ChanServ changed the topic of #linux-rockchip to: Rockchip development discussion | Wiki at | Logs at | ML at
Ancient has quit [Ping timeout: 240 seconds]
Ancient has joined #linux-rockchip
nighty has joined #linux-rockchip
Ancient has quit [Ping timeout: 240 seconds]
bbelos has quit [Ping timeout: 240 seconds]
bbelos has joined #linux-rockchip
Ancient has joined #linux-rockchip
muvlon has joined #linux-rockchip
wens has joined #linux-rockchip
wzyy2 has joined #linux-rockchip
cnxsoft has joined #linux-rockchip
muvlon has quit [Quit: Leaving]
Ancient|2 has joined #linux-rockchip
Ancient has quit [Ping timeout: 240 seconds]
muvlon has joined #linux-rockchip
vagrantc has quit [Quit: leaving]
wzyy2 has quit [Read error: Connection reset by peer]
wzyy2 has joined #linux-rockchip
cnxsoft1 has joined #linux-rockchip
cnxsoft has quit [Read error: Connection reset by peer]
cnxsoft1 is now known as cnxsoft
muvlon has quit [Ping timeout: 245 seconds]
muvlon has joined #linux-rockchip
<LongChair> stdint: morning, i'd have a few questions regarding drm if you have a min
<stdint> ok
<LongChair> ok first question is about vsync. the layer we add with drmModeAddFB2 that contains the frame. is it displayed when the function is called of after vsync ?
<LongChair> because it seemed that i couldn't use teh pageflip function on the fbid which is created by drmModeAddFB2
<LongChair> and depending on when that layer is displayed, you would need to call that function either before vsync or just right after it
<LongChair> and i have foudn really very little doc on the DRM API for userland
<LongChair> stdint: see what i mean ?
<stdint> LongChair, yes I want to implement pageflip function, but I have the same question
<stdint> lacks of document, I don't know the function could improve the performance
<stdint> you must wait the event
<stdint> you could read kmssink in gstreamer
<LongChair> i already did
<stdint> drmModeSetPlane() is the real function that put the buffer into screen
<stdint> drmModeAddFB2 just import the external buffer(even it is not) from outside
<LongChair> ok
<LongChair> so there is no way to have the buffer prepared and get it displayed when the page is flipped ?
<stdint> you could import them before using drmModeAddFB2
<stdint> but not much help actually
<LongChair> yeah usually with drm you would use "dumb" buffers, and you coudl page flip on those
<LongChair> which would make buffer switching perfect
<stdint> page flip have nothing to do with dumb buffer
<stdint> it is the driver level problem
<stdint> the mpp create buffer in the same way as create dumb buffer
<stdint> creating dumb buffer
<LongChair> drmModePageFlip allows to pas a dunb buffer id to which you want to flip
<LongChair> so if you create 2 dumb buffers, you can draw to one offscreen, and then flip on vsync
<LongChair> because drmmodepageflip will accept that dumb buffer id
<LongChair> stdint: but it doesn't seem to accept the buffer id returned by drmModeAddFB2
<stdint> I see how does the drmmodepageflip work, but I didn't see many driver have implemented that
<LongChair> mpv does
<LongChair> I got it changed a bit to handle rockchip
<stdint> no, I mean the kernel driver
<LongChair> ah
<LongChair> stdint: because even in gst code, we do drmModeAddFB2 , then drmModeSetPlane, and then wait vsync
<LongChair> if the layer is shown on teh drmModeSetPlane, then it means that the display timing is pretty bad
<LongChair> ok i'll make some more experimentation then
<LongChair> now i had a second question :)
<LongChair> when i get typically a 4K image (3840 x 2160)
<LongChair> and have a 1920x1080p display
<LongChair> it seems that the drmModeSetPlane requires at least the dest with and height to be aligned on 16
<LongChair> and typically 1080 isn't aligned on 16
<stdint> I am not sure whether the VOP has this limit
<stdint> but the vpu does
cnxsoft has quit [Remote host closed the connection]
<LongChair> stdint: i had artefacts when not aligning it
<LongChair> and it's not very nice, because, i needed to display the 1080p image o a 1072 or 1088 area .. so causes soem stretching
cnxsoft has joined #linux-rockchip
<stdint> it is not a big deal, what you need to do is set the width to original width but stride to aligned stride
<LongChair> where ?
<LongChair> because when i'm decoding a 2160p image it's no problem
<LongChair> 2160 is 16 aligned
<LongChair> so i get that 2160 buffer and want to display on a 1080p display
<LongChair> at that time i have no option to fiddle with aligned heights
<stdint> stride is set in drmModeAddFB2()
<stdint> drmModeSetPlane would set the dest width and src width
<LongChair> the thing is that in drmModeAddFB2, both height and with are 16 bytes aligned as i have 3840x2160 there
<LongChair> but drmModeSetPlane would then do the 1080p scaling
<stdint> I don't meet this problem
<LongChair> so i would do something like drmModeSetPlane(.... 0,0,3840,2160, 0 << 16, 0, << 16, 1920 << 16, 1080 << 16 )
<LongChair> and if the dest hieght is not 16 byes aligned .. i have artefacts on teh few first and few last lines
<LongChair> will fix if i replace 1080 with 1072 or 1088
stdint has quit [Quit: F*ck CCP]
wadim_ has joined #linux-rockchip
cnxsoft has quit [Quit: cnxsoft]
cnxsoft has joined #linux-rockchip
stdint has joined #linux-rockchip
matthias_bgg has joined #linux-rockchip
mrjay has joined #linux-rockchip
olerem has joined #linux-rockchip
holin has joined #linux-rockchip
<olerem> hallo all, i'm looking for information about availability of rockchip SoCs. Do this company provides long term support?
<mrjay> mmind00: i think it's a problem with unpacking ramdisk ... it might be a kernel issue or uboot issue
<holin> hi, would anyone here happen to know whether mainline kernel can be run on ASUS Chrombook C100 (Flip) with the google bootloader (not libreboot)? When I try anything else than chromium 3.14 kernel I only get a black screen and either the machine powers off or gets seemingly stuck.
<holin> chromium 3.14 would be fine, too, if i found a combination of mali / vpu drivers that would work with it and give acceleration under X11
eballetbo has joined #linux-rockchip
<phh> olerem: it depends on what you call long term support, but unless you pay a lot/buy a lot, the answer is no
<olerem> phh: i mean product life cycle
<olerem> so far i know, Allwinner has 3-4 Jears. iMX* ~10 Jears
<olerem> Some IT SoCs 10~15 Jears
<phh> I don't know about IoT like rk3036, but for MID don't expect more than allwinner I'd say
<olerem> phh: ok, thx
<wadim_> olerem: you should contact rockchip directly
<stdint> as long as the kernel is upstream, it is not a big deal
<stdint> iMX doesn't announce their new SoC frequently
<stdint> but rockchip release too frequently
<LongWork> stdint: what can RK3288 is suppsoed to decode up to in 4K ? 30 fps ? 60 fps
<stdint> 60fps
<LongWork> hmmm
<LongWork> i seemed to have trouble to keep the pace in some videos
<stdint> but for H.264, only 30fps
<LongWork> could you try that video with gst ?
<LongWork> see if that plays smoothly ?
<LongWork> because i didn't seem to be able to decode it fast enough
<wzyy2> I think most rockchip SOC have same life cycle with Allwinner.
<stdint> LongWork, that website is very slow
<LongWork> wierd .. was quite fast here
<stdint> there are some chips having very long life cycle
<LongWork> it has a lot of nice 4K materials to test with
<stdint> LongWork, yes, someone has given that to me
<stdint> I would suggest not to try those high bitrate video
<phh> erm, 60Mbps, high bitrate?
<LongWork> 60 mbps is not so high for a 4K video
<stdint> I remember that I was asked to support 3036 before
<stdint> 50Mpbs for AVC
<stdint> 80Mbps for HEVC
<LongWork> yeah so that video should play fine
<stdint> but 4k@25fps for AVC
<LongWork> at least it should decode fine, but i seemed to have like a very low decode rate with that one ... like maybe 10 fps ...
<stdint> it can't be
<stdint> back in half a hour
<LongWork> I could have something off in my code ... but works well with 24p video ...
<LongWork> like that one decodes fine :
mrjay has quit [Quit: Page closed]
beeno has joined #linux-rockchip
mfischer_ has quit [Ping timeout: 276 seconds]
daniels is now known as daniels_
daniels_ is now known as daniels
fischerm has joined #linux-rockchip
Omegamoon has joined #linux-rockchip
<stdint> LongChair, Oh, it is 10 bit
<stdint> only 50Mbps for HEVC high 10 profile
<LongWork> ah
<LongWork> could you still test that video and tell me how it plays on gst ?
Laibsch has joined #linux-rockchip
<LongWork> stdint: regarding tinker, would you advise to work with 4.4-release branch or the miniarm branch ?
muvlon has quit [Ping timeout: 245 seconds]
muvlon has joined #linux-rockchip
<ayaka> LongChair, miniarm branch I think
<ayaka> LongWork, sorry
<ayaka> could you upload that video into somewhere like google driver
nighty has quit [Quit: Disappears in a puff of smoke]
Laibsch has quit [Read error: Connection reset by peer]
wzyy2 has quit [Read error: Connection reset by peer]
<LongChair> ayaka: i'm not sure i'll have enough space there but will try
<LongChair> also one issue i have when connecting a 4K display atm is :
<LongChair> [ 39.732639] dwhdmi-rockchip ff980000.hdmi: Pixel clock 594000000 - unsupported by HDMI
<LongChair> would be nice to get fixed :)
<LongChair> ayaka: also does the miniarm branch had the vpu-service ?
Ancient|2 is now known as Ancient
<beeno> mac-l1, are you here?
ChatSharp has joined #linux-rockchip
ChatSharp has quit [K-Lined]
<beeno> Can anyone tell me where I can find the best content/docs for the vpu-api
<phh> which one? :D
paulk-collins has joined #linux-rockchip
<beeno> phh, you answering me? I'm looking at <rockchip/vpu_api.h> and macffplay source
<LongWork> beeno: i owuldn't expect too much docs :) there are a few API which are more self explanatory though
<LongWork> phh: seems like i got ffmpeg / mpv work together not too bad with drm / dmabuf
<LongWork> still need to solve the overlay issue
<phh> LongWork: yup i've read, that's cool
<phh> erm, I need my subtitles :P
<phh> LongWork: though I still don't have a proper kernel with vpu working
<LongWork> i used the 4.4-release
<LongWork> but i should give a try to the minarm branch
<phh> I have no board which works out of the box with 4.4
<LongWork> ah
<LongWork> 4K doesn't seem very well supported in terms of display though
<phh> well, wzyy2 did mention that rga is needed for 4k rendering
<LongWork> i still can't make sure that my stuff works right with a real 4K display mode as i can't propery switch to it
olerem has quit [Quit: WeeChat 1.6]
olerem has joined #linux-rockchip
olerem has left #linux-rockchip [#linux-rockchip]
<LongWork> i suppose rga should be underlying the drm API ...
<LongWork> redering 4K video in a 1080p mode works ... but soem dragons could show up when using a 4K mode
<beeno> Is the following command necessary or is it optional for performance reasons --> ctx->control(vpu_ctx,VPU_API_SET_VPUMEM_CONTEXT,(void*)vpu_display_pool);
mnemoc has joined #linux-rockchip
menomc has quit [Remote host closed the connection]
cnxsoft has quit [Quit: cnxsoft]
beeno has quit [Ping timeout: 260 seconds]
beeno has joined #linux-rockchip
holin has quit [Quit: leaving]
Laibsch has joined #linux-rockchip
jwerner has quit [Ping timeout: 240 seconds]
jwerner has joined #linux-rockchip
jwerner has quit [Ping timeout: 240 seconds]
jwerner has joined #linux-rockchip
JohnDoe_71Rus has joined #linux-rockchip
alex_ has quit [Ping timeout: 240 seconds]
alex_ has joined #linux-rockchip
wadim_ has quit [Remote host closed the connection]
wzyy2 has joined #linux-rockchip
athidhep has joined #linux-rockchip
jwerner has quit [Ping timeout: 252 seconds]
jwerner has joined #linux-rockchip
vagrantc has joined #linux-rockchip
Omegamoon has left #linux-rockchip [#linux-rockchip]
holin has joined #linux-rockchip
holin has quit [Quit: leaving]
beeno has quit [Ping timeout: 260 seconds]
wzyy2 has quit [Remote host closed the connection]
utente has joined #linux-rockchip
utente has left #linux-rockchip [#linux-rockchip]
mac-l1 has joined #linux-rockchip
mac-l1 has quit [Client Quit]
paulk-collins has quit [Remote host closed the connection]
Laibsch has quit [Read error: Connection reset by peer]
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria]
Laibsch has joined #linux-rockchip
leah2 has quit [Ping timeout: 245 seconds]
Laibsch has quit [Read error: Connection reset by peer]
Laibsch has joined #linux-rockchip
vagrantc has quit [Quit: leaving]