| 09:17 | deuter | ho |
| 10:18 | S_Hennig | Hi, there! Anybody here who has done FlexCAN with a HummingBoard? |
| 10:37 | rabeeh | S_Hennig: hi |
| 10:37 | rabeeh | there are few that has done flexcan |
| 10:37 | rabeeh | they use sd3_clk/cmd on the 26 pin header |
| 10:38 | rabeeh | kernel 3.0.35 based |
| 10:44 | S_Hennig | @rabeeh: Thanks for the info. Can you give some info on what / how to change the MUXer settings for the pins? |
| 11:00 | deuter | when openelec will switch to vivante 5.x from 3.10.53-ga ? |
| 11:04 | rabeeh | deuter: jnettlet is working on the port to LK 3.14.x |
| 11:05 | deuter | aha, ok, I ported it to 3.18 |
| 11:06 | deuter | rabeeh: do you have some plans for newer kernel like >=3.18 |
| 11:06 | fritsch | deuter: no plans at all yet ... |
| 11:06 | fritsch | deuter: and we don't know why we should |
| 11:07 | deuter | deuter: hehe, I have many problems with udoo's wifi on kernel up to 3.16 |
| 11:07 | deuter | fritsch: ^ |
| 11:07 | fritsch | best idea for wifi: don't use wifi |
| 11:07 | rabeeh | S_Henning: following are the two patches - |
| 11:08 | rabeeh | https://dl.dropboxusercontent.com/u/72661517/tbr/change-sd3_cmd_clk_to_be_flexcan.patch |
| 11:08 | deuter | fritsch: or dont use computer ;) |
| 11:08 | rabeeh | https://dl.dropboxusercontent.com/u/72661517/tbr/add-flexcan0-to-hummingboard.patch |
| 11:08 | rabeeh | the first changes the io mux and the second adds flexcan0 |
| 11:10 | S_Hennig | @rabeeh: Thanks a lot! |
| 11:13 | rabeeh | fritsch: good to see you are online |
| 11:13 | rabeeh | i had a question; i'm trying to visually see artifacts of the 23.976fps but i'm still unable |
| 11:14 | rabeeh | fritsch: i have a basic implementation that i'm getting 74.175mhz clock; and then tweaked hdmi cts/n dividers and i can play 23.976fps |
| 11:15 | rabeeh | but still; the main issue is that i can measure success / fail since i can't see the artifact first place |
| 11:15 | rabeeh | i'm using a PAL TV |
| 11:15 | rabeeh | any words of wisdom? |
| 11:29 | fritsch | rabeeh: if you press "o" |
| 11:29 | fritsch | what's the fps you get reported? |
| 11:29 | rabeeh | 23.976 |
| 11:29 | rabeeh | i'v tried the motion bar test patterns too; i can't see any micro stutters at all |
| 11:29 | rabeeh | and my eyes are good :) |
| 11:31 | fritsch | that would be very, very good news |
| 11:31 | fritsch | can you show me the patch? |
| 11:31 | fritsch | cause I am very interested how you told that to the driver |
| 11:32 | fritsch | as the spec mentions: no fractional modes at all |
| 11:33 | fritsch | and what's the value of fr? |
| 11:34 | fritsch | fps should slightly "change" from now and then |
| 11:34 | fritsch | i think you look at "fr" currntly |
| 11:34 | fritsch | cause fps only has "two digits" |
| 11:34 | fritsch | so you cannot see 23.976 there |
| 11:50 | rabeeh | sorry; got distracted |
| 11:51 | rabeeh | fritsch: i'v measured the pixel clock and it's ~74.175mhz vs. ~74.25mhz |
| 11:52 | rabeeh | so i'm able to change the video pll pixel clock for sure; afterwards i'v feed the new n/cts dividors to get correct audio clock |
| 11:53 | rabeeh | but as i said; 23.976fps plays without issue; but i can't really tell if it's good or bad since even running 23.976fps on 24fps configuration i should be seeing one frame repetition every ~16 seconds |
| 11:53 | rabeeh | which i'm not seeing |
| 11:53 | rabeeh | so i don't really have a benchmark; the only thing i can think of is - |
| 11:53 | rabeeh | 1. my eyes are tired looking at tv continuous time and my brain can't comprehend micro stutters anymore |
| 11:54 | rabeeh | 2. the TV is hiding all of this and it's doing frame rate change without asking me |
| 11:59 | fritsch | rabeeh: the codec screen is enough |
| 12:00 | fritsch | rabeeh: we measure the fps in there |
| 12:00 | fritsch | on current version without your patches the fps is nailed to 24.0 |
| 12:00 | fritsch | which is the problem |
| 12:00 | fritsch | when your patch works this fps value should be something like 23.98 |
| 12:01 | fritsch | problem is not that every 40 seconds a frame is skipped |
| 12:01 | fritsch | but rather the audio clock |
| 12:01 | fritsch | which should drop a packet every minute |
| 12:01 | fritsch | cause the video is played "too fast" |
| 12:01 | fritsch | in 24fps instead of 23.976 on average |
| 12:01 | fritsch | rabeeh: so, play some file, press "print screen" and show me the screenshot, please |
| 12:01 | rabeeh | fritsch: hmm.... unclear |
| 12:02 | fritsch | fps we measure |
| 12:02 | fritsch | while playing |
| 12:02 | fritsch | -> screenshot will tell |
| 12:02 | fritsch | -> afk for lunch |
| 12:02 | rabeeh | if i press 'o' then it shows 23.976fps |
| 12:03 | fritsch | it does not have 3 digits |
| 12:03 | fritsch | in the fps row |
| 12:03 | rabeeh | this was for ages like this; this only reports what the content is and unrelated to the actual rendering |
| 12:03 | fritsch | give me a screenshot please |
| 12:03 | rabeeh | oh; sorry |
| 12:03 | fritsch | naaa - you watch the wrong row |
| 12:03 | rabeeh | 23.98 |
| 12:03 | fritsch | :-) |
| 12:03 | fritsch | that's very good new |
| 12:03 | fritsch | s |
| 12:03 | fritsch | now compare with older |
| 12:03 | fritsch | you see 2w4.0 |
| 12:03 | rabeeh | i need to reinstall to check :) |
| 12:04 | rabeeh | go grab lunch; will have results in 15m |
| 12:04 | rabeeh | :) |
| 12:11 | vpeter | and then there are posts on forum that solidrun doesn't do much for customers :) |
| 12:12 | vpeter | ^^^^ this is obviously an opposite of that. |
| 12:15 | rabeeh | vpeter: we are good at engineering; we are lousy at public relations :) |
| 12:15 | rabeeh | that i can guarantee |
| 12:16 | vpeter | hidden behing a monitor :) |
| 12:18 | rabeeh | well... sort of |
| 12:18 | rabeeh | anyhow; it shows mostly 24fps on OE 5.0.2 |
| 12:19 | rabeeh | it jumps between 23.2 to 24.15 time to time |
| 12:24 | rabeeh | actually i would say that 50% it's 24fps and 50% it's 23.98fps; now lets see if fritsch can explain this :) |
| 12:40 | fritsch | rabeeh: yeah - that's fine |
| 12:40 | fritsch | rabeeh: no content is perfect :-) |
| 12:40 | fritsch | do you have "Sync Playback to display" enabled? |
| 12:40 | fritsch | do that, please |
| 12:41 | fritsch | method video clock |
| 12:41 | fritsch | and see if that changes anything |
| 12:43 | rabeeh | i'v switched back to my hacked version of OE |
| 12:43 | rabeeh | it renders fixed 23.98 for ~15 seconds then drops for half a second for 23.95 and goes back |
| 12:44 | fritsch | cool - any skips? |
| 12:44 | rabeeh | Sync Playback to display is disabled |
| 12:44 | rabeeh | shall i enable? |
| 12:44 | fritsch | yes, with method video clock |
| 12:44 | fritsch | it does not have a "reeal video clock" |
| 12:44 | fritsch | so it's sw integrated |
| 12:45 | rabeeh | video clock with audio resample? |
| 12:45 | rabeeh | or drop/dupe audio? |
| 12:45 | fritsch | use audio resample |
| 12:45 | fritsch | that will automatically resample for non passtrhough audio |
| 12:45 | fritsch | and passthough audio will be dropped / duped |
| 12:45 | fritsch | as we cannot resample that (obviously) |
| 12:46 | fritsch | rabeeh: btw. can you please share the 352Mhz patch, so that the lowest state can be reached again? |
| 12:46 | rabeeh | jnettlet is still baking that |
| 12:46 | fritsch | oki |
| 12:46 | fritsch | thx |
| 12:46 | fritsch | so - what is your experience with the sync playback enabled? concerning fps? |
| 12:46 | jnettlet | it will be in the next kernel release |
| 12:46 | rabeeh | ok; current status 23.976fps content, 5.1 passthrough |
| 12:47 | rabeeh | fps:23.98 |
| 12:47 | rabeeh | skip:5 |
| 12:48 | rabeeh | how much should i really wait to check if there are skips? |
| 12:49 | rabeeh | fritsch: want to try? |
| 12:50 | rabeeh | after 4 minutes skip:6 |
| 12:56 | fritsch | if the fps is not nailed to 24.0 |
| 12:56 | fritsch | it's a real, real improvement |
| 12:56 | fritsch | rabeeh: if you post the kernel patch - I will apply it |
| 12:56 | fritsch | and do testbuilds |
| 12:56 | fritsch | if currently have the wolfgar / smallint rework compiled |
| 12:57 | fritsch | with the fixes from this morning |
| 12:57 | fritsch | 1080p50 |
| 12:57 | fritsch | now really works |
| 12:57 | fritsch | at least on the ddr3 64 bit cuboxes |
| 12:57 | fritsch | the i2 has no chance sadly |
| 12:59 | fritsch | but I want to see it first |
| 13:00 | fritsch | for wetek we did it in a special way |
| 13:00 | fritsch | ask the video for the fps |
| 13:00 | fritsch | and set the video clock acoordingly |
| 13:00 | fritsch | so the the kernel only needs to maintain the normal 24 mode it gets |
| 13:00 | fritsch | but on video start we trick it |
| 13:00 | fritsch | by changing the clocks |
| 13:00 | fritsch | disadvantage: only works if fps are known correctly |
| 13:01 | fritsch | vfr videos or files where fps needs to be estimated are a bit harder to do |
| 13:09 | rabeeh | i2 is 64bit ddr too :) |
| 13:09 | fritsch | from your homepage |
| 13:09 | fritsch | it tells differently |
| 13:10 | rabeeh | fritsch: as i posted on the forums; on the solo i can play 1080p60 |
| 13:10 | fritsch | http://www.solid-run.com/products/cubox-i-mini-computer/cubox-i-specifications/ |
| 13:10 | fritsch | yeeah and the peopz that tried to verify with your very same image could not do it :-) |
| 13:10 | fritsch | without hard freezing |
| 13:11 | fritsch | see: i.mx6 dual lite |
| 13:11 | rabeeh | oh; a bug |
| 13:11 | fritsch | 32 bit 512 MB 800 Mbps |
| 13:11 | rabeeh | 64bit 1GByte 800 Mbps |
| 13:11 | fritsch | that's the i2ex |
| 13:11 | fritsch | but still theoretical throughput is at 80% |
| 13:11 | fritsch | in comparison with the quads |
| 13:12 | fritsch | with the new approach the memory is the bottleneck now |
| 13:12 | rabeeh | that's also wrong |
| 13:12 | rabeeh | i2ex is 64bit 1GByte 1066 Mbps |
| 13:13 | fritsch | okay, then go - fix it :-) |
| 13:13 | fritsch | hehe |
| 13:13 | rabeeh | they lately removed the i1 from the site and probably things shifted |
| 13:13 | django | Still selling the MicroSOM though. |
| 13:13 | fritsch | ah - wolfgar mentioned something else |
| 13:13 | fritsch | concerning the architecture |
| 13:13 | fritsch | of the GC880 systems |
| 13:13 | fritsch | concerning bus usage |
| 13:14 | fritsch | but, yeah we get off topic |
| 13:14 | fritsch | would like to see the 23.976 fps patch wip |
| 13:14 | fritsch | to see if I can learn something from your hacking skills :-) |
| 13:14 | fritsch | as I could not get it going |
| 13:14 | fritsch | hehe |
| 13:14 | django | rabeeh: Do you know if there's any news about the 352MHz VPU patch? |
| 13:15 | rabeeh | django: in what means? |
| 13:15 | rabeeh | fritsch: you need to be hardcore i.mx6 (and with a scope) to make 23.976 speakup |
| 13:15 | rabeeh | django: it's working; the only pitfall is that it removes a lower freq from the processor freq scaling algorithm |
| 13:15 | django | rabeeh: there was talk about a patch that would allow for higher spec hardware video encoding. |
| 13:15 | rabeeh | jnettlet is about to fix it |
| 13:16 | rabeeh | encoding? |
| 13:16 | django | yes |
| 13:16 | rabeeh | which resolution? |
| 13:16 | rabeeh | and bitrate? |
| 13:16 | django | Well, as I understood now it's possible to do 1080p30 |
| 13:16 | django | Don't know what bitrate |
| 13:16 | rabeeh | i think i.mx6 is limited to 1080p30 h.264 encoding; i'm not sure about getting it higher with vpu@352 |
| 13:17 | fritsch | django: here: https://github.com/fritsch/OpenELEC.tv/commits/imx-master |
| 13:17 | fritsch | django: if you need a build with all of the current rework, I can link that too |
| 13:18 | fritsch | rabeeh: yeah - that's why I am curious (concerning the 23.976) |
| 13:18 | django | fritsch: I need it in on a lightweight distribution with gstreamer |
| 13:18 | fritsch | django: https://github.com/fritsch/OpenELEC.tv/commit/d7ca9a0f5364897e533c6e437ff2d54bc1cd0a86 |
| 13:18 | django | openelec isn't really suited |
| 13:18 | fritsch | just pick it |
| 13:18 | fritsch | all there |
| 13:18 | fritsch | https://github.com/fritsch/OpenELEC.tv/commit/d7ca9a0f5364897e533c6e437ff2d54bc1cd0a86.patch <- |
| 13:19 | fritsch | one part is kernel config |
| 13:19 | fritsch | other part is patch |
| 13:19 | django | Yeah, but I think they're being worked into the new kernel |
| 13:19 | django | So I'm just waiting. |
| 13:19 | fritsch | nope, I don't think so |
| 13:19 | django | I think jnettlet talked about it a while back |
| 13:19 | fritsch | jnettlet is working out the issue that the lowest freq will vanish |
| 13:19 | fritsch | when the 352 is enabled |
| 13:19 | rabeeh | django: how light? |
| 13:19 | rabeeh | yocto is light enough? |
| 13:20 | django | Well, not light specifically, but headless |
| 13:20 | django | like deb/suse/ubuntu |
| 13:20 | rabeeh | why don't you try the obs debs / suse / ubuntus |
| 13:20 | rabeeh | ? |
| 13:20 | rabeeh | they all have gst stuff; and can run headless |
| 13:20 | rabeeh | my guess is that you are looking for transcoding? |
| 13:20 | django | Yeah |
| 13:21 | django | I have to be able to do 1080p50, but will mostly do 720p50/25 and 720i50/25 |
| 13:21 | fritsch | django: encoding? |
| 13:21 | fritsch | live? |
| 13:21 | django | Yeah |
| 13:21 | rabeeh | django: which processor? |
| 13:21 | django | Yeah |
| 13:21 | fritsch | ouh |
| 13:21 | django | i4Pro |
| 13:21 | fritsch | which h264 settings? |
| 13:22 | fritsch | medium level 4.1@high? |
| 13:22 | fritsch | :-) |
| 13:22 | fritsch | that's really doable? |
| 13:22 | django | Don't know. |
| 13:22 | rabeeh | i would guess base profile |
| 13:22 | django | Will have to test later |
| 13:23 | django | Have a cluster which is doing transport streams with veryslow 4.1@high |
| 13:24 | fritsch | nice |
| 13:24 | fritsch | vaapi encode |
| 13:24 | fritsch | also can do that |
| 13:24 | fritsch | but is in a very early satge |
| 13:24 | fritsch | stage |
| 13:26 | django | That's decoding? |
| 13:26 | fritsch | nope encoding |
| 13:26 | fritsch | libyami |
| 13:26 | fritsch | or something |
| 13:26 | fritsch | during 0.35 vaapi the encoding stuff was integrated |
| 13:26 | fritsch | rabeeh: i need to go, when you have something kernel wise for testing - ping me |
| 13:27 | fritsch | rabeeh: then I will have a look |
| 13:41 | MarcusVinter | What res is u-boot? 1024x768? |
| 15:47 | django | I think it's 720 |
| 15:47 | django | 1280x720 |
| 15:48 | MarcusVinter | thank you, much appreciated. |
| 17:38 | django | It might be adaptive though, I'm not entirely sure. |
| 18:09 | MarcusVinter | django fyi it was 1024x768 :P |
| 18:47 | MarcusVinter | Just a quick question again, is there a way to remove the build name from u-boot start screen? the "U-Boot 2013.10-rc4-dirty (build date and time) " line? |
| 18:48 | MarcusVinter | I'm in mx6_cubox-i.h and cant seem to see it |
| 21:45 | cbxbiker61 | finally.... I've been working for 5 days to make the transition from kde4 to kde5 |
| 22:02 | dz0ny | cbxbiker61: went from unity to gnome3 in 3h :> |
| 22:04 | cbxbiker61 | heh, yeah but i build from scratch, and there were tons of new package build scripts i had to create/test |
| 22:04 | cbxbiker61 | i like to have "full" control of my systems ;) |
| 22:06 | cbxbiker61 | i should be set for simple upgrades the next 4 or 5 years |
| 22:11 | dz0ny | :) |