ChanServ changed the topic of #lima to: Development channel for open source lima driver for ARM Mali4** GPUs - Kernel has landed in mainline, userspace driver is part of mesa - Logs at https://people.freedesktop.org/~cbrill/dri-log/index.php?channel=lima and https://freenode.irclog.whitequark.org/lima - Contact ARM for binary driver support!
niceplace has quit [Ping timeout: 260 seconds]
camus1 has joined #lima
kaspter has quit [Remote host closed the connection]
camus1 is now known as kaspter
dev1990 has quit [Quit: Konversation terminated!]
dev1990 has joined #lima
kaspter has quit [Ping timeout: 246 seconds]
kaspter has joined #lima
niceplace has joined #lima
camus1 has joined #lima
kaspter has quit [Read error: Connection reset by peer]
camus1 is now known as kaspter
rando25892 has quit [Ping timeout: 256 seconds]
rando25892 has joined #lima
Elpaulo has quit [Quit: Elpaulo]
jernej has quit [Ping timeout: 264 seconds]
jernej has joined #lima
monstr has joined #lima
kaspter has quit [Quit: kaspter]
_whitelogger has joined #lima
kaspter has joined #lima
<enunes> rellla: not much low hanging either but some help with https://gitlab.freedesktop.org/mesa/mesa/-/issues/3592 could be useful
<enunes> I think there may be a gtk4 bug in the mix since trying to make it use gles2 doesnt work
<enunes> maybe with the latest version
<enunes> but then its more time to spend debugging already
<rellla> enunes: i'm afraid of setting up anything other than deqp :p
<rellla> i'm about to pick up the outstanding 2 patches from cwabbott here https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2315
<rellla> i rebased them and got them working. there was a bug in lima_nir_split_loads which caused a nir_validate assert, but i fixed that.
<enunes> seems helpful too
<rellla> i have still 2 regressions left and have to look into them. then give shader-db a chance ...
<linkmauve> enunes, the visual glitch looks similar to the one on my Olimex board: https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/2792
<rellla> after that, i will either try to implement shader cache or look into the outstanding gpir issue, which causes the 512 instr crash
<linkmauve> I haven’t been able to reproduce on my PinePhone though.
<enunes> linkmauve: I can reproduce it locally on X bug not on wayland
<enunes> I think I was able to get an apitrace
<linkmauve> Ah, for me it’s all Wayland.
<enunes> but it needs time to debug, I'm working on something else now with the time I can get
<enunes> so some help would be good on that one
<enunes> also to untangle the gtk4 situatiln with gles2
<linkmauve> With the MR I just linked, it at least runs there.
<enunes> ah I see
<enunes> yeah that patch may be helpful
<enunes> linkmauve: is there a way to make gtk4 work on gles2 without rebuilding it with debug options though?
<enunes> it seems to request gl 3 by default
<enunes> and that debug option is not available in distribution builds
<linkmauve> enunes, pass GDK_GL=gles in the environment.
kaspter has quit [Quit: kaspter]
chewitt has joined #lima
<enunes> linkmauve: I found some info about GDK_GL online but it seemed to apply only for gtk <4, should it still work on 4?
<enunes> if I set GDK_GL=gles GDK_DEBUG=opengl it still requests desktop gl 3 or 4
<linkmauve> It does make gtk4-demo run on GLES instead of GL here it seems.
<linkmauve> Oh?
<linkmauve> Hmm, maybe I should rebuild Mesa without GL 2.1 support.
<linkmauve> But then, it should have worked because before this patch, libepoxy was telling me that the symbols for perf queries weren’t available.
<enunes> actually I just tried it on my PC right now
<enunes> GDK_GL=gles GDK_DEBUG=opengl gtk4-demo --run gears
<enunes> OpenGL version: 4.6 (core)
<linkmauve> Here it says “Unable to create a GL context” on my phone.
<linkmauve> “Gdk-Message: 14:43:30.316: Creating EGL context version 3.2 (debug:no, forward:no, legacy:no, es:no)” what does that mean?
<enunes> GDK_DEBUG=gl-gles,opengl gtk4-demo --run gears
<enunes> Creating EGL context version 2.0 (debug:no, forward:no, legacy:no, es:yes)
<enunes> OpenGL ES version: 3.2 (core)
<enunes> that is a little better, it does set the es bit
<linkmauve> Oh, it runs, slowly.
<linkmauve> 5.6fps
<linkmauve> 8% of CPU used, and IIRC no way to get that information from Lima.
<linkmauve> Makes the whole phone laggy too.
afaerber has quit [Ping timeout: 260 seconds]
<ente> it's a linkmauve :o
<linkmauve> ente… Do I know you from somewhere? Your nick doesn’t ring anything sorry. :x
<ente> fw.fr
<ente> dunno if I still have the muc open
<linkmauve> Oh, I’ve left this one long ago. ^^
afaerber has joined #lima
yann has quit [Read error: No route to host]
yann has joined #lima
yann has quit [Ping timeout: 240 seconds]
yann has joined #lima
yann has quit [Ping timeout: 272 seconds]
yann has joined #lima
yann has quit [Ping timeout: 264 seconds]
<anarsoul> rellla: IIRC two remaining patches cause significant issues with gpir compiler
yann has joined #lima
<rellla> anarsoul: i think i could fix it somehow. i will push the code tomorrow. there is one more >512 instr test failing and one “invariant“ test.
<rellla> for the latter, i'm still wondering, why it fails. i will post some more info tomorrow
<rellla> except these, there are no deqp regressions
monstr has quit [Remote host closed the connection]
drod has joined #lima
drod has quit [Read error: Connection reset by peer]
niceplace has quit [Ping timeout: 265 seconds]
drod has joined #lima
drod has quit [Client Quit]
_whitelogger has joined #lima
putti_ has joined #lima
lvrp16 has quit [Ping timeout: 240 seconds]
Putti has quit [Ping timeout: 240 seconds]
lvrp16 has joined #lima