<anarsoul>
rellla: btw have you pin-pointed which tests cause GP hang?
<anarsoul>
I guess there's no low hanging fruits left :)
<anarsoul>
well, dEQP-GLES2.functional.depth_range.write.reverse is easy to fix
<anarsoul>
just swap near/far if far < near
<anarsoul>
and some other depth_range tests
warpme_ has quit [Quit: Connection closed for inactivity]
yuq825 has quit [Remote host closed the connection]
yuq825 has joined #lima
dddddd has quit [Remote host closed the connection]
_whitelogger has joined #lima
megi has quit [Ping timeout: 265 seconds]
Barada has joined #lima
jailbox has quit [Ping timeout: 268 seconds]
yann has quit [Ping timeout: 258 seconds]
Elpaulo has joined #lima
warpme_ has joined #lima
yuq825 has quit [Remote host closed the connection]
yuq825 has joined #lima
yann has joined #lima
arti has quit [Quit: arti]
arti has joined #lima
yuq825 has quit [Remote host closed the connection]
yuq825 has joined #lima
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo has joined #lima
yuq825 has quit [Remote host closed the connection]
megi has joined #lima
yuq825 has joined #lima
dddddd has joined #lima
buzzmarshall has joined #lima
Barada has quit [Quit: Barada]
yuq825 has quit [Quit: Leaving.]
Elpaulo has quit [Ping timeout: 240 seconds]
<rellla>
i got all the blend.* tests fixed :)
<rellla>
anarsoul: was one of the low hanging fruits btw...
yann has quit [Ping timeout: 265 seconds]
<anarsoul>
:)
<anarsoul>
rellla: great!
<anarsoul>
rellla: hm, maybe I misread the patch, I'll look into it again my coffee
<anarsoul>
don't change anything right now
<rellla>
using your ci it fixes 107 tests
<rellla>
thanks
<anarsoul>
that's not my CI btw, it's BayLibre's
yann has joined #lima
mripard has quit [Ping timeout: 258 seconds]
ezequielg has quit [Read error: Connection reset by peer]
marex-cloud has quit [Ping timeout: 252 seconds]
robher has quit [Ping timeout: 245 seconds]
warpme_ has quit [Ping timeout: 272 seconds]
anarsoul|c has quit [Ping timeout: 268 seconds]
anarsoul|c has joined #lima
robher has joined #lima
warpme_ has joined #lima
ezequielg has joined #lima
ezequielg has quit [Ping timeout: 260 seconds]
buzzmarshall has quit [Remote host closed the connection]
Elpaulo has joined #lima
warpme_ has quit [Quit: Connection closed for inactivity]
anarsoul|c has quit [Quit: Connection closed for inactivity]
<anarsoul>
enunes: have you tested your branch with deqp?
<enunes>
anarsoul: yes, same set of tests as the earlier comment
<enunes>
mesa 20.0 is looking great for lima
<anarsoul>
yeah
<enunes>
I tried some actual graphical apps, games, firefox some days ago, I'm surprised by how much seems to work
<anarsoul>
how's firefox?
<enunes>
on wayland seemed ok, I force enabled webgl and despite it complaining of some opengl parameter gets, I saw spinning cube
<anarsoul>
what about X11? :)
<enunes>
I tried xfce and some apps seemed ok but overall desktop with some flickering
<anarsoul>
darn
<anarsoul>
GPU errors?
<enunes>
some apps failed to allocate memory, but I didn't investigate into that
<anarsoul>
ah
<anarsoul>
we have to do something about kmsro
<enunes>
I used to see ppir regalloc fail with that in the past, with my branch at least those are gone, and other than flickering, no other misrendering or black windows
<anarsoul>
we quite often allocate BOs through it but it's not really necessary
<anarsoul>
since they're not passed to display driver
<enunes>
one thing I always have to do is tweak vm.min_free_kbytes so it doesn't run into page allocation errors, is this something you do too?
<anarsoul>
nope
<anarsoul>
update your kernel? :)
<enunes>
with recent kernels, currently I'm running 5.5-rc6
<anarsoul>
oh
<anarsoul>
it's OK for me with 2G of RAM
<enunes>
2G RAM too
<enunes>
what is vm.min_free_kbytes on your system?
<anarsoul>
5344
<anarsoul>
that's on 5.5-rc2
<enunes>
yeah I guess with that piglit randomly runs into some page allocation failures sometimes
<anarsoul>
(with some patches on top)
<enunes>
I tweaked this a while ago on 5.4 or earlier
<enunes>
I gave it a comfortable 67584 and never saw those failures again