<plaes>
mripard, wens: I marked merged patches in the mainlining effort page
<plaes>
..and updated 4.5 section too
naobsd has joined #linux-sunxi
<rellla>
GeneralStupid: compiling ffmpeg and mpv worked for me some time ago. without logs it's impossible to help you...
<rellla>
though it's more a debian/arm related question, not linux-sunxi specific.
<rellla>
<GeneralStupid>jelle: and sorry, i dont guess why nobody simply uploads his built mpv dpkg's...
<rellla>
because nobody has such files available atm?
<mripard>
plaes: I guess we can remove them entirely from the wip section
<mripard>
it's not a work in progress anymore
leviathancn has joined #linux-sunxi
leviathancn has quit [Ping timeout: 240 seconds]
<GeneralStupid>
rellla: ... i guess my sd card is full... thats the issue ...
<jelly>
oh dear, another kickstarter for a cheap "open source" aw based device (not going to link it, do a web search on PINE64 if you want)
viccuad has joined #linux-sunxi
<JohnDoe_71Rus>
jelly: some day ago
<NiteHawk>
was on the ML, too
<GeneralStupid>
maybe it would be time to kickstart kodi for allwinner device...
<jelle>
which allwinner though :p
<rellla>
GeneralStupid: uahhh. compiling ffmpeg on sd card \o/
<GeneralStupid>
rellla: nice idea, i know. But i wanted it quick and dirty and now i get the result
leviathancn has joined #linux-sunxi
viccuad has quit [Ping timeout: 272 seconds]
premoboss has quit [Remote host closed the connection]
Gerwin_J has quit [Quit: Gerwin_J]
leviathancn has quit [Ping timeout: 256 seconds]
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
viccuad has joined #linux-sunxi
IgorPec2 has joined #linux-sunxi
leviathancn has joined #linux-sunxi
IgorPec has quit [Ping timeout: 246 seconds]
IgorPec2 has quit [Ping timeout: 246 seconds]
leviathancn has quit [Read error: No route to host]
SadSmile has quit [Quit: Leaving]
leviathancn has joined #linux-sunxi
nove has joined #linux-sunxi
avph has quit [Ping timeout: 264 seconds]
avph has joined #linux-sunxi
cnxsoft has quit [Remote host closed the connection]
afaerber has joined #linux-sunxi
afaerber has quit [Quit: Ex-Chat]
lanodan has quit [Ping timeout: 245 seconds]
leviathancn has quit [Ping timeout: 272 seconds]
reinforce has quit [Quit: Leaving.]
leviathancn has joined #linux-sunxi
Ueno_Otoko- has quit [Ping timeout: 250 seconds]
afaerber has joined #linux-sunxi
afaerber has quit [Quit: Ex-Chat]
afaerber has joined #linux-sunxi
IgorPec has joined #linux-sunxi
MaDMaLKaV has quit [Ping timeout: 256 seconds]
delsio has joined #linux-sunxi
MaDMaLKaV has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
delsio has quit [Ping timeout: 240 seconds]
reinforce has joined #linux-sunxi
FDCX__ has joined #linux-sunxi
FDCX__ has quit [Ping timeout: 240 seconds]
leviathancn has quit [Ping timeout: 240 seconds]
FDCX__ has joined #linux-sunxi
<rellla>
ssvb: whats the order you are doing for the blit fallbacks in fbturbo, e.g. when g2d blit is not possible?
<rellla>
i seem to mix up neon, vfp, arm, pixman ...
delsio has joined #linux-sunxi
<ssvb>
rellla: there are a bunch of alternative implementations, and the fastest one gets selected
<ssvb>
rellla: for cortex-a7 it is vfp and for cortex-a8 it is neon
<rellla>
ssvb: should pixman take care of that?
<ssvb>
rellla: maybe in the future this can be moved into pixman
ganbold has quit [Quit: This computer has gone to sleep]
<rellla>
i think of adding a fallback for blits in libvdpau-sunxi
<ssvb>
rellla: basically, that code tries to minimize the performance penalty when doing uncached memory reads
<rellla>
i haven't looked into pixman, but it should organize the fallbacks itself, shouldn't it?
<ssvb>
rellla: if we add a special property to pixman_image_t (so that the user can mark the image as uncached), then pixman may try to do the same tricks itself
<ssvb>
normal cached memory and uncached (writecombine) memory need different implementations for better performance
<ssvb>
currently pixman does not differentiate between them
reinforce1 has joined #linux-sunxi
<ssvb>
rellla: what exactly are you trying to implement?
reinforce1 has quit [Client Quit]
<rellla>
i'm just making thoughts on how to provide functionality of accelerated blits, when G2D is absent -> H3
<ssvb>
and also A13 :)
<rellla>
to make osd possible without GPU acceleration
<rellla>
so i looked into your code and got a bit confused :)
<GeneralStupid>
hey guys
<GeneralStupid>
i compiled mpv and it runs awesome on just 900 Mhz ! :-D
<ssvb>
rellla: you can try to implement OSD using pixman, and then we can make sure that the required operations are fast
<rellla>
ssvb: ok. i will do the pixman part, and then let's look for the speed.
<ssvb>
rellla: even now the blending functions are using NEON, so probably it will be not too bad
<ssvb>
rellla: oh wait, is the OSD rendering done in YUV format?
<rellla>
how did you find out the threshold? is it clever to avoid g2d blits for "small" blits?
<rellla>
GeneralStupid: congrats, two days later :p
<ssvb>
the performance can be measured, so the threshold is just selected in an experimental way :)
<GeneralStupid>
rellla: i had these issue that it not compiles... BECAUSE OF DISKSPACE...
<rellla>
haha
<GeneralStupid>
rellla: and then i realized that i have a second orangepi with 1 TB USB Disks...
FDCX__ has quit [Ping timeout: 244 seconds]
the193rd_ has joined #linux-sunxi
the193rd has quit [Ping timeout: 245 seconds]
the193rd_ is now known as the193rd
yann|work has quit [Ping timeout: 240 seconds]
<ssvb>
rellla: I have briefly looked at the libvdpau-sunxi code
<ssvb>
rellla: seems like it is possible to do all the rga handling by pixman in normal memory, and then copy it to some physically contiguous location (usable by disp) in the rgba_flush function
<ssvb>
rellla: oh, in fact if this memory is already cached, then everything should work nicely as is with a little bit of tweaking
<rellla>
ssvb: that means, we still use the ve_malloc'ed memory?
<ssvb>
rellla: well, yes, if you want to
<ssvb>
rellla: still I don't quite understand what kind of role is the CPU cache flushing playing there between G2D and CEDAR, maybe jemk can comment
<rellla>
in the meantime i go reading pixman source code ...
leviathancn has joined #linux-sunxi
<rellla>
ssvb: ok. seeing, that i need some more time for understanding how pixman works... alpha blending, scaling and blitting can be done, right? :)
<rellla>
can you give me an entry point?
<ssvb>
rellla: yes, you create pixman_image_t for every image buffer, and then call pixman_image_composite function to do various operations with these images
Gerwin_J has quit [Quit: Gerwin_J]
Dodger78 has quit [Remote host closed the connection]
leviathancn has quit [Ping timeout: 256 seconds]
leviathancn has joined #linux-sunxi
viccuad has quit [Ping timeout: 240 seconds]
caog has quit [Quit: Ex-Chat]
leviathancn has quit [Ping timeout: 256 seconds]
arete74 has quit [Read error: Connection reset by peer]
arete74 has joined #linux-sunxi
buZz has quit [Quit: Lost terminal]
buZz has joined #linux-sunxi
buZz is now known as Guest77067
Guest77067 is now known as buZz
maz_ has quit [Quit: Leaving]
clonak has quit [Ping timeout: 272 seconds]
delsio has quit [Ping timeout: 250 seconds]
<rellla>
ssvb: would it be overkill to use cairo as the frontend?
ricardocrudo has quit [Read error: Connection reset by peer]
<ssvb>
rellla: that would be an extra dependency and a bit of extra overhead
<rellla>
hm. but it seems to be documented well :)
<ssvb>
rellla: I can make a patch for libvdpau-sunxi, this looks rather simple
<rellla>
ha. for you it would be. you are familiar with pixman api ...
<rellla>
i don't give veto if you want to provide some code :p
<ssvb>
rellla: yeah, pixman is not documented well because it was originally a part of Xorg, which got moved into a separate library to use it also as a software rendering backend for cairo
<ssvb>
rellla: so all the people involved used to be already familiar with this code :)
<ssvb>
but now pixman is also used by wayland, qemu and maybe some other projects
<rellla>
if i understand correctly, cairo is the software counterpart to 2D-OpenGL or OpenVG?
<rellla>
*cpu-based
<ssvb>
cairo has many backends, it can also use GL instead of pixman or generate something like SVG files as output
clonak has joined #linux-sunxi
<ssvb>
rellla: how do you usually test subtitles support?
<ssvb>
rellla: can you recommend some test samples?
destroyedlolo has joined #linux-sunxi
<rellla>
do you mean osd in general or "real" subtitles
<destroyedlolo>
Hello,
<ssvb>
I mean something that stresses all code paths in the rgba code
<destroyedlolo>
Can someone help me to use ft5x touchscreen on my A13 tablette ?
<destroyedlolo>
I posted some input on the group.
<rellla>
ssvb: i have www.tvdr.de running here, which can display very easy and very heavy designed osd
<rellla>
VDR causes video framedrops, if the blits aren't minimized, because they take too much time with g2d
<destroyedlolo>
I removed IRQF_TRIGGER_FALLING, I can see ft5x driver is linked to /dev/input/event1 but cat on this file doesn't return anything :(
<rellla>
ssvb: the last commit in staging branch tries to optimize that issue. only blit if sth changed.
<ssvb>
rellla: can you document the test procedure somewhere in the wiki? or maybe it is already documented?
clonak has quit [Read error: Connection reset by peer]