- mjw (QUIT: Quit: Leaving) (~mjw_@herd.wildebeest.org) | 00:11 | |
+ specing_ (~specing@unaffiliated/specing) | 02:32 | |
- specing (QUIT: Ping timeout: 265 seconds) (~specing@unaffiliated/specing) | 02:35 | |
* specing_ -> specing | 02:35 | |
- reverbdog (QUIT: Ping timeout: 240 seconds) (~thor@51.158.168.14) | 03:20 | |
+ reverbdog (~thor@51.158.168.14) | 03:23 | |
- erlehmann (QUIT: Ping timeout: 240 seconds) (~erlehmann@dynamic-046-114-038-119.46.114.pool.telefonica.de) | 03:24 | |
- reverbdog (QUIT: Read error: Connection reset by peer) (~thor@51.158.168.14) | 03:36 | |
+ reverbdog (~thor@51.158.168.14) | 03:36 | |
+ erlehmann (~erlehmann@dynamic-046-114-032-005.46.114.pool.telefonica.de) | 03:37 | |
+ B[] (~Thunderbi@219-89-228-78-adsl.sparkbb.co.nz) | 06:20 | |
+ rasmus-mk (~rasmus@c83-253-223-217.bredband.comhem.se) | 07:38 | |
+ odnes (~odnes@109-178-131-92.pat.ren.cosmote.net) | 08:18 | |
- Asmadeus (QUIT: Ping timeout: 258 seconds) (~asmadeus@unaffiliated/asmadeus) | 09:20 | |
+ Asmadeus (~asmadeus@unaffiliated/asmadeus) | 09:26 | |
- odnes (QUIT: Remote host closed the connection) (~odnes@109-178-131-92.pat.ren.cosmote.net) | 09:46 | |
+ odnes (~odnes@109-178-131-92.pat.ren.cosmote.net) | 09:47 | |
+ mmicko (~mmicko@hhsrv1.vastheman.com) | 10:10 | |
- odnes (QUIT: Ping timeout: 260 seconds) (~odnes@109-178-131-92.pat.ren.cosmote.net) | 11:03 | |
- jvalleroy (QUIT: Quit: http://quassel-irc.org - Chat comfortably. Anywhere.) (~quassel@unaffiliated/jvalleroy) | 11:36 | |
+ jvalleroy (~quassel@li1266-111.members.linode.com) | 11:37 | |
- jvalleroy (QUIT: Changing host) (~quassel@li1266-111.members.linode.com) | 11:37 | |
+ jvalleroy (~quassel@unaffiliated/jvalleroy) | 11:37 | |
+ odnes (~odnes@109-178-131-92.pat.ren.cosmote.net) | 14:06 | |
+ odnes_ (~odnes@109-178-131-92.pat.ren.cosmote.net) | 14:12 | |
- odnes (QUIT: Read error: Connection reset by peer) (~odnes@109-178-131-92.pat.ren.cosmote.net) | 14:13 | |
+ specing_ (~specing@unaffiliated/specing) | 14:32 | |
- specing (QUIT: Ping timeout: 240 seconds) (~specing@unaffiliated/specing) | 14:35 | |
* specing_ -> specing | 14:35 | |
- q66 (QUIT: Quit: q66) (~q66@enlightenment/developer/q66) | 14:35 | |
+ q66 (~q66@enlightenment/developer/q66) | 14:36 | |
- odnes_ (QUIT: Ping timeout: 246 seconds) (~odnes@109-178-131-92.pat.ren.cosmote.net) | 14:38 | |
devyl | Hm moonlight (qt nvidia game streaming app) doesnt recognize any h264/h265 hardware decoder. The i.mx8 vpu has one. Isnt it implemented on debian jet or do i miss something (config, environment var)? Moonlight itself runs nicely with my elite controller x) but with sw decoding 720p/30fps/~20mbps is the max setting for smooth gaming. ;) | 15:23 |
---|---|---|
_Bnu | Ah yeah, something isn't quite working with the hardware decoding at the moment, though I don't know exactly what. | 15:26 |
_Bnu | Not sure if it's VAAPI or that other thing or exactly where the problem lies, but it seems to affect the web browsers as well. | 15:27 |
mntmn | h265 is not really there yet | 15:43 |
mntmn | (work in progress) | 15:43 |
mntmn | h264, hantro, is in the kernel | 15:43 |
mntmn | but it is implemented as v4l2 m2m api which is kind of new | 15:44 |
mntmn | it is supported by newer gstreamer versions (i doubt that by the ones in debian) | 15:44 |
mntmn | you'll have to do some research on this devyl, i can't help you with a recipe at the moment. | 15:44 |
mntmn | devyl: ezequielg here might know more about the current status | 15:45 |
ezequielg | hey guys | 15:47 |
ezequielg | h264 "stable" support is merged in the kernel and gstreamer (i think it's queued for 1.20). | 15:47 |
ezequielg | ffmpeg patches are either merged or being discussed. | 15:47 |
ezequielg | mpeg2, vp8 and h265 uapi is in "unstable" api form. | 15:48 |
ezequielg | vp8 and mpeg2 apis will be stable very soon. | 15:48 |
ezequielg | imx8mq currently supports mpeg2, vp8 and h264. | 15:48 |
ezequielg | imx8mq h265 support is posted in ML see https://www.collabora.com/news-and-blog/blog/2021/02/17/adding-hevc-support-for-imx-8m/. | 15:49 |
ezequielg | the series has evolved and i think it might land for 5.13. | 15:49 |
ezequielg | we are also working on 10-bit support, through raster-scan P010 and compressed reference buffers. | 15:49 |
ezequielg | we are not there yet, but this is work towards 4K, via compressed buffers in the dcss. | 15:50 |
ezequielg | perhaps towards end of this year everything will land? | 15:50 |
ezequielg | we are working on VP9 in parallel and on rockchip SoCs, and encoding... so: many fronts, few people :) | 15:51 |
ezequielg | mntmn: if you can collect the top features your users want, i.e. firefox? chromium? .. 4K? which codec? I'm interested in that kind of stuff to plan our R&D. | 15:52 |
ezequielg | being spreaded very thinly means we must pick the battles carefully, and sadly some features are sacrificied. | 15:52 |
ezequielg | devyl: hopefully this answers your questions :-) | 15:55 |
mntmn | ezequielg: great answers, thank you. i'll do a bit of an analysis of where h264 (at first) decoding makes the most impact | 16:18 |
reverbdog | devyl: i'm also interested in getting moonlight working on the reform! | 17:05 |
reverbdog | can it use v4l2-m2m h264 through ffmpeg? i don't have a reform yet. but when mine arrives i'll be trying with you | 17:09 |
mntmn | what is moonlight exactly? some kind of remote play thing? | 17:13 |
mntmn | i see https://moonlight-stream.org/ | 17:13 |
mntmn | but i need my own pc for that, yeah? | 17:13 |
reverbdog | mntmn: yep. i have an old windows 10 pc that's dedicated to gaming and it's mostly headless. i use moonlight to stream GPU-intensive games to my laptop | 17:17 |
mntmn | that's cool | 17:17 |
mntmn | is there a client like that for ps4? | 17:17 |
reverbdog | but moonlight only works with nvidia GPUs | 17:17 |
mntmn | ok | 17:17 |
mntmn | sorry i meant a host | 17:17 |
mntmn | i know it has remote play but i don't know if there are any foss clients for ps4 remote play. | 17:17 |
reverbdog | interesting, no idea | 17:18 |
reverbdog | didn't know ps4 had remote play | 17:18 |
mmicko | mntmn: There is https://git.sr.ht/~thestr4ng3r/chiaki | 17:18 |
mmicko | but did not test that to be honest | 17:19 |
reverbdog | devyl: maybe your build of ffmpeg doesn't include support for v4l2-m2m, but from what i'm reading support is available? | 17:19 |
mntmn | mmicko: oh that's cool. i'll try that (i have a ps4, but no gaming pc) | 17:20 |
mmicko | https://git.sr.ht/~thestr4ng3r/chiaki/refs/v2.1.1 contains AppImage so should work from any Linux machine | 17:20 |
mmicko | x64 not all :) | 17:24 |
+ odnes (~odnes@109-178-131-92.pat.ren.cosmote.net) | 17:39 | |
+ reform32104 (~matthias@ipb21a7ce2.dynamic.kabel-deutschland.de) | 17:41 | |
reform32104 | EXIT | 17:42 |
reform32104 | exit | 17:42 |
- reform32104 (QUIT: Client Quit) (~matthias@ipb21a7ce2.dynamic.kabel-deutschland.de) | 17:42 | |
devyl | thanks for the answers... :) thats what i use mntmn : https://flathub.org/apps/details/com.moonlight_stream.Moonlight | 17:44 |
mntmn | i just tried chiaki | 17:44 |
mntmn | (built from source on reform) | 17:44 |
mntmn | so it works! but only with LIBGL_ALWAYS_SOFTWARE i see the stream | 17:45 |
mntmn | something goes wrong when displaying the stream through GL | 17:45 |
devyl | i play cyberpunk 2077 (...) though moonlight on my macbook... and hopefully soon on my reform ;) | 17:46 |
mntmn | haha i'm also testing cp2077 right now... but the mystery ps4 version | 17:46 |
devyl | sackboy adventure is a great fun on ps4 ;) | 17:47 |
mntmn | works also with ps4 controller connected to reform | 17:48 |
- DanC (QUIT: Quit: ZNC 1.6.6+deb1ubuntu0.2 - http://znc.in) (~DanC@136.33.249.0) | 17:49 | |
mntmn | would be good to fix GL though... | 17:49 |
+ dckc (~DanC@2605:a601:ac60:1000:dc7:8456:eb59:d948) | 17:50 | |
reverbdog | hm never heard of flathub. so that's a repository of statically linked linux binaries? | 17:50 |
devyl | flatpak :) like snap on ubuntu | 17:51 |
reverbdog | it's kinda hard to see details of their build | 17:52 |
devyl | more or less a second distribution independed repository ... | 17:52 |
reverbdog | for moonlight | 17:52 |
devyl | afk ;) | 17:52 |
mntmn | ah chiaki wants to use GLSL 1.50 | 17:55 |
mntmn | that's the problem | 17:55 |
reverbdog | ah i found it https://github.com/flathub/com.moonlight_stream.Moonlight/blob/master/com.moonlight_stream.Moonlight.json | 17:56 |
reverbdog | if your build uses that configuration, then ffmpeg v4l2-m2m support isn't the issue. although it appears that it was only toggled 13 days ago | 17:57 |
reverbdog | i wonder how moonlight negotiates codecs | 17:58 |
mntmn | from which version on is v4l2-m2m support in ffmpeg? | 18:00 |
mntmn | hmm. testing: ffplay -vcodec h264_v4l2m2m ~/Videos/lumixtest1.mp4 | 18:05 |
mntmn | [h264_v4l2m2m @ 0x7f483489f0] Could not find a valid device | 18:05 |
mntmn | [h264_v4l2m2m @ 0x7f483489f0] can't configure decoder | 18:05 |
reverbdog | looks like... 3.4? | 18:06 |
mntmn | hmm strange that it doesn't work then. | 18:07 |
reverbdog | permissions thing? | 18:08 |
mntmn | ezequielg: is there any more work/patch needed to allow ffmpeg/ffplay to use h264_v4l2m2m on imx8mq? | 18:08 |
mntmn | reverbdog: nope | 18:08 |
+ Neelfyn (uid180106@gateway/web/irccloud.com/x-jxvyuufqsexfcpuw) | 18:44 | |
ezequielg | I would have to check that. Is it merged upstream? | 19:21 |
mntmn | afaik yes | 19:21 |
ezequielg | Keep in mind the hw decoder mode is called something like "drm" which can be confusing? | 19:22 |
mntmn | ezequielg: if you google h264_v4l2m2m you will find a lot of examples of people trying it on various platforms | 19:22 |
mntmn | with ffplay | 19:22 |
ezequielg | I can find you my notes tomorrow or Monday | 19:28 |
mntmn | cool | 19:29 |
ezequielg | But iirc it's something like ffmpeg -hwaccel drm | 19:29 |
mntmn | ohh ok | 19:29 |
mntmn | that does not appear to be in the ffmpeg version on debian | 19:31 |
mntmn | but also the gstreamer-1.0 in debian sid: WARNING: erroneous pipeline: no element "v4l2h264dec" | 19:32 |
_Bnu | I think I did try VLC already with the "drm" hardware decoding and it didn't appear to work. | 19:44 |
mntmn | _Bnu: what does vlc use behind the scenes? ffmpeg or something unique? | 19:49 |
- adjtm_ (QUIT: Remote host closed the connection) (~adjtm@188.26.223.61) | 19:53 | |
+ adjtm_ (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 19:54 | |
- adjtm_ (QUIT: Remote host closed the connection) (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 19:56 | |
+ adjtm_ (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 19:57 | |
_Bnu | I don't actually recall for certain, but I believe it uses the same code base, ie libavcodec or whatever it's called these days. | 20:00 |
- adjtm_ (QUIT: Remote host closed the connection) (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 20:06 | |
+ adjtm_ (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 20:06 | |
+ adjtm (~adjtm@188.26.223.61) | 20:11 | |
- adjtm_ (QUIT: Ping timeout: 240 seconds) (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 20:12 | |
- adjtm (QUIT: Remote host closed the connection) (~adjtm@188.26.223.61) | 20:13 | |
+ adjtm (~adjtm@188.26.223.61) | 20:14 | |
mntmn | _Bnu: understood | 20:19 |
- B[] (QUIT: Ping timeout: 240 seconds) (~Thunderbi@219-89-228-78-adsl.sparkbb.co.nz) | 20:23 | |
mntmn | ok chiaki works with export MESA_GLSL_VERSION_OVERRIDE=150 | 20:32 |
+ reform13803 (~root@2a01:e0a:1ef:7180:219:b8ff:fe08:dddf) | 20:42 | |
mntmn | hi reform13803 | 20:42 |
+ S0rin (~S0rin@ns3074960.ip-217-182-174.eu) | 20:43 | |
S0rin | Hello, I have my reform showing me blinking battery even after a night of charge is there something I can do to fix it ? | 20:44 |
mntmn | S0rin: welcome, check this out https://community.mnt.re/t/how-long-for-battery-controller-to-learn-battery-capacity/129/2 | 20:46 |
reform13803 | quit | 20:46 |
mntmn | reform13803: you have to type /quit | 20:46 |
- reform13803 (QUIT: Client Quit) (~root@2a01:e0a:1ef:7180:219:b8ff:fe08:dddf) | 20:46 | |
mntmn | S0rin: use it on battery for a few hours and charge full again. | 20:47 |
- adjtm (QUIT: Remote host closed the connection) (~adjtm@188.26.223.61) | 20:47 | |
+ adjtm (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 20:48 | |
S0rin | mntmn: Sorry it was me on the reform (reform13803) | 20:48 |
S0rin | mntmn: Yes I will let run on battery for a while and then charge again thank you | 20:49 |
mntmn | ok! | 20:50 |
+ adjtm_ (~adjtm@188.26.223.61) | 21:12 | |
- adjtm (QUIT: Ping timeout: 246 seconds) (~adjtm@2a0c:5a80:1402:200:248c:77bd:484b:2528) | 21:13 | |
S0rin | I've seen in the handbook that it is possible to change the CPU soc, is there a page with links where I can see details ? | 21:23 |
- odnes (QUIT: Remote host closed the connection) (~odnes@109-178-131-92.pat.ren.cosmote.net) | 21:23 | |
mntmn | S0rin: what's your question as to the soc? | 21:48 |
S0rin | mntmn: I would like to know the price and spec of alternative CPU | 21:50 |
mntmn | S0rin: there is no price yet, it is still in development. | 21:50 |
mntmn | S0rin: the first prototypes are just being assembled. the module is based on NXP LS1028A | 21:51 |
S0rin | mntmn: Yes I have seen that in the handbook can't find it online so maybe I am a bit early on this | 21:52 |
mntmn | S0rin: as i said, the module is still in development. | 21:52 |
mntmn | S0rin: this is the chip https://www.nxp.com/products/processors-and-microcontrollers/arm-processors/layerscape-processors/layerscape-1028a-applications-processor:LS1028A | 21:52 |
S0rin | mntmn: I get it thank you for the precision ;) | 21:52 |
mntmn | the module is interesting if you want higher single core performance and 8 or 16GB ram, and more pcie address space (allowing for eGPU). but it will be quite expensive. | 21:54 |
S0rin | mntmn: yes yes I can guess that but I wonder the price range | 21:55 |
S0rin | And yes more RAM can be a strong argument | 21:55 |
mntmn | i suspect something between 500 and 700 euro | 21:55 |
mntmn | but i really don't know the real numbers yet | 21:55 |
S0rin | mntmn: Ok and what would be the cost of the actual CPU if buy single online (to compare) | 21:57 |
mntmn | idk maybe 150 euro? the expensive part is the 16GB ram i guess | 21:58 |
mntmn | ah no the chip is around 70 EUR. depending on volum | 21:59 |
mntmn | +e | 21:59 |
S0rin | mntmn: hum ok so maybe an upgrade version of the actual with just more ram can be more well priced ? | 21:59 |
S0rin | mntmn: Yes sure I understand | 21:59 |
mntmn | really, i don't know yet. | 22:00 |
S0rin | mntmn: that's ok, thank you for the informations :) | 22:01 |
mntmn | i think there will be more info in around 2-3 months | 22:01 |
mntmn | the module assembly is delayed @ rbz because of component shortage | 22:02 |
mntmn | everything is pretty difficult right now | 22:02 |
doppler | yeah, this chip shortage is a pain in the neck | 22:02 |
mntmn | i'm really happy that we early on bought everything for 450 reforms | 22:03 |
doppler | good call | 22:03 |
mntmn | so we will be able to build enough for all the campaign + after campaign sales plus some extra | 22:03 |
adjtm_ | mntmn, the problem of Layerscape is that they are too focused on networking devices, even the parts with fewer interfaces are too much for everything except routers | 22:03 |
adjtm_ | and I suspect that network interface contribute a lot to power and price | 22:03 |
mntmn | adjtm_: well as i said, the memory is the expensive part here | 22:03 |
mntmn | and the board also needs a more expensive process | 22:04 |
adjtm_ | is LS1028A's gpu vivante? | 22:04 |
mntmn | yes | 22:06 |
mntmn | i think the same as in imx8m | 22:06 |
S0rin | It this one https://cnx-software.ru/2019/05/07/miriac-mpx-ls1028a-nxp-qoriq-ls1028a-som-prednaznachen-dlya-prilozhenij-tsn/ | 22:07 |
mntmn | S0rin: no | 22:08 |
mntmn | S0rin: i mean, it is not that module. it is the chip, though. i can't read russian though. | 22:08 |
mntmn | that module comes only with 4GB ram max | 22:09 |
- rasmus-mk (QUIT: Quit: Konversation terminated!) (~rasmus@c83-253-223-217.bredband.comhem.se) | 22:10 | |
S0rin | mntmn: I can't read russian too, but found that in duckduck go images | 22:10 |
rvense | mntmn: was someone working on an fpga-based module? | 22:10 |
mntmn | rvense: yes, me | 22:11 |
mntmn | kintex-7 | 22:11 |
rvense | ah, cool | 22:11 |
rvense | how far along is that? | 22:11 |
rvense | schematics.. layout... fully functional amiga laptop? | 22:13 |
mntmn | i have some basic schematics and most of the chips selected, but not completed power stuff | 22:13 |
rvense | ah, ok | 22:13 |
mntmn | it will take a while. most time i have to spend on building reforms now | 22:13 |
rvense | yeah, of course. it's a niche within a niche i'm sure. | 22:14 |
mntmn | meanwhile the imx8mq runs amiberry quite well! | 22:14 |
rvense | nice | 22:14 |
mntmn | i have amiga emulation, classic mac, playstation 1, c64, doxbox, even windows 10 arm on my reform ;) | 22:15 |
mntmn | dosbox | 22:15 |
S0rin | mntmn: That's cool, and let's say qemu i386 / amd64 for let's small dev vm did someone test that | 22:17 |
S0rin | Like most docker image won't work right | 22:17 |
mntmn | qemu-user-static works ok for x86 but a full blown vm is still too slow | 22:18 |
mntmn | exagear will probably work better but i haven't tested the huawei (?) version yet | 22:18 |
mntmn | it's also all proprietary stuff | 22:18 |
mntmn | i think faster x86 emulation for arm is still in development | 22:19 |
mntmn | https://github.com/FEX-Emu/FEX | 22:20 |
mntmn | i think this is the most interesting one | 22:21 |
S0rin | mntmn: it's like wine for x86 | 22:22 |
S0rin | it | 22:22 |
S0rin | it's cool | 22:22 |
mntmn | it's like qemu-user-static | 22:23 |
mntmn | wine is a windows emulator | 22:23 |
S0rin | mntmn: Yes sure bad comparison | 22:28 |
- terpri_ (QUIT: Remote host closed the connection) (~terpri@2603-6081-2241-0752-96b8-6dff-feb3-87d3.res6.spectrum.com) | 23:24 | |
+ terpri_ (~terpri@2603-6081-2241-0752-96b8-6dff-feb3-87d3.res6.spectrum.com) | 23:24 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!