regarding the timeline semaphore feature, we could implement it for
the kgsl backend in the future, and probalby it should be using the
existing code in tu_drm.
See #4738, #4907
Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11488>
The compiler *can* eventually chew through all the copy prop, constant
folding, and dead_cf necessary to use just our constant index, but we can
save a whole lot of hassle by chasing the MOVs up front and finding the
constant.
dEQP-VK.ubo.3_level_array.scalar.row_major_mat4.both goes from 2.0s to
1.6s on a release build (3.1s to 2.1s for a debug build like we use in CI).
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11613>
Move and rename warn_non_conformant_implementation() to common location
of src/vulkan/util/vk_util.c as vk_warn_non_conformant_implementation().
In freedreno/ci, move MESA_VK_IGNORE_CONFORMANCE_WARNING to common
location of .baremetal-deqp-test-freedreno-vk.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11563>
Improves drawoverhead perf through Zink up to 260%
Before:
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 1518
After:
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 3981
This brings it close to Freedreno, which has around 4300.
In vkQuake vs params re-emission now occurs in 0.23% of draw calls.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11556>
Hyunjun Ko
Danylo Piliaiev
Connor Abbott
Rob Clark
Matt Turner
Emma Anholt
Alejandro Piñeiro
Eduardo Lima Mitev
Jason Ekstrand
Jonathan Marek
Michel Dänzer