Right now if the shader indirects on some large constant array, we see NIR
load_consts (usually from the const file) of its contents into general
registers, then indirection on the GPRs. This often results in register
allocation failures, as it's easy to go beyond the ~256 dwords of
registers per invocation.
By moving the large constants to a UBO, we can load an arbitrary number of
them. They also can be theoretically moved to the constant reg file (~2k
dwords), though you're unlikely to hit this path without an indirect load
on your large constant, and we don't yet let UBO indirect loads get moved
to constant regs.
This possibly won't work out right if we have 16-bit load_constants, but
without other MRs in flight we won't see 16-bit temps to be lowered to
this.
This allows 2 kerbal-space-program shaders to compile that previously
would fail, and fixes the new dEQP-VK and -GLES2 tests I wrote that
dynamically index a 40-element temporary array of float/vec2/vec3/vec4
with constant element initializers.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/2789
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5810>
I want the new version to show the fix in the fd-largeconsts branch (and
make sure the pass keeps working, and make sure other drivers get around
to fixing the issue). While I'm here, cherry-pick in the VK test along
with the GLES one, and also the fix for clip_three on ARMs.
Since the VK and GL test lists were changing, I took the opportunity to
reset freedreno xfails lists to just the tests that are being run with the
CTS uprev, and increase its coverage to 1/10th of the CTS across two
boards (since we just freed up a bunch of runtime with the grouped gles
"other" job).
For panfrost, I didn't spend the time characterizing the t720 fragment_ops
flakes like I did for the deqp-runner change. Given that the random
behavior changes between CTS versions, it doesn't seem to be worth the
time to do so.
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6971>
I found the C++ runner hard to develop on, and we had stability issues and
outstanding feature needs that made me want something I felt good about
hacking on. Thus, Rewrite It In Rust of the deqp runner.
The new runner includes:
- Skip lists don't reshuffle the test list.
- Known-flake handling without resorting to skip lists (fixing our main CI
reliability issue on a3xx right now).
- Per-thread Vulkan shader caches should speed up VK CI runtime.
- Tracking of crashes separate from fails (so we can see progress on that
front).
- Logging of deqp stderr spam (particularly assertion failures!) in the CI
log.
- Integrated QPA filtering so we don't have bash perf issues for it.
- Logging of what caselist to go look at for a given error report (in red,
so it's easier to find in your CI log).
- The code is 1/3 unit tests, and easy to extend for more coverage.
- Non-LAVA CI runs create a failures.csv in artifacts that you can check
in as your deqp-*-fails.txt file.
- Test runtime is included in results.csv so you can debug how to speed up
your CI job.
- Pretty summary at the end of the run of slow/flaky/failed tests.
Since this is a new runner with a different RNG, the test groups are
shuffled one more time. This seems to result in some panfrost T720
stability issues (See its new deqp-panfrost-t720-flakes.txt), and one new
flake in freedreno a630.
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7434>
This allows overlapping inputs, which is required for the next patch which
makes it so setup_input may be called multiple times for each input.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6181>
Given that a530 doesn't have cpufreq, we really don't have the time to be
running the validator on all of deqp. This also helps explain why I had
to go to such a small fraction on the a3xx gles3 run (which we can now
increase). However, a3xx gles2 seems to be fast enough that we can leave
it enabled and get coverage for older chips.
Because we run more tests now, clear out some stale xfails from the a3xx
list.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5089>
A3xx apparently has higher alignment requirements than later gens for
indirect const uploads. It also has fewer of them. Add compiler
parameters for both settings, and set accordingly for a3xx and a4xx+.
This fixes all the ubo test failures caused by this optimization.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5077>
This causes failures on a3xx resulting in the non-sensical dEQP failures
on packUnorm2x16. The same test uses ldlv on a4xx+, so just disallow
(sat) on bary.f on all generations.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5074>
Rendering doesn't work, but having the format in place avoids an assert
when selecting the texture format in st_format. I believe it's required
for GLES3, so more tracing is required to determine what bit we're
missing to make rendering work.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5073>
This fixes a number of dEQP tests:
dEQP-GLES3.functional.fbo.blit.conversion.r8*
dEQP-GLES3.texture.specification.basic_teximage2d.r8*
and others. The reason why this enum showed up in traces for R8 is that
it was an "upgraded" texture to R8G8.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5073>
We were failing to tell the allocator about the restriction that scalar
texture instructions (allocated as scalar regs) couldn't be allocated such
that the start of the full unwritemasked vector started before r0. There
was a patch in select_reg_callback on a6xx that tried to work around that,
but you could still end up backed into a corner you shouldn't be because
we didn't tell the RA what it needed.
Fixes compiler assertion failures on a300-a400's blit_z shader, used for
Z32F gmem blits.
Looks like as a result we get tighter register allocation but more nops:
instructions in affected programs: 757945 -> 760356 (0.32%)
nops in affected programs: 317983 -> 320468 (0.78%)
non-nops in affected programs: 27525 -> 27451 (-0.27%)
mov in affected programs: 3098 -> 3023 (-2.42%)
dwords in affected programs: 109664 -> 110656 (0.90%)
last-baryf in affected programs: 112701 -> 112847 (0.13%)
full in affected programs: 4326 -> 4011 (-7.28%)
sstall in affected programs: 120550 -> 120836 (0.24%)
(ss) in affected programs: 13939 -> 13918 (-0.15%)
(sy) in affected programs: 3006 -> 2786 (-7.32%)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4562>
We haven't had it enabled due tointermittent failures. Those failures
are, as far as I can tell, due to GPU faults from buffer overflows where a
failing test in a thread stomps an otherwise passing thread's buffers. By
running deqp single-threaded, we can get more consistent failures, at the
cost of needing to do a tiny subset of the tests to keep runtime down.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4685>
Now that we have 7 (soon 8) boards available, there's capacity to be
testing GLES 3.0. However, due to (it looks like) buffer overflows in the
driver, we end up with flaky test results: 1/60 jobs spuriously failed,
and another 6/60 jobs reported flakes. At 6 jobs per pipeline, that's way
too high of a failure rate to enable for non-freedreno developers. Leave
the job present but disabled so that we can do manual test runs for
regressions.
Reviewed-by: Daniel Stone <daniels@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3661>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3661>
This is based on the fix I used for the same problem on V3D. In this
case, it fixes all but the the
dEQP-GLES2.functional.texture.filtering.2d.*_npot cases of
dEQP-GLES2.functional.texture.filtering.2d.*'s failures.
Acked-by: Rob Clark <robdclark@chromium.org>
Since freedreno's kernel and GPU reset seem to be totally solid, we
don't need to have the complexity of the LAVA setup that panfrost has.
Instead, we can register some boards as shared gitlab runners and have
the jobs run out of a docker container just like we do for llvmpipe.
Just make sure that the DRI device node is passed through to the
containers in the gitlab config ('devices = ["/dev/dri"]' under
runners.docker).
If a runner fails (networking dies, kernel panic, etc.) it'll take out
one build but the rest can keep going since gitlab-runner is what
pulls jobs. Since the runner pulls jobs, it also means that they can
live behind firewalls instead of needing some public address to be
accessed by gitlab.fd.o.
For now, enable it just on db410c (A307) and cheza (A630) as those are
the hardware that I have plenty of. A307 is only testing GLES2 since
running all of GLES3 takes too long for the number of boards I've
brought up.
Acked-by: Rob Clark <robdclark@chromium.org>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Juan A. Suarez Romero
Eric Anholt
Rob Clark
Jonathan Marek
Ilia Mirkin