Many tests had been fixed but weren't being run due to test reshuffles
from uprevs. Add some explanations for what remains.
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15133>
On Bifrost and Valhall, push uniforms are loaded into Fast Access Uniform
Random Access Memory (FAU-RAM). FAU-RAM is organized as an array of 64-bit
slots. A given tuple (Bifrost) or instruction (Valhall) may access at most a
single 64-bit slot. If an instruction requires uniforms from multiple 64-bit
slots, a uniform-to-register move must be inserted to avoid the hazard. However,
if an instruction requires a pair of 32-bit uniforms from the same 64-bit slot,
no move is required.
To reduce the number of moves we emit, this commit adds an optimization pass
that reorders pushed uniforms, trying to group uniforms used by the same
instruction. The pass works by creating a graph of pushed uniforms, where edges
denote the "both 32-bit uniforms required by the same instruction" relationship.
We perform depth-first search on this graph to find the connected components,
where each connected component is a cluster of uniforms that are used together.
We then select pairs of uniforms from each connected component. The remaining
unpaired uniforms (from components of odd sizes) are paired together
arbitrarily.
In principle, we should weight the graph by number of occurences and choose
pairs that maximize the total selected edge weight. This is left for
future work, as it is nontrivial -- selecting these edges optimally appears to
be NP-hard at first blush.
Implementation note: As position and varying shaders share FAU on Bifrost, extra
care is taken with a `push_offset` shader stage info parameter that ensures
varying shaders do not reorder uniforms selected by the previous position
shader.
total instructions in shared programs: 2503343 -> 2451758 (-2.06%)
instructions in affected programs: 1553309 -> 1501724 (-3.32%)
helped: 14256
HURT: 8
helped stats (abs) min: 1.0 max: 80.0 x̄: 3.62 x̃: 3
helped stats (rel) min: 0.06% max: 36.36% x̄: 7.31% x̃: 6.67%
HURT stats (abs) min: 1.0 max: 2.0 x̄: 1.38 x̃: 1
HURT stats (rel) min: 1.30% max: 12.50% x̄: 4.99% x̃: 3.85%
95% mean confidence interval for instructions value: -3.66 -3.58
95% mean confidence interval for instructions %-change: -7.41% -7.20%
Instructions are helped.
total tuples in shared programs: 2008399 -> 1969627 (-1.93%)
tuples in affected programs: 1146344 -> 1107572 (-3.38%)
helped: 12867
HURT: 147
helped stats (abs) min: 1.0 max: 61.0 x̄: 3.03 x̃: 2
helped stats (rel) min: 0.17% max: 42.86% x̄: 6.79% x̃: 4.65%
HURT stats (abs) min: 1.0 max: 3.0 x̄: 1.20 x̃: 1
HURT stats (rel) min: 0.29% max: 20.00% x̄: 2.12% x̃: 1.19%
95% mean confidence interval for tuples value: -3.03 -2.93
95% mean confidence interval for tuples %-change: -6.82% -6.57%
Tuples are helped.
total clauses in shared programs: 408005 -> 401708 (-1.54%)
clauses in affected programs: 90760 -> 84463 (-6.94%)
helped: 6006
HURT: 164
helped stats (abs) min: 1.0 max: 9.0 x̄: 1.08 x̃: 1
helped stats (rel) min: 0.45% max: 33.33% x̄: 12.44% x̃: 14.29%
HURT stats (abs) min: 1.0 max: 1.0 x̄: 1.00 x̃: 1
HURT stats (rel) min: 1.64% max: 25.00% x̄: 9.81% x̃: 5.26%
95% mean confidence interval for clauses value: -1.03 -1.01
95% mean confidence interval for clauses %-change: -12.03% -11.66%
Clauses are helped.
total cycles in shared programs: 203308.37 -> 202737.83 (-0.28%)
cycles in affected programs: 19264.71 -> 18694.17 (-2.96%)
helped: 3024
HURT: 41
helped stats (abs) min: 0.041665999999999315 max: 2.5416680000000014 x̄: 0.19 x̃: 0
helped stats (rel) min: 0.17% max: 33.33% x̄: 3.83% x̃: 2.83%
HURT stats (abs) min: 0.041665999999999315 max: 0.125 x̄: 0.06 x̃: 0
HURT stats (rel) min: 0.30% max: 5.88% x̄: 1.41% x̃: 0.93%
95% mean confidence interval for cycles value: -0.19 -0.18
95% mean confidence interval for cycles %-change: -3.89% -3.64%
Cycles are helped.
total arith in shared programs: 76265.67 -> 74669.25 (-2.09%)
arith in affected programs: 45001.50 -> 43405.08 (-3.55%)
helped: 12945
HURT: 97
helped stats (abs) min: 0.041665999999999315 max: 2.5416680000000014 x̄: 0.12 x̃: 0
helped stats (rel) min: 0.17% max: 50.00% x̄: 8.06% x̃: 4.88%
HURT stats (abs) min: 0.041665999999999315 max: 0.125 x̄: 0.05 x̃: 0
HURT stats (rel) min: 0.21% max: 33.33% x̄: 2.16% x̃: 0.96%
95% mean confidence interval for arith value: -0.12 -0.12
95% mean confidence interval for arith %-change: -8.16% -7.81%
Arith are helped.
total quadwords in shared programs: 1796563 -> 1766803 (-1.66%)
quadwords in affected programs: 948830 -> 919070 (-3.14%)
helped: 12078
HURT: 219
helped stats (abs) min: 1.0 max: 42.0 x̄: 2.49 x̃: 2
helped stats (rel) min: 0.10% max: 33.33% x̄: 5.57% x̃: 5.26%
HURT stats (abs) min: 1.0 max: 4.0 x̄: 1.21 x̃: 1
HURT stats (rel) min: 0.33% max: 6.67% x̄: 2.00% x̃: 1.14%
95% mean confidence interval for quadwords value: -2.46 -2.38
95% mean confidence interval for quadwords %-change: -5.52% -5.36%
Quadwords are helped.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14163>
Gives us memory back faster which is useful for pathalogical CTS
tests.
The GLSL IR was previously used after converting to NIR for things
like building the GL resource list but we have had a NIR version
for this for some time and I don't believe there are any other
use cases left for keeping the old IR hanging around this long.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15127>
These runners are configured to have a single job take up the whole
runner, which means we get to use threads to our hearts content. The pile
of cores means we don't need to spawn separate jobs to try to load-balance
across fdo's shared runner capacity. Having dedicated runners means we
won't get our MRs blocked as much waiting on non-Mesa testing happening on
fd.o.
We manage to complete all of this llvmpipe testing in about 6:15.
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14962>
This was only used for meta operations. DCC/FMASK/FCE pipelines
only declare one color attachment and the color writemask of the
second color attachment is 0 for the HW CB resolve.
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/14650>
5nm PHY is a variation of 7nm PHY, they use the same register
definitions. To remove duplication, drop 5nm defs.
Cc: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15051>
This check for geom shader needed to be expanded for tess support.
dEQP-VK.transform_feedback.simple.depth_clip_control_tese with lvp
Fixes: dacf8f5f5c ("draw: hook up final bits of tessellation")
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15128>
There are up to 4 instructions in the latter stage (if a branch is included),
not 3. Bump the limit to fix memory corruption.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reported-by: Icecream95 <ixn@disroot.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15147>
This callback is only intended for software rasterizers, layered drivers, and
other special drivers that go through the software winsys path. Remove the
unimplemented stubs from the Intel drivers.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Dave Airlie <airlied@redhat.com> [crocus]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15118>
Unify the exit paths to clean up the logic. There are logically three modes we
support (KMS without renderonly, KMS with renderonly, and FD); these each
correspond to a leg of a small if statement. Outside of the small if's,
everything else should be identical.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: James Jones <jajones@nvidia.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15120>
When handle->type is WINSYS_HANDLE_TYPE_FD, the caller wants a file descriptor
for the BO backing the resource. We previously had two paths for this:
1. If rsrc->scanout is available, we prime the GEM handle from the KMS device
(rsrc->scanout->handle) to a file descriptor via the KMS device.
2. If rsrc->scanout is not available, we prime the GEM handle from the GPU
(bo->gem_handle) to a file descriptor via the GPU device.
In both cases, the caller passes in a resource (with BO) and expects out a file
descriptor. There are no direct GEM handles in the function signature; the
caller doesn't care which GEM handle we prime to get the file descriptor. In
principle, both paths produce the same file descriptor for the same BO, since
both GEM handles represent the same underlying resource (viewed from different
devices).
On grounds of redundancy alone, it makes sense to remove the rsrc->scanout path.
Why have a path that only works sometimes, when we have another path that works
always?
In fact, the issues with the rsrc->scanout path are deeper. rsrc->scanout is
populated by renderonly_create_gpu_import_for_resource, which does the
following:
1. Get a file descriptor for the resource by resource_get_handle with
WINSYS_HANDLE_TYPE_FD
2. Prime the file descriptor to a GEM handle via the KMS device.
Here comes strike number 2: in order to get a file descriptor via the KMS
device, we had to /already/ get a file descriptor via the GPU device. If we go
down the KMS device path, we effectively round trip:
GPU handle -> fd -> KMS handle -> fd
There is no good reason to do this; if everything works, the fd is the same in
each case. If everything works. If.
The lifetimes of the GPU handle and the KMS handle are not necessarily bound. In
principle, a resource can be created with scanout (constructing a KMS handle).
Then the KMS view can be destroyed (invalidating the GEM handle for the KMS
device), even though the underlying resource is still valid. Notice the GPU
handle is still valid; its lifetime is tied to the resource itself. Then a
caller can ask for the FD for the resource; as the resource is still valid, this
is sensible. Under the scanout path, we try to get the FD by priming the GEM
handle on the KMS device... but that GEM handle is no longer valid, causing the
PRIME ioctl to fail with ENOENT. On the other hand, if we primed the GPU GEM
handle, everything works as expected.
These edge cases are not theoretical; recent versions of Xwayland trigger this
ENOENT, causing issue #5758 on all Panfrost devices. As far as I can tell, no
other kmsro driver has this 'special' kmsro path; the only part of
resource_get_handle that needs special handling for kmsro is getting a KMS
handle.
Let's remove the broken, useless path, fix Xwayland, bring us in line with other
drivers, and delete some code.
Thank you for coming to my ted talk.
Closes: #5758
Fixes: 7da251fc72 ("panfrost: Check in sources for command stream")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reported-and-tested-by: Jan Palus <jpalus@fastmail.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: James Jones <jajones@nvidia.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Tested-by: Dan Johansen <strit@manjaro.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15120>
We always blit from a particular level, so it's a waste to compute the LOD.
This corresponds to a simple texture instruction with implement 0 LOD, which is
the optimal texturing path on Bifrost -- it maps to TEXS_2D but does not require
helper invocations.
Functional change on Bifrost: Blit shaders no longer set .computed_lod or
shader_contains_barrier.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15123>
There are always set to true. Don't pollute the driver code with them, make
their existence a local detail to pre-Valhall XML and that's it.
Functional change: "four components per vertex" is now set on vertex job DCDs.
This should be a no-op.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15123>
This reduces code duplication and will ease Valhall porting. Functional changes
on v7:
* Shader contains barrier is now set (perf loss, fixed later in series)
* Shader register allocation is now set (perf win)
* Point sprite inverted, no-op for blit shaders
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15123>
cont -> skip, last -> kill, and fix the special case handling. It's just an
enum. Makes the disassembly easier to read and closer to Bifrost.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15123>
For ordinary feedback loops GMEM is a lot faster than sysmem since
we don't set SINGLE_PRIM mode.
For feedback loops with ordered rasterization GMEM should also be
faster.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15106>
Trivially implemented by using A6XX_GRAS_SC_CNTL_SINGLE_PRIM_MODE.
This extension is useful for emulators e.g. AetherSX2 PS2 emulator and
could drastically improve performance when blending is emulated.
Relevant tests:
dEQP-VK.rasterization.rasterization_order_attachment_access.*
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15106>
Otherwise a shader invocation would read the value which should have
been set AFTER this shader invocation.
Fixes tests:
dEQP-VK.rasterization.rasterization_order_attachment_access.depth.samples_1.multi_draw_barriers
dEQP-VK.rasterization.rasterization_order_attachment_access.stencil.samples_1.multi_draw_barriers
Fixes: 71595a189a
("tu: Fix feedback loops in sysmem mode")
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15106>
XeHP scratch space is handled differently. Commit ae18e1e707
implemented support for it, but handled it differently between render
and compute shaders: it calculates scratch_addr differently and
doesn't pin the buffer on compute. Make it work on compute shaders by
calling pin_scratch_space() from iris_compute_walker(), which fixes
both the address and the pinning.
This commit can be verified by the two-year-old-but-still-unreviewed
Piglit MR 234. You can also verify this by running a very simple
compute shader with INTEL_DEBUG=spill_fs.
References: https://gitlab.freedesktop.org/mesa/piglit/-/merge_requests/234
Fixes: ae18e1e707 ("iris: Add support for scratch on XeHP")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15070>
This raises our vertex input bindings limit from 28 to 31, and our
vertex input attribute limit from 28 to 29. We could theoretically
go higher, but it will take additional work.
The 3DSTATE_VERTEX_BUFFERS and 3DSTATE_VERTEX_ELEMENTS limits are 33
vertex buffers, and 34 vertex elements. But we need up to two vertex
elements for system values (FirstVertex, BaseVertex, BaseInstance,
DrawID), and we currently use two vertex bindings for those.
There is another hidden limit: our compiler backend only supports the
push model for VS inputs currently. 3DSTATE_VS only allows URB Read
Lengths between [0, 15], which is measured in pairs of inputs, which
means we can theoretically push no more than 32 vertex elements. This
is no artifical limit either, as a vec4 element takes up 4 registers
in the payload, and 32 * 4 = 128, the entire size of our register file.
Plus, the VS Thread payload needs at least g0 and g1 for other things,
so we can really only push 31.
We can theoretically support one additional binding, by combining our
two SGV bindings into a single upload. In order to support additional
vertex elements, we would need to add support to the backend compiler
for the pull model for VS inputs.
References: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5917
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14991>
I got confused while writing this somehow because of the null descriptor
feature, which enables drivers to consume a null descriptor, which has no
relation to a descriptor layout containing no descriptors
failing to accurately use zero descriptors can put layouts over the maximum
per-stage limits, which causes tests to crash
fixes (lavapipe):
KHR-GL46.shading_language_420pack.binding_uniform_block_array
KHR-GL46.multi_bind.dispatch_bind_buffers_base
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15067>
I chose to implement this as a global flag in the device, because
otherwise we would end up with extra draw overhead trying to avoid it in
the implicit-sync WSI case, and you're probably going to end up needing
implicit sync anyway because you used one of the BOs in any of the
submitted cmdbufs. To do better than this, we would probably want a
skip-implicit-sync flag on the BOs in the BO list, rather than global on
the submit.
Reports about venus on turnip say that this flag reduces worst-case
QueueSubmit time in a game workload from ~10ms to ~4ms.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14838>
Extend our existing bi_scoreboard infrastructure with a simple data flow
analysis pass that calculates which dependency slots need waiting. We
still lack a heuristic for selecting dependency slots.
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14298>