Mister_Magister | mal: could you help me with fixing ui performance and/or getting surfaceflinger rendering? | 08:21 |
---|---|---|
T42 | Udbdwpqcc was added by: Udbdwpqcc | 11:08 |
T42 | <adampigg> mal: i think that macro idea works well, much neater code and now no warnings | 11:57 |
T42 | <adampigg> mister_magister: yottagram binary should be prefixed with harbour- | 12:07 |
Mister_Magister | no it should not | 12:08 |
Mister_Magister | give me one readon | 12:08 |
Mister_Magister | reason* | 12:08 |
T42 | <adampigg> it doesnt play well with the heuristics in taskswitcher | 12:10 |
T42 | <adampigg> your music app does, its good to be consistent | 12:10 |
Mister_Magister | music app is the only app out of 7 that is in store, thats why it has harbour- | 12:11 |
Mister_Magister | otherwise there's no reason | 12:11 |
Mister_Magister | i hate this harbour- prefix as its stupid | 12:14 |
T42 | <adampigg> well, for a start it quickly identifies user intended apps from system apps | 12:17 |
Mister_Magister | desktop linux has no problems with that | 12:18 |
Mister_Magister | you can always make PR ;) | 12:20 |
T42 | <adampigg> i may...u have to try actively to make it not have a harbour prefix tho when creating a new app | 12:21 |
Mister_Magister | like i said, i don't really care about harbour prefix as i hate it | 12:22 |
Mister_Magister | i never had any issue with not having harbour- so maybe problem is in your taskswitcher | 12:25 |
T42 | <adampigg> well, that can certainly be argued, but its tricky to identify all running apps and their icons without getting false results | 13:06 |
Mister_Magister | shouldn't you like⦠switch windows instead of apps | 13:08 |
T42 | <adampigg> lipstick doesnt have an api to get windows afaik | 13:11 |
T42 | <adampigg> but yes, that would be ideal | 13:11 |
T42 | <edp_17> All apps should be consistent, otherwise there will be nothing else but chaos. I thought the 'harbour' prefix for community apps was a requirement rather than an advice. I think this example clearly shows the main problem of doing developments without projects and project management. Each developers has own view of it and this way real | 13:12 |
T42 | ly hard to make something that is consistent. | 13:12 |
T42 | <adampigg> +1 | 13:15 |
Mister_Magister | @edp_17 harbour- is requirement for harbour, you are missunderstanding | 13:34 |
Mister_Magister | its not requirement for community apps | 13:34 |
T42 | <edp_17> What's the difference (community app vs harbour)? | 13:39 |
Mister_Magister | difference is if you want to put your app into harbour or not | 13:41 |
T42 | <edp_17> What is harbour? Is it Jolla store? | 13:44 |
T42 | <adampigg> yes, so, for official store, app needs prefixed | 13:45 |
T42 | <adampigg> no such requirement for openrepos, but still, consistency | 13:46 |
T42 | <edp_17> I think doesn't matter if a community app goes into harbour/openrepos or not, all should follow the same standards. Only that way the apps and developers can rely on the others. If an app is not official Jolla app, that should have the same prefix. This is not question of harbour, openrepos or hosted on my own server point of view, thi | 13:53 |
T42 | s is a question of consistency point of view. We can break this harbour prefix IF it is agreed and accepted by all developers. Otherwise we are only cowboys aimlessly running around and shooting in the dark. π | 13:53 |
Mister_Magister | @edp_17 and i don't think so | 13:56 |
Mister_Magister | sailfishos is linux, linux has no requirements to add stupid harbour- prefix because someone at jolla had that whim | 13:56 |
T42 | <edp_17> I think this is a problem on the Linux side too. If something is not right we should keep doing it just because we have been doing that for decades. π | 13:58 |
Mister_Magister | i don't think its a problem | 14:23 |
T42 | <adampigg> Will be if u ever want it in harbour ;) | 14:23 |
T42 | <edp_17> +1π | 14:36 |
Mister_Magister | @adampigg only one of my apps is in harbour and it has harbour- prefix | 14:58 |
Mister_Magister | rest of apps don't need it | 14:58 |
T42 | <edp_17> If you add the audio and video call functions into your Yottagram, you can name the app any like you want. π | 15:57 |
T42 | <edp_17> (Unless it starts with harbour- π ) π | 15:58 |
T42 | <edp_17> *Until | 15:58 |
T42 | <sayanthtezro> Is tgere any gsi available? | 16:20 |
T42 | <sayanthtezro> I need to install sailfish os on my redmi note 7 pro | 16:21 |
Mister_Magister | @edp_17 audio is more likely | 16:26 |
Mister_Magister | @edp_17 can you⦠build rpm packages? | 16:45 |
mal | Mister_Magister: so what UI issue do you have? | 17:03 |
Mister_Magister | mal: https://pbs.twimg.com/media/ElKyElgXgAAKMDO?format=jpg&name=4096x4096 poor performance | 17:03 |
Mister_Magister | when i change govenor to ondemand or switch to big cores only its better, still lagging but thats not solution | 17:04 |
mal | which device? | 17:04 |
Mister_Magister | Zenfone 5z | 17:04 |
mal | what do you have in compositor config? | 17:04 |
Mister_Magister | snapdragon 845, hybris 16 (i need 17) | 17:04 |
Mister_Magister | only standard hwc thingy | 17:04 |
mal | show it | 17:04 |
Mister_Magister | i tried buffer and the other one but it only made it worse | 17:05 |
Mister_Magister | mal: https://github.com/VerdandiTeam/droid-config-5z/blob/master/sparse/var/lib/environment/compositor/droid-hal-device.conf | 17:05 |
mal | which android base? | 17:06 |
Mister_Magister | i mentioned hybris 16 :) | 17:06 |
mal | ah, missed that | 17:06 |
mal | which values did you try for QPA_HWC_IDLE_TIME | 17:07 |
Mister_Magister | i don't really remember | 17:08 |
Mister_Magister | i can try now if you want | 17:08 |
T42 | <Mattia990> Latest libhybris on which version of Android is based? | 17:08 |
T42 | <Mattia990> Oreo? | 17:08 |
Mister_Magister | 16 @mattia990 | 17:08 |
mal | Mister_Magister: are graphics otherwise ok? | 17:09 |
T42 | <Mattia990> Android 16 lol, no it's Oreo I think | 17:09 |
Mister_Magister | yeah they're fine but i want to speed it up :) | 17:09 |
Mister_Magister | @mattia990 hybris 16, android 7 = 14 so 16 = 9 | 17:09 |
T42 | <Mattia990> Oh yeah, you are right | 17:10 |
mal | Mister_Magister: is there noticeable lag when swiping for example? is the transition otherwise normal meaning no frames shown at wrong times | 17:10 |
Mister_Magister | yeah, when scrolling for example it just stutters for no reason | 17:10 |
mal | does QPA_HWC_SYNC_BEFORE_SET have any effect | 17:11 |
Mister_Magister | should i set it to 1? | 17:11 |
mal | yes | 17:11 |
Mister_Magister | amount of red bars is around same but yeah it seems a lot smoother (i've switched govenor back to schedutil too" | 17:14 |
Mister_Magister | it's not ideal though, got any more tips? | 17:15 |
Mister_Magister | as there are still, smaller, stutters | 17:15 |
Mister_Magister | but it is better for sure, lot better | 17:16 |
mal | random small stutters happen almost on all devices | 17:20 |
Mister_Magister | yeah i want to optimize it as much as i can | 17:20 |
Mister_Magister | just asking if you have other, undocummented tips :) | 17:21 |
mal | you could try increasing QPA_HWC_IDLE_TIME a bit, default is 5, don't increase too much | 17:22 |
mal | also try QPA_HWC_BUFFER_COUNT=3 | 17:24 |
Mister_Magister | QPA_HWC_IDLE_TIME=10 seems to have effect | 17:25 |
Mister_Magister | it is now stable around 60fps and red bards are lot smaller and rare | 17:26 |
Mister_Magister | when im doing nothing they're almost gone | 17:26 |
Mister_Magister | mal: 10 is not too much? | 17:27 |
Mister_Magister | same settings seem to help a lot in my other device too! | 17:28 |
Mister_Magister | QPA_HWC_BUFFER_COUNT=3 seem to make it worse | 17:30 |
mal | Mister_Magister: at 60 fps there is about 16 ms time for each frame so you can see what is the best value | 17:34 |
Mister_Magister | 16 is best value? | 17:34 |
mal | Mister_Magister: just to be sure try QPA_HWC_BUFFER_COUNT=3 without QPA_HWC_SYNC_BEFORE_SET for reference | 17:35 |
mal | no 16 might be too much | 17:35 |
Mister_Magister | i tried it ealier and there was no diff / it was worse :/ | 17:35 |
mal | you have to leave enough time for other parts | 17:35 |
mal | ok | 17:35 |
mal | so go with QPA_HWC_IDLE_TIME=10 and QPA_HWC_SYNC_BEFORE_SET=1 | 17:35 |
Mister_Magister | yep | 17:35 |
Mister_Magister | i'm now trying different values on other phone | 17:36 |
Mister_Magister | this one is harder | 17:36 |
Mister_Magister | mal: since we're talking what is status of 3.4 on obs and hybris 17? | 17:36 |
Mister_Magister | https://build.merproject.org/package/show/nemo:testing:hw:asus:5z/nemo-qml-plugin-systemsettings# kinda want my settings | 17:36 |
mal | those env vars are in hwcomposer plugin, not very well documented though | 17:37 |
mal | 3.4 is coming, there has been some issues getting it there, hopefully it will be fixed soon | 17:37 |
Mister_Magister | yeah hwc plugin is quite weird | 17:37 |
Mister_Magister | i wish i'll see someday performance of sfos on huawei p8lite using surfaceflinger renderer | 17:38 |
Mister_Magister | that thing beat jolla phone even, so smooth | 17:38 |
Mister_Magister | seems for my other device 10 is too much | 17:40 |
mal | which android base is that p8lite? | 17:40 |
Mister_Magister | oh it never went past getting ui | 17:41 |
mal | what was the issue with that? | 17:41 |
Mister_Magister | mal: adding 32bit support in graphics driver in kernel | 17:41 |
mal | is it related to 64-bit? | 17:41 |
Mister_Magister | yep | 17:41 |
mal | so soon you can try aarch64 sfos on that | 17:41 |
Mister_Magister | as if i would bother | 17:42 |
mal | why not? | 17:42 |
Mister_Magister | i alreay got too many ports. don't wanna go back to burnout | 17:42 |
Mister_Magister | and ports is not only thing i have | 17:42 |
Mister_Magister | not to mention noone will use it | 17:42 |
Mister_Magister | not even me | 17:42 |
mal | ok | 17:42 |
Mister_Magister | mal: currently i need hybris 17 on 5z to get ofono play with rild | 17:43 |
mal | I should try to fix some last issues on my aarch64 based port | 17:44 |
mal | what is in hybris-17 that helps with ofono? | 17:44 |
Mister_Magister | mal: using hybris 17 i can use newer vendor that doesn't send data registration status over proprietary signals from different service | 17:44 |
*** monich_ is now known as monich | 17:45 | |
mal | oh, ok | 17:45 |
* Mister_Magister hopefully, that is | 17:45 | |
Mister_Magister | not to mention i gotta finish multiple camera patch | 17:48 |
Mister_Magister | thanks mal <3 | 17:50 |
lbt | PSA: services will be down for a bit - replacing a failed drive - back soon | 19:37 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!