chewitt has quit [Quit: Zzz..]
NeuroScr has quit [Quit: NeuroScr]
stikonas has quit [Ping timeout: 258 seconds]
stikonas has joined #panfrost
indy has joined #panfrost
vstehle has quit [Ping timeout: 268 seconds]
stikonas has quit [Remote host closed the connection]
anarsoul|c has quit [Quit: Connection closed for inactivity]
alyssa is now known as alrlied
alrlied is now known as alyssa
<
alyssa>
Oops, I'm starting a nontrivial refactor of the scheduler
<
alyssa>
I feel like my coding ability now is just soo much better than this time last year
<
alyssa>
And it shows in the codebase
<
HdkR>
Woo improving over time
<
alyssa>
Wonder if now's a good time to write a better scheduler entirely ....
vstehle has joined #panfrost
robclark has quit [Ping timeout: 252 seconds]
ric96 has quit [Ping timeout: 252 seconds]
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo has joined #panfrost
ric96 has joined #panfrost
robclark has joined #panfrost
BenG83 has quit [Ping timeout: 258 seconds]
pH5 has joined #panfrost
robertfoss has quit [Ping timeout: 245 seconds]
robertfoss has joined #panfrost
gcl_ is now known as gcl
_whitelogger has joined #panfrost
indy has quit [Read error: Connection reset by peer]
herbmillerjr has quit [Ping timeout: 258 seconds]
indy has joined #panfrost
<
tomeu>
alyssa: I'm personally quite happy with panfrost's code quality, specially given its age
<
tomeu>
but when you write your 3rd driver, I think it will be much better :)
<
HdkR>
Sounds like a good time for a Vulkan driver at that point :P
<
tomeu>
isn't always a good time for a Vulkan driver? :)
<
HdkR>
Maybe post-shader compiler bringup anyway :P
<
HdkR>
Test harness in GL takes so much less time than Vulkan
<
HdkR>
It's definitely a bit higher on my priority list for me at least :D
<
alyssa>
tomeu: 160 char lines are a regular occurrence.. =P
<
tomeu>
I don't care that much about that particular nit :)
raster has joined #panfrost
<
HdkR>
160 is nothing on my 4k display
<
HdkR>
I've got 294 columns :P
<
HdkR>
Or 146 when I'm running my common split size
marcodiego has joined #panfrost
yann has joined #panfrost
raster has quit [Remote host closed the connection]
BenG83 has joined #panfrost
jeez_ has joined #panfrost
jeez_ has quit [Ping timeout: 264 seconds]
pH5 has quit [Quit: bye]
tlwoerner has quit [Quit: Leaving]
tlwoerner has joined #panfrost
yann has quit [Ping timeout: 244 seconds]
suihkulokki has quit [Read error: error:1408F119:SSL routines:ssl3_get_record:decryption failed or bad record mac]
suihkulokki has joined #panfrost
suihkulokki has quit [Read error: error:1408F119:SSL routines:ssl3_get_record:decryption failed or bad record mac]
suihkulokki has joined #panfrost
herbmilleriw has quit [Remote host closed the connection]
herbmilleriw has joined #panfrost
NeuroScr has joined #panfrost
stikonas has joined #panfrost
yann has joined #panfrost
stikonas_ has joined #panfrost
stikonas has quit [Ping timeout: 246 seconds]
jernej has joined #panfrost
<
robher>
alyssa: can I assume GROW_ON_GPF functionality doesn't need to be exported (aka prime dmabuf) or need CPU accesses?
<
anarsoul>
wouldn't it be difficult to inspect its contents if it won
<
anarsoul>
if it won't support CPU accesses?
BenG83 has quit [Quit: Leaving]
NeuroScr has quit [Quit: NeuroScr]
<
robher>
anarsoul: if only debug, that could be addressed separately.
<
alyssa>
robher: Definitely never exported.
<
alyssa>
And no, not CPU accessible
<
alyssa>
(Not even for debug -- if there's a debug issue, I'd just do a special build that doesn't set 'growable')
<
alyssa>
So yes, those are both good assumptions :)
<
robher>
alyssa: would it be possible to also not even expose to userspace? Seem userspace just stuffs in fixed values.
<
robher>
alyssa: though that would introduce the kernel touching the execution buffer which we avoid currently. So maybe not the best idea.
rhyskidd has quit [Ping timeout: 258 seconds]
rhyskidd has joined #panfrost
NeuroScr has joined #panfrost