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 and - Contact ARM for binary driver support!
kaspter has joined #lima
kaspter has quit [Excess Flood]
kaspter has joined #lima
Barada has joined #lima
Barada has quit [Quit: Barada]
monstr has joined #lima
dev1990 has joined #lima
camus has joined #lima
kaspter has quit [Ping timeout: 240 seconds]
camus is now known as kaspter
ddevault_ has joined #lima
ddevault has quit [Quit: Why do I even put this quit message in if I never quit]
ddevault_ is now known as ddevault
Barada has joined #lima
BenG83 has joined #lima
BenG83 has quit [Quit: Leaving]
Barada has quit [Quit: Barada]
Barada has joined #lima
chewitt has quit [Quit: Adios!]
Barada has quit [Quit: Barada]
Ntemis has joined #lima
kaspter has quit [Quit: kaspter]
monstr has quit [Remote host closed the connection]
Ntemis has quit [Read error: Connection reset by peer]
<enunes> linkmauve: seems like a great improvement on gtk 4.2, I'm getting 60ish fps on gears, transitions seems to be sane
<linkmauve> Woah!
<linkmauve> I never got 60 fps anywhere yet.
<linkmauve> Still about 37 fps in gears.
<linkmauve> On a MP2.
<anarsoul> likely enunes uses a board with mali450
<enunes> yeah this is s905 mali450
<linkmauve> I might have a s905x.
<linkmauve> Never managed to boot it yet.
<enunes> still even here I was not getting more than 40 weeks ago
<anarsoul> it also has faster memory
<anarsoul> gpu is really memory bandwidth constrained on A64
<enunes> and this is weston, it still refuses to start for me on X but I dont know if I have something else wrong
<anarsoul> so overclocking the memory has immediate effect on GPU performance
<anarsoul> enunes: btw, regarding - did you dump what near and far values are set in render->depth_range?
<enunes> anarsoul: I did inspect some dumps, the values seemed to be the different-by-1 that we write, still it didnt work
<enunes> then I tried to subtract 'near' by 2 but that didnt seem to work either, increasing 'far' seemed to work so I sent that for the user to test
<enunes> I dont know what is wrong with those magic values
<anarsoul> sounds like we have a bug somewhere else
<enunes> yeah it could be
<anarsoul> not sure if depth is represented anywhere by fp16
<anarsoul> but if it is - it could be the issue
<enunes> I think the apps shouldn't be setting near == far anyway
<enunes> so honestly I just wanted to build on the existing workaround and move on
<anarsoul> app may not know what should be epsilon value for the driver
<enunes> I dont know where that epsilon thing came from, I couldnt find other drivers doing it
<linkmauve> enunes, weston should be ideal in terms of memory bandwidth, it will attempt to use DRM planes for foreground applications if they are available and the formats match, so it will not go back to the GPU if it can avoid it.
<enunes> linkmauve: on X it just gives me 'The program 'gtk4-demo' received an X Window System error.', do you get the same?
<linkmauve> You can enable composition colour (Super+Shift+Space+s, do it again to disable it) to see which surfaces get composited and which ones get promoted to a plane.
<linkmauve> I don’t even have Xwayland installed on any of my boards.
<linkmauve> I can try a bit later.
<enunes> not xwayland, this is just raw X or with something like xfce
<linkmauve> Ah uh, that’s gonna be much harder to setup, I haven’t done that in years.
<enunes> its ok, I was just curious to see if we can close now
<enunes> I think Xorg worked before so I wonder what is going on now
<adjtm_> Replicant 11 is using mesa on a nearly mainline kernel. Lima is super snappy, and everything I have tested so far (including firefox and webview) works perfectly except that crashes while starting
<adjtm_> so congratulations and thank you to everyone involved
<adjtm_> I will debug gears and send a bug report if it's a lima's issue
<anarsoul> adjtm_: great!
<anarsoul> adjtm_: out of curiosity, what mesa version do you use?
<adjtm_> let me see...
<linkmauve> enunes, ah right, in I only considered Wayland, not X11…
<linkmauve> Sorry.
<enunes> linkmauve: I just posted a comment on the lima issue dont know if you saw it; it works on Xorg too but still needs the debug build and env var
<enunes> and performs pretty badly with a compositing manager on, while on wayland it is quite great
<adjtm_> anarsoul, 21.0.0-devel
<adjtm_> from 2020-11-15
<anarsoul> adjtm_: you should probably update your mesa :)
<linkmauve> enunes, the need for a debug build would make sense, the fallback never got written for GLX probably.
<anarsoul> I'm not sure if android enables disk shader cache, but you would at least benefit from in-memory shader cache
<adjtm_> mesa is right now the least of our problems in replicant 11, that's why we had not much hurry to update it
<adjtm_> :)
<adjtm_> llvmpipe also works reasonably well
<enunes> linkmauve: makes sense then, well I guess everything going on in the pinephone as the user of that is only on wayland anyway
<adjtm_> anarsoul, this is galaxy s3 lte with Exynos 4412 (quad cortex-a9) which has Mali-400MP4
<anarsoul> adjtm_: you should probably cooperate with GloDroid to get Android working on Pinephone :)
<linkmauve> It’s the same on my other computers, I haven’t had to care about Xorg in years, and I use Xwayland less and less.
<adjtm_> anarsoul, I'm working on it
putti_ has joined #lima
enty has joined #lima
fourkbomb has joined #lima
Net147_ has joined #lima
ecloud_ has joined #lima
Net147 has quit [Ping timeout: 265 seconds]
ecloud has quit [Write error: Connection reset by peer]
Putti has quit [Ping timeout: 265 seconds]
forkbomb has quit [Read error: Connection reset by peer]
eightdot has quit [Ping timeout: 240 seconds]
ente has quit [Read error: Connection reset by peer]
smaeul has quit [Ping timeout: 240 seconds]
smaeul has joined #lima