#
✕ nick:alyssa
#asahi-gpu · 2023-06-30 19:50
alyssaella-0: so.. now that we're over the Big GLES Hurdle, it's time for me to start switching gears towards VK
#asahi-gpu · 2023-06-30 00:19
alyssaI think Apple has an optimization where they're only blending one sample instead of all 4 if all the samples have the same value, or something like that. IDK. not helpful here.
#asahi-gpu · 2023-06-30 00:17
alyssathis seems to materialize a sample mask in r1l.. somehow
#asahi-gpu · 2023-06-30 00:16
alyssa> 1c: 8000 <disassembly failed>
#asahi-gpu · 2023-06-30 00:16
alyssa> 16: 59050000103c threadgroup_atomic add, None, r0l, r0l, r0, 1, 2, 60, 0
#asahi-gpu · 2023-06-30 00:16
alyssaalso it's doing... who knows
#asahi-gpu · 2023-06-30 00:16
alyssaexcept the 4800c200 wait_pix 512, 3 crap which maybe I need to stop ignoring.
#asahi-gpu · 2023-06-30 00:16
alyssalooking at traces in the macos side for inspiration.. nothing striking
#asahi-gpu · 2023-06-29 23:54
alyssarepro.
#asahi-gpu · 2023-06-29 23:53
alyssaI do find this flake concerningly difficult to repo.
#asahi-gpu · 2023-06-29 23:51
alyssaalthough seemingly the fail might've been carried over
#asahi-gpu · 2023-06-29 23:50
alyssaignore me.
#asahi-gpu · 2023-06-29 23:50
alyssa---Fuck, never mind, just got the fail
#asahi-gpu · 2023-06-29 23:50
alyssaWith that, I can't seem to repro the Fail, although I can still hit timeouts
#asahi-gpu · 2023-06-29 23:50
alyssajannau: I've just pushed 2c0d4fc71fb ("DONOTMERGE: agx: Don't emit wait_pix multiple times")
#asahi-gpu · 2023-06-29 18:55
alyssaoh that's probably the thing Lina just fixed
#asahi-gpu · 2023-06-29 18:55
alyssaone bug at a time. beh
#asahi-gpu · 2023-06-29 18:55
alyssaruh roh why am I seeing faults trying to open keyboard shortcuts in gnome
#asahi-gpu · 2023-06-29 18:07
alyssaI just meant running the khr tests at all (with a standard config)
#asahi-gpu · 2023-06-29 18:07
alyssaYeah, I don't think that's worth it to replicate with deqp-runner
#asahi-gpu · 2023-06-29 18:07
alyssaoh, yeah
#asahi-gpu · 2023-06-29 18:07
alyssa17:43 jannau | they are repeated with different --deqp-gl-config-id options
#asahi-gpu · 2023-06-29 18:07
alyssaah
#asahi-gpu · 2023-06-29 17:42
alyssano?
#asahi-gpu · 2023-06-29 17:42
alyssafwiw there are a lot more deqps than khr-gles's, so it's not adding much wall clock time to run the KHR tests too
#asahi-gpu · 2023-06-29 17:41
alyssashould probably do that
#asahi-gpu · 2023-06-29 17:41
alyssaso if you want you can `cat` together the different mustpass files for a big combined run
#asahi-gpu · 2023-06-29 17:41
alyssathe glcts binary contains all of them
#asahi-gpu · 2023-06-29 17:41
alyssaOh. Yeah, it is
#asahi-gpu · 2023-06-29 17:40
alyssaeric_engestrom: ^^ any ideas?
#asahi-gpu · 2023-06-29 17:40
alyssajannau: to confirm that's with x11-egl?
#asahi-gpu · 2023-06-29 17:39
alyssaI wonder what that's about
#asahi-gpu · 2023-06-29 17:39
alyssajannau: and woo
#asahi-gpu · 2023-06-29 17:38
alyssathe CTS run contains both
#asahi-gpu · 2023-06-29 17:38
alyssathe aosp_mustpass is for dEQP tests
#asahi-gpu · 2023-06-29 17:38
alyssajannau: IIRC the khronos_mustpass is for KHR-GLES* tests
#asahi-gpu · 2023-06-29 16:59
alyssaor like lucky 1 tbf
#asahi-gpu · 2023-06-29 16:59
alyssatoday's lucky 10,000
#asahi-gpu · 2023-06-29 16:57
alyssaanholt's cts runner is a lot newer than piglit tbf
#asahi-gpu · 2023-06-29 16:57
alyssafair
#asahi-gpu · 2023-06-29 16:51
alyssait'd be great if we passed stress too but it's not part of the CTS currently
#asahi-gpu · 2023-06-29 16:51
alyssaand the stress tests are also not included in mustpass, only the functional ones contribute to conformance
#asahi-gpu · 2023-06-29 16:50
alyssathe image atomic comp swap return value tests are known broken and not included in the mustpass list, piglit is only running them because piglit is wrong
#asahi-gpu · 2023-06-29 16:50
alyssaand run-deqp31 would indicate that as passing
#asahi-gpu · 2023-06-29 16:50
alyssathat is passing to me
#asahi-gpu · 2023-06-29 16:50
alyssajannau: as for the deqp-gles31 results--
#asahi-gpu · 2023-06-29 16:49
alyssabloop
#asahi-gpu · 2023-06-29 16:48
alyssa--deqp-log-images=disable --deqp-log-shader-sources=disable
#asahi-gpu · 2023-06-29 16:48
alyssaMESA_SHADER_CACHE_DISABLE=1 EGL_PLATFORM=surfaceless deqp-runner run --deqp ~/GLCTS/glcts --output output --jobs 8 --renderer-check G13 --caselist ~/GLCTS/gl_cts/data/mustpass/gles/aosp_mustpass/main/gles31-master.txt --skips ~/mesa/.gitlab-ci/all-skips.txt -- --deqp-surface-type=pbuffer --deqp-gl-config-name=rgba8888d24s8ms0 --deqp-surface-width=256 --deqp-surface-height=256
#asahi-gpu · 2023-06-29 16:48
alyssaI have `run-deqp31` aliased to

plus ›