ChanServ changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - Logs https://oftc.irclog.whitequark.org/panfrost - <macc24> i have been here before it was popular
rasterman has quit [Quit: Gettin' stinky!]
camus has quit [Remote host closed the connection]
camus has joined #panfrost
camus1 has joined #panfrost
camus has quit [Read error: Connection reset by peer]
camus has joined #panfrost
camus1 has quit [Ping timeout: 480 seconds]
megi1 has quit [Ping timeout: 480 seconds]
nlhowell has quit [Ping timeout: 480 seconds]
megi1 has joined #panfrost
fd has joined #panfrost
fd has quit []
camus1 has joined #panfrost
camus has quit [Read error: Connection reset by peer]
<tomeu> urja: indeed, kernelci has found 5.15 to be quite unstable on the veyron: https://lava.collabora.co.uk/scheduler/device_type/rk3288-veyron-jaq?dt_search=5.15
camus1 has quit [Remote host closed the connection]
camus has joined #panfrost
<bbrezillon> alyssa,tomeu: if you have no objection, I'd like to land MR 13016
<mmind00> tomeu: with an earlycon you'd probably see it failing in pinctrl as well ... fixes are in the pinctrl and gpio trees for 5.15
<tomeu> bbrezillon: sure, it's basically just waiting for your blit MR to be landed
<tomeu> and alyssa to ack the move
<tomeu> though, why would she be against it? :p
<bbrezillon> tomeu: the blit MR just landed ;)
<tomeu> bbrezillon: rebased already :p
<mmind00> tomeu: correct ... see the lore.kernel.org thread I linked to yesterday
<tomeu> cool, guess the only question that remains is why this wasn't reported from -next...
<mmind00> tomeu: it's very much veyron-specific ... only veyron uses pinctrl-hogs that use the output-enable pinconfig
<tomeu> yeah, but kernelci should be running the tests on it as it does on any other board
<bbrezillon> tomeu: thanks
pjakobsson has quit [Remote host closed the connection]
pjakobsson has joined #panfrost
<alyssa> bbrezillon: wfm
<alyssa> "though, why would she be against it? :p" Yeah that :-p
<tomeu> will take that as an ack :p
<alyssa> sure :)
<alyssa> tomeu: Actually, what's more worrying is that you seem to be under the impression I am the panfrost/ci maintainer 😉
<tomeu> alyssa: of course you are!
camus has quit [Ping timeout: 480 seconds]
camus has joined #panfrost
camus has quit [Remote host closed the connection]
camus has joined #panfrost
<tomeu> bbrezillon: btw, this is almost 1 hour into a full run: https://paste.debian.net/1213408/
camus has quit [Ping timeout: 480 seconds]
<bbrezillon> tomeu: this should fix a few of those crashes => https://gitlab.freedesktop.org/bbrezillon/mesa/-/tree/panvk
camus has joined #panfrost
<tomeu> bbrezillon: I meant it more as a guide on how to work towards the CTS being more useful
<bbrezillon> tomeu: well, having the assert doesn't really help if I don't know the test that triggered it :P
<bbrezillon> (except for those panvk_stub()s)
<tomeu> ah, I can give you that information if you want :)
<bbrezillon> tomeu: that can be useful, yes, I doubt I'll get to it this week though
<bbrezillon> still have a pile of things to upstream in this panvk branch
<bbrezillon> tomeu: BTW, did you start working the secondary command buf implementation based on vk_commands_gen.py?
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
<tomeu> bbrezillon: not yet
nlhowell has joined #panfrost
megi1 has quit [Ping timeout: 480 seconds]
megi1 has joined #panfrost
Danct12 has joined #panfrost
vstehle1 has joined #panfrost
vstehle has quit [Ping timeout: 480 seconds]
rasterman has joined #panfrost
nlhowell is now known as Guest1127
nlhowell has joined #panfrost
Guest1127 has quit [Ping timeout: 480 seconds]
floof58 has quit []
floof58 has joined #panfrost
<urja> .... so, i cherry-picked those two patches (pinctrl and gpio) on top of (now) 5.15-rc3 and yay we have picture... just for the eMMC to fail to initialize
<urja> https://urja.dev/nextcloud/index.php/s/knei2Ayw2gafpCo/preview there's where it stops (to wait for a root device that is never to appear because the eMMC failed to init)
<urja> and all it really has to say is "error -110"
<urja> i havent got the spoons to debug this today.... with my luck it'd be as moot as my sunday :P
Guest6915 has quit [Read error: Connection reset by peer]
<alyssa> urja: you might have more success on #linux-rockchip?
<urja> fair
<urja> where is that even these days? (i think that was one of the channels that i just left in The Move)
<urja> found (on libera)
jernej_ is now known as jernej
Net147 has quit [Ping timeout: 480 seconds]
JulianGro has quit [Remote host closed the connection]
JulianGro has joined #panfrost
Net147 has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
Net147 has quit [Ping timeout: 480 seconds]
Net147 has joined #panfrost
rasterman has quit [Quit: Gettin' stinky!]
nlhowell has joined #panfrost
nlhowell is now known as Guest1138
nlhowell has joined #panfrost
camus1 has quit [Remote host closed the connection]
camus has joined #panfrost
camus1 has joined #panfrost
<tlwoerner> urja: (off-topic) i see the same error (error -110 whilst initialising SD card) on my nanopi-m4 with linux-5.14.6
<tlwoerner> urja: if i remove my uSD card and re-insert it, magically it works
camus has quit [Ping timeout: 480 seconds]
<urja> can't exactly remove the eMMC and re-insert :P
<tlwoerner> lol
<urja> (also, -110 is just "ETIMEDOUT" which can mask pretty much any kind of "it doesnt answer in this configuration")
tchebb has quit [Ping timeout: 480 seconds]
<alyssa> tomeu: If your intention with the piglit/ci MR is to dangle a list of fails on G52 in front of me and nerdsnipe me into fixing them, mission accomplished.
<alyssa> my board is online and running piglit and I am digging into these fails :_p
camus1 has quit [Remote host closed the connection]
camus has joined #panfrost
<alyssa> looking for low hanging bugs
WoC has quit [Remote host closed the connection]
WoC has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
nlhowell has joined #panfrost
<cphealy> IIRC, there was a significant performance regression between Mesa 21.1 and 21.2 with the glmark2-es2 "buffer" benchmarks on Mali G31 where one of them dropped from ~51 fps to ~3 fps. Perhaps this is the low hanging fruit you are looking for? ;-)
nlhowell has quit [Ping timeout: 480 seconds]
tomeu has quit [Read error: Connection reset by peer]
italove has quit [Read error: Connection reset by peer]
ndufresne has quit [Write error: connection closed]
fahien2 has quit [Remote host closed the connection]
megi1 has quit [Ping timeout: 480 seconds]
tomeu has joined #panfrost
ndufresne has joined #panfrost
italove has joined #panfrost
fahien2 has joined #panfrost
ndufresne is now known as Guest1159
<alyssa> cphealy: what wait uh
italove8 has joined #panfrost
ndufresne5 has joined #panfrost
<alyssa> cphealy: buffer is indeed slow. what.
camus1 has joined #panfrost
Guest1159 is now known as ndufresne
<alyssa> shaders seem ok
<cphealy> alyssa: does this mean you reproduced it?
camus has quit [Ping timeout: 480 seconds]
<alyssa> cphealy: not the regression but yes the G52 being 3x slower than T860
ndufresne5 has quit []
italove8 has quit []
<cphealy> What I saw was that of the three buffer benchmarks, the first one regressed from 50->33, the second one regressed from 51->3 and the third one regressed from 58->35. DDK was 56, 58, and 60 for comparison.
<alyssa> it's not clear how this can be
<alyssa> names please?
<cphealy> benchmark names you mean?
<alyssa> oh I see
<cphealy> [buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map
<cphealy> [buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata
<cphealy> [buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map
<cphealy> first, second, third
megi1 has joined #panfrost
<alyssa> okay I do see pathologically bad perf on the second
<alyssa> but completely CPU bound
<alyssa> that's a job for sysprof
<alyssa> thanks for the report :]
<cphealy> N/P
<alyssa> ugh my internet stiuation is no bueno
<alyssa> cphealy: This is baffling
<alyssa> Pretty much entirely bound by the memcpy in u_default_buffer_subdata
<alyssa> I don't see a good reason for behaviour to diverge between the arches, at all.
cyrozap has quit [Remote host closed the connection]
<alyssa> will chew on this.
<alyssa> and certianly no good reason for a regression
<anholt_> pfdev->coherent different between the two?
<alyssa> tomeu: fixed .... one piglit!
<alyssa> oh boy
<alyssa> quality nerdsnipe 10/10
<alyssa> (crash -> pass)
<alyssa> shaders@glsl-fs-pointcoord is upside down. Oh fun.
<cphealy> alyssa: so I just need a faster CPU, huh? ;-)
<alyssa> cphealy: I mean that's why I don't notice as much here
<cphealy> so, you need a slower CPU then... :-)
<alyssa> oh one bug masking another, gosh
<alyssa> conformance was just the start
karolherbst has quit [Quit: Konversation terminated!]
cyrozap has joined #panfrost
Lyude has quit [Ping timeout: 480 seconds]
karolherbst has joined #panfrost
JulianGro has quit [Ping timeout: 480 seconds]
JulianGro has joined #panfrost
jernej_ has joined #panfrost
jernej has quit [Read error: Connection reset by peer]
jernej has joined #panfrost
<alyssa> glsl-vs-if-bool
<alyssa> this is the sort of 1-bit boolean trouble I've been scared of. Oooof.
<alyssa> I sketched in my head a better way to deal with this over the summer. But... ugh. It's too late for me to rewrite booleans again.
<alyssa> (in the night. Maybe on like. Wednesday.)
jernej_ has quit [Ping timeout: 480 seconds]