stikonas has quit [Remote host closed the connection]
_whitelogger has joined #panfrost
_whitelogger has joined #panfrost
atler has quit [Killed (verne.freenode.net (Nickname regained by services))]
atler has joined #panfrost
vstehle has quit [Ping timeout: 264 seconds]
archetech has joined #panfrost
<indy>
macc24, looks like. anyway i refuse polute system with gnome3/kde. are there any other possibilities for wayland + panfrost?
<macc24>
indy: sway
davidlt has joined #panfrost
<macc24>
that's what i use on all my panfrost machines
<macc24>
phosh worked for me too, and icecream95 has launched wayfire on bifrost
<indy>
macc24, and replacement for lxdm? (no gdm pls)
<macc24>
just use getty?
<macc24>
with something like 'tty | grep tty1 && exec sway' in bashrc
cphealy has joined #panfrost
<indy>
looks good, but... migration from windowmaker will be nut
<archetech>
make your own weston desktop per arch wiki :)
<macc24>
and write your own kernel :)
archetech has quit [Quit: Konversation terminated!]
<indy>
there few thing i need - hexchat (using irssi in gnu screen/tmux is good for knotting fingers)
<indy>
what is replacement for glxgears?
vstehle has joined #panfrost
<icecream95>
indy: Xwayland works pretty well, so most applications using X should be usable
orkid has quit [Ping timeout: 272 seconds]
<indy>
icecream95, any hwoto available? :) in already installed xwayland and weston
<icecream95>
indy: For weston, you need to run weston --xwayland
<indy>
huh xwayland has no man page
<icecream95>
indy: The manual (and program itself) is called "Xwayland", with a capital 'X'
<indy>
i know, dpkg -L xwayland listed only /usr/bin/Xwayland and 'mandatory' files in /usr/share/doc/xwayland
<indy>
no man pages
<icecream95>
My copy of the Xwayland manual page is from compiling xserver from git, so maybe it was only added recently
BenG83 has joined #panfrost
BenG83 has quit [Quit: Leaving]
stikonas has joined #panfrost
karolherbst has joined #panfrost
raster has joined #panfrost
m][sko has joined #panfrost
m][sko has quit [Quit: Connection closed]
alyssa has joined #panfrost
Elpaulo has joined #panfrost
<macc24>
G31 may not be as slow as i thought it is
<macc24>
if not for fact that the guy who made panel ran on it messed up colors it would be barely playable on 640x480
<macc24>
still, jokes about G31 will never die
wwilly__ has joined #panfrost
wwilly__ is now known as wwilly
wwilly_ has quit [Ping timeout: 272 seconds]
<KungFuJesus>
daniels: is hardware cursor support possible on panfrost?
<alyssa>
unrelated
<alyssa>
rockchip/amlogic/etc display drivers need to support, or not, panfrost has no say in it
<amonakov>
yeah, it was a surprise for me when learning about Arm socs (that the "3D core" and "2D/video core" are quite independent)
<daniels>
right, Rockchip at least has kernel support for cursor planes; I don't believe Amlogic does
<KungFuJesus>
So it's up to the kernel devs? Or is the hardware cursor completely foreign to the way the architecture was written?
<daniels>
it's totally unrelated
<daniels>
hardware cursor support (as well as HDMI/DisplayPort/etc) is provided by whoever made your SoC - Rockchip, Amlogic, MediaTek, whoever - each of them have their own implementation of a display controller (using DRM/KMS), so that capability is up to them
<daniels>
the Mali GPU is a totally separate hardware block which only provides accelerated rendering
<KungFuJesus>
ok but how much of that is baked into the hardware? I'm just wondering if it's possible somebody could patch Amlogic's drivers to support it some time in the future or if there's something very specific to the SoC that makes it impossible
wwilly_ has joined #panfrost
wwilly has quit [Ping timeout: 246 seconds]
wwilly_ has quit [Client Quit]
<daniels>
it's purely hardware functionality
<daniels>
you can't have a software implementation of a hardware cursor
<daniels>
you can have software cursor support, but well, it's not the same thing :)
wwilly_ has joined #panfrost
<bbrezillon>
Amlogic might not have a dedicated cursor planes, but I'm pretty sure they have more than one HW plane, isn't the compositor able to repurpose one of those for the cursor?
<bbrezillon>
*Amlogic SoCs
<macc24>
mt8183 has hardware cursor
<robmur01>
FWIW, even on PC GPUs the 2D and 3D bits tend to be distinct once you look closely enough, they're just handled by different parts of one massive kernel driver :)
<alyssa>
robmur01: mali-dp+mali-gpu when :p
<robmur01>
alyssa: Xiaomi Surge, that was when :P
<robmur01>
even had our VPU too...
<alyssa>
:p
<alyssa>
Juno? :p
<robmur01>
nah, predates Mali-DP so has the crusty old not-a-real-product HDLCD
<robmur01>
you'll have to wait for some kind of new testchip to come out... :P
<alyssa>
robmur01: I wonder what that could be referring to.
<robmur01>
or get lots of monies then contact sales with your desire to license GPU+DPU to build a discrete PCIe endpoint :D
warpme_ has quit [Quit: Connection closed for inactivity]
<alyssa>
a discrete tiler would be... unusual
<amonakov>
modern desktop nvidia and amd are tiling though. not sure about intel
Lyude has quit [Read error: Connection reset by peer]
Lyude has joined #panfrost
guillaume_g has quit [Quit: Konversation terminated!]
<daniels>
bbrezillon: yeah compositors _can_ but don't always do it
<daniels>
amonakov: well, they're tiling in the sense that they store their framebuffers in a tiled format and they do per-tile rasterisation + shading, but the last I saw the various engines still had direct access to the entire framebuffer - unlike Mali where you have an intermediate tile buffer where the old framebuffer is loaded into the tile buffer, the tile is fully shaded (into the tile buffer), then written out to sysmem