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
vstehle has quit [Ping timeout: 480 seconds]
alpernebbi has quit [Ping timeout: 480 seconds]
WoC has joined #panfrost
alpernebbi has joined #panfrost
WoC has quit [Remote host closed the connection]
WoC has joined #panfrost
hyrc_ has joined #panfrost
hyrc has quit [Ping timeout: 480 seconds]
vstehle has joined #panfrost
camus1 has joined #panfrost
camus has quit [Read error: Connection reset by peer]
Rathann|Mobile has joined #panfrost
MajorBiscuit has joined #panfrost
camus has joined #panfrost
camus1 has quit [Read error: Connection reset by peer]
atler is now known as Guest295
atler has joined #panfrost
Guest295 has quit [Ping timeout: 480 seconds]
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
camus1 has quit [Remote host closed the connection]
camus has joined #panfrost
camus has quit [Remote host closed the connection]
camus has joined #panfrost
rasterman has joined #panfrost
floof58 has joined #panfrost
floof58_ has quit [Ping timeout: 480 seconds]
Rathann|Mobile has quit [Quit: Leaving]
CME has quit [Remote host closed the connection]
CME has joined #panfrost
pjakobsson has quit [Remote host closed the connection]
pjakobsson has joined #panfrost
vstehle has quit [Quit: WeeChat 3.3]
nlhowell has joined #panfrost
vstehle has joined #panfrost
camus1 has joined #panfrost
camus has quit [Remote host closed the connection]
<rasterman>
alyssa: boo
<alyssa>
rasterman: boo
<rasterman>
(and i guess this goes for daniels too and bbrezillon)
<rasterman>
i have 3 patches for you for panfrost/mesa brewing
<rasterman>
just a heads up
<alyssa>
Tasty
<rasterman>
1. add new dpu to gallium
<rasterman>
(kmsro)
<rasterman>
2. fix your failure cause for gpuid in panfrost (ie if gpuid not known) .. it'd be nice to not segv...
<rasterman>
3. add a new gpuid :)
<alyssa>
for #3, the policy is just "if it works, I'll merge"
<alyssa>
hell I have an "add Mali-G76" patch here somewhere but nobody could confirm it worked
<rasterman>
i was actually mulling ... perhaps panfrost should have some env vars to be able to specifically fa a gpuid in an env var...
<alyssa>
PAN_GPU_ID?
<alyssa>
robmur01: If you want to have a field day with dEQP and an FPGA, we can get All the Malis supported :-p
<rasterman>
mainly because in future... it'd be nice for regular users to easily confirm "hey this gpuid works just the same as this other one"
<rasterman>
oh its there?
<rasterman>
oh and u have the g76 one?
<alyssa>
I have unmerged patches for everything :-p
<rasterman>
0x7211 right?
<alyssa>
uhhh 0x7001
<alyssa>
maybe there are multiple IDs for G76
<rasterman>
oh...
<rasterman>
i cant help you with that
<alyssa>
I'm still not 100% on how Bifrost/Valhall GPU IDs are assigned tbh
<rasterman>
well in theory i could with my fpga and many many many hours of hutning down bit files and getting them to work at all etc...
<rasterman>
anyway
<daniels>
ugh, renderonly
<rasterman>
i have a patch that is tested and works...
<robmur01>
I don't have G76 or anything newer, since it had all moved to a new FPGA system by then :(
<rasterman>
like works in that i'v erun gnome shell, enlightenment, xorg with compositing, glmark, even 0ad (strategy game) works...
<rasterman>
have been beating my head mostly on non-panfrost issues :)
<rasterman>
alyssa: also given the number of places to go add gpu id's .. was mulling if panfrost needed some better "gpu" struct where you just had all of these things in a central struct so you add them in one place... :)
<rasterman>
anyway... was a minor quibble. i'll get you some patches to look at - but know that i actually tested these and they run/work on some new shiny silicon
vstehle has quit [Quit: WeeChat 3.3]
vstehle has joined #panfrost
WoC has quit [Ping timeout: 480 seconds]
<alyssa>
rasterman: Yeah, adding a GPUs struct has been a wishlist item of mine it's just so trivial that I've never gotten around to it
<rasterman>
i got a nice sequence there... awesome
<rasterman>
daniels: :) tnx :)
<daniels>
heh np
<rasterman>
not like you're going ot find too many other peolpe to verify and test it :)
<rasterman>
you'll have to take my word for it that it works ... i guess 0ad running in x11 with compositing should be enough
<rasterman>
i first tested gnome shell.. but man was it... hmmm .. jerky. moving windows around was amazingly not smooth.
<rasterman>
it is because gnome doesnt vsync or something?
<rasterman>
it's odd just stuffing e on the box made it smooth... :( i expected gnome shell and e to be ballpark similar
<rasterman>
e even does nasrty things like dithers the gl rendering which adds a measurable overhead
<rasterman>
(it helps the win shadows look ultra-smood though)
<daniels>
honestly, who knows with x11 ...
<daniels>
depends on which DDX you're using, as some of them love to give you a default visual which means you end up doing too many copies
<macc24>
"some of them love to give you a default visual" O.o? the hell
<rasterman>
¯\_(ツ)_/¯
WoC has joined #panfrost
<rasterman>
dunno what is really up anyway... maybe e/efl get lucky in their choices :)
WoC has quit [Remote host closed the connection]
WoC has joined #panfrost
<rasterman>
i know that i put in explicit visal/cmap choosing code to choose what exactl matches - no zbuffer or stencils etc. and the right rgb bits
WoC has quit [Quit: Leaving]
WoC has joined #panfrost
tlwoerner has quit [Quit: Leaving]
<alyssa>
daniels: gnome-shell has been jankier for me in wayland than x11 (on both panfrost and llvmpipe)
<alyssa>
I do not know what mutter botched for that to happen
<macc24>
alyssa: jankier?
<alyssa>
lower fps
<alyssa>
or something
<alyssa>
tbf, I think also tearing so maybe it's the difference of 1 extra copy
<daniels>
eh? there are no extra copies
<daniels>
(in fact one fewer compared to Shell on X11)
<daniels>
tearing wouldn't entirely surprise me if you were running Xwayland clients
<alyssa>
daniels: then I don't know what's wrong in shell/mutter
<macc24>
alyssa: i might be able to try replicating that on a chromebook tomorrow
<alyssa>
(Tearing on X11 but not Wayland, I meant)
<macc24>
oh x11
<daniels>
they've definitely had issues in the past with the main thread being overloaded, or just having blocking stuff run within it, or frame pacing, or whatever
<alyssa>
nod
<daniels>
(oh yeah, tearing in X11 is basically guaranteed)
<daniels>
but tbf Mutter is now the gold standard for aggressively-performant frame pacing and timing
<alyssa>
also suspicious that the cursor plane isn't getting used with their Wayland backend
<macc24>
alyssa: i think it is
<macc24>
when it broke on mt8183 i noticed on gnome wayland first
<alyssa>
also possible this is fixed on new GNOMEs and I need to update
karolherbst_ has joined #panfrost
karolherbst has quit [Ping timeout: 480 seconds]
nlhowell has quit [Remote host closed the connection]
nlhowell has joined #panfrost
moa has joined #panfrost
kenzie3 has joined #panfrost
ezequielg_ has joined #panfrost
narmstrong_ has joined #panfrost
robclark_ has joined #panfrost
cwabbott_ has joined #panfrost
robher_ has joined #panfrost
taowa_ has joined #panfrost
jstultz_ has joined #panfrost
rcf1 has joined #panfrost
jolan_ has joined #panfrost
wicastC has joined #panfrost
daniels_ has joined #panfrost
cyrozap-ZNC has joined #panfrost
rcf has quit [synthon.oftc.net reflection.oftc.net]
bluebugs has quit [synthon.oftc.net reflection.oftc.net]
jolan has quit [synthon.oftc.net reflection.oftc.net]
robink_ has quit [synthon.oftc.net reflection.oftc.net]
kenzie has quit [synthon.oftc.net reflection.oftc.net]
robher has quit [synthon.oftc.net reflection.oftc.net]
daniels has quit [synthon.oftc.net reflection.oftc.net]
ezequielg has quit [synthon.oftc.net reflection.oftc.net]
narmstrong has quit [synthon.oftc.net reflection.oftc.net]
taowa has quit [synthon.oftc.net reflection.oftc.net]
cwabbott has quit [synthon.oftc.net reflection.oftc.net]
ckeepax has quit [synthon.oftc.net reflection.oftc.net]
jstultz has quit [synthon.oftc.net reflection.oftc.net]
robclark has quit [synthon.oftc.net reflection.oftc.net]
cyrozap has quit [synthon.oftc.net reflection.oftc.net]
wicast has quit [synthon.oftc.net reflection.oftc.net]
cwabbott_ has quit []
cwabbott has joined #panfrost
daniels_ is now known as daniels
robclark_ has quit []
ckeepax has joined #panfrost
robink has joined #panfrost
robclark has joined #panfrost
robher_ has quit []
robher_ has joined #panfrost
robher_ has quit []
robher has joined #panfrost
narmstrong_ has quit []
narmstrong has joined #panfrost
camus has joined #panfrost
camus1 has quit [Remote host closed the connection]
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
jstultz_ has quit []
jstultz_ has joined #panfrost
jstultz_ has quit []
jstultz has joined #panfrost
MajorBiscuit has quit [Ping timeout: 480 seconds]
tlwoerner has joined #panfrost
karolherbst_ is now known as karolherbst
rasterman has quit [Quit: Gettin' stinky!]
<alyssa>
Alyssa's public panfrost to-do list, maybe if she makes it public she'll be forced to actually do it
<alyssa>
[ ] Finish writing that blog post she said she would (..teaser?)
<alyssa>
[ ] Finish scoreboarding patches. Needs some dEQP love.
<alyssa>
[ ] Finish boolean rework series. Code is untested. Hopefully I didn't break dEQP too bad.
<alyssa>
[ ] Clean up Valhall kernel patches, get them on the ML.
nlhowell has quit [Ping timeout: 480 seconds]
<alyssa>
[ ] Clean up Valhall Mesa patches, land as much as reasonable now that I can actually test them.
<alyssa>
[ ] Debug more deqp's on Valhall. Textures are broken badly, etc. >90% pass rate on Valhall should be doable soonish.
<alyssa>
[ ] Test dschuermann's vectorization stuff so I can feel good about giving them my reviewed-by
<alyssa>
[ ] Investigate downstream bug fixes, upstream where it makes sense to do so.
<alyssa>
[ ] Investigate Piglit fails on Bifrost.
<alyssa>
[ ] Write value discard optimization for valhall compiler once enough of deqp is passing to test said opt
<alyssa>
(also, like, school. How dreadfully boring :-p)
<dschuermann>
is the priority as listed? :P
<dschuermann>
because then I can probably target mesa 25.3 for the vectorization stuff ;)
<dschuermann>
err, sorry if that sounded mean. you are actually amazing fast working. I only wished the whole stupid thing was already upstream, and no idea who else to annoy with it
<HdkR>
>90% pass rate on Valhall might make some people ridiculously happy
<dschuermann>
true that... :)
<dschuermann>
I'll just ping you when I get into urgent need for that MR, okay? (there are some other requirements before it becomes useful anyways)