alyssa changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - https://gitlab.freedesktop.org/panfrost - Logs https://freenode.irclog.whitequark.org/panfrost - <daniels> avoiding X is a huge feature
stikonas has quit [Remote host closed the connection]
vstehle has quit [Ping timeout: 258 seconds]
NeuroScr has quit [Quit: NeuroScr]
davidlt_ has joined #panfrost
davidlt_ is now known as davidlt
vstehle has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo has joined #panfrost
buzzmarshall has quit [Remote host closed the connection]
davidlt has quit [Ping timeout: 272 seconds]
somy has joined #panfrost
stikonas has joined #panfrost
stikonas has quit [Remote host closed the connection]
<shadeslayer[m]> Oh ffs
<HdkR> oh noe
<shadeslayer[m]> that's what I was referring to
<shadeslayer[m]> tomeu: ^^ thoughts?
raster has joined #panfrost
<shadeslayer[m]> unless I guess we don't want to modify the context in pan_blit
belgin has joined #panfrost
<tomeu> shadeslayer: I'm having trouble understanding what is being asked
<belgin> hello
belgin has quit [Quit: Leaving]
<shadeslayer[m]> <tomeu "shadeslayer: I'm having trouble "> I'm having trouble understanding why ctx->in_wallpaper isn't modified in panfrost_blit_wallpaper
<shadeslayer[m]> tomeu: but I guess because we don't want to modify the context state in a blit function?
embed-3d has quit [Ping timeout: 258 seconds]
davidlt has joined #panfrost
raster has quit [Remote host closed the connection]
embed-3d has joined #panfrost
<alyssa> shadeslayer[m]: That code is overhauled anyway from the scoreboarding patch
<alyssa> WHich I maybe should've broken up but history was a mess and I was in a hurry to get it sent off to the list before disconnecting for the day so I just squashed it all
mearon has quit [Ping timeout: 258 seconds]
<tomeu> alyssa: if you have one or more branches with the patches you sent to the ml, would be great if you could push them to my repo for CI
<alyssa> tomeu: Sure :)
<tomeu> thanks!
<alyssa> tomeu: (I'll be on and off a lot today, have appointments)
<tomeu> np, I won't be around for much longer today
<shadeslayer[m]> <alyssa "shadeslayer: That code is overha"> Awesome :)
buzzmarshall has joined #panfrost
<tomeu> alyssa: btw, have pushed some of your patches to this branch whcih is rebased on top of master, as otherwise the CI wouldn't be green: https://gitlab.freedesktop.org/tomeu/mesa/commits/master-ci
<tomeu> armhf passed, there's quite a bit of backlog for arm64
afaerber has quit [Quit: Leaving]
raster has joined #panfrost
afaerber has joined #panfrost
yann has joined #panfrost
<tomeu> alyssa: the CI is happy for this branch to be pushed: https://gitlab.freedesktop.org/tomeu/mesa/commits/master-ci
raster has quit [Remote host closed the connection]
* alyssa eyes
<tomeu> alyssa: btw, the branches ubos and compiler-opts-v2 didn't pass because they are based on a master revision that was still broken
<tomeu> if you rebase, they should pass if they don't introduce regressions
<alyssa> OK
yann has quit [Ping timeout: 244 seconds]
<alyssa> tomeu: Just landed the stuff in master-ci
<tomeu> nice!
<alyssa> https://lists.freedesktop.org/archives/dri-devel/2019-June/223323.html is an interesting patch that cropped up today.
<alyssa> For context, this is to expose a fourcc code for Utgard-style tiling.
<alyssa> This is critical for lima, since that's how they share buffers.
<alyssa> It's somewhat less critical for Panfrost, since Midgard introduces support for AFBC (which is much better performance than uncompressed tiling).
<alyssa> (Though it still matters for us, since not all formats that can be tiled can be compressed. sRGB is a good example where this may matter.)
<alyssa> The patch doesn't say too much about the tiling scheme itself, so if you're curious, lima and panfrost share a free software implementation of the shceme:
<alyssa> I should comment that routine a bit better though :)
<alyssa> (And by that I mean at all)
<alyssa> Huge thank you to libv and anarsoul for working on that file :)
NeuroScr has joined #panfrost
* anarsoul vaguely remembers that he copied this code from panfrost and tweak it a bit. Credit should go to libv and alyssa.
<alyssa> anarsoul: You did load() routines, that's a big deal
stikonas has joined #panfrost
yann has joined #panfrost
rhyskidd_ has joined #panfrost
NeuroScr has quit [Quit: NeuroScr]
rhyskidd has quit [Ping timeout: 272 seconds]
rhyskidd_ is now known as rhyskidd
NeuroScr has joined #panfrost
davidlt has quit [Ping timeout: 258 seconds]
buzzmarshall has quit [Remote host closed the connection]
buzzmarshall has joined #panfrost
pH5 has joined #panfrost
* alyssa thought she had Inspiration but nope
<anarsoul> alyssa: Friday's almost over
<alyssa> Is it? :(
<alyssa> *FINALLY* got blitting doing something reasonable
<alyssa> Er mipmaps
<alyssa> Needs a lot of cleanup but the theory is solid I think
<alyssa> And we're passing a ton of new dEQP tests
<alyssa> So that's kind of a breakthrough
<alyssa> Even though it was kind of trivial and it's still broken in more ways than one
NeuroScr has quit [Quit: NeuroScr]
mearon has joined #panfrost
<sravn> alyssa: Considering there are only 635 cases listed in expected-failures, it is quite good to pass a ton of nw dEQP tests (assuming a ton equals 1000)
<sravn> Sorry, could not resist. Nice to follow the daily progress of panfrost!
<alyssa> sravn: :D
<alyssa> I think there may be some failures not listed in that file that we don't bother trying to run
pH5 has quit [Quit: -_-]
NeuroScr has joined #panfrost
NeuroScr has quit [Quit: NeuroScr]
<HdkR> alyssa: Don't worry about unaligned UBO loads btw. std140 UBO alignment enforces vec4 alignment :P
<alyssa> HdkR: But does NIR?
<HdkR> If it doesn't then there should be an option for it since old hardware will break
<HdkR> There are a couple of /very/ new extensions that lift the alignment requirements
<alyssa> Meeep
stikonas has quit [Remote host closed the connection]