buZz | ok guess i was seeing things ; PP real 31m51 user 28m1 sys 2m24 , n900 real 76m50 user 65m23 sys 6m16 | 00:06 |
---|---|---|
buZz | even the ratios seem to align :P | 00:06 |
sicelo | poor n900 :p | 00:10 |
buZz | still , just ~2x faster , not 20x like cpumark said | 00:10 |
uvos | thats compileing? | 00:34 |
uvos | probubly io bound | 00:34 |
uvos | raw cpu perf should be way more than 2x | 00:35 |
uvos | d4 is allready around 3x on just one core | 00:35 |
buZz | yeah , make -j1 | 00:35 |
uvos | ok | 00:36 |
uvos | yeah the n900's cpu is pittifully slow | 00:36 |
uvos | but the other devices arnt slow but as mutch of a ratio | 00:36 |
uvos | s/but/by | 00:37 |
uvos | ie ram/gpu/io | 00:37 |
buZz | still, imho quite impressive for n900 vs pp | 00:40 |
buZz | i'll try d4 later | 00:40 |
Wizzup | buZz: I mean this comparison isn't too interesting no? | 00:45 |
buZz | i'm just doing it cause freemangordon asked | 00:46 |
buZz | was* | 00:46 |
buZz | the d4 is now just fascination by me | 00:47 |
buZz | would it be ~40m or ~60m :P | 00:47 |
buZz | Wizzup: this isnt really work :P its just 'time make -j1' alt-tab back to series | 00:48 |
Wizzup | ;p | 00:49 |
buZz | i mean, would be funny if its faster than pp :P | 00:54 |
buZz | d4 real 29m user 18m sys 2m7 , omg, it -is- faster than pp | 01:20 |
Wizzup | i/o matters a lot here | 01:21 |
buZz | totally, but io matters a lot for user experience too | 01:21 |
Wizzup | what I am getting is that it could be kernel related | 01:21 |
buZz | oh, yeah can be, its just 'current state' i guess | 01:22 |
freemangordon | Wizzup: when you have some spare time, could you please open addressbook and compare scrolling 'choppiness' with and without (ctrl-shift-n) compositing | 07:31 |
freemangordon | I see no difference here, but maybe it is just me | 07:31 |
freemangordon | ugh, pixmaps are created/destroyed on evry scroll frame | 08:31 |
freemangordon | no wonder it is slow | 08:31 |
freemangordon | ok, that's crazy, somebody is doing crazy things here, unfortunately mapping pixmap into PRV memory can take up to 15ms depending on the size | 08:53 |
rafael2k | morning | 08:54 |
freemangordon | hi! | 08:54 |
rafael2k | hi there, libcamera crew unbroke compilation with gcc 8.3... this is good news for us | 08:55 |
rafael2k | I'm creating new test packages for arm64 | 08:56 |
freemangordon | :) | 08:56 |
freemangordon | oh, sometimes it takes 20 ms to map a BO | 08:56 |
rafael2k | freemangordon: which device are you working on? | 09:00 |
freemangordon | d4 | 09:00 |
rafael2k | right | 09:01 |
freemangordon | but I guess it is the same on PP | 09:01 |
freemangordon | as I am debugging performance issues in xorg driver | 09:01 |
rafael2k | indeed, performance is not stellar... but at least does not crash anymore. | 09:01 |
freemangordon | it turns out thaty someone constantly creates/destroys fullscreen pixmaps when scrolling in addressbook, for example | 09:01 |
freemangordon | well, almost fillscreen | 09:02 |
rafael2k | how to open the addressbook? just to be sure I'm in the right place | 09:03 |
rafael2k | today I broght the pp to the work | 09:03 |
freemangordon | ummm. click on contacts | 09:03 |
freemangordon | keep in mind on PP it is even worse, because of the SW rotation | 09:03 |
freemangordon | on d4 it scrolls with 20-30 fps | 09:04 |
freemangordon | on PP i guess it is with 15 | 09:04 |
rafael2k | I have no contacts | 09:05 |
rafael2k | :P | 09:05 |
rafael2k | I need to fill there first | 09:05 |
freemangordon | well, yeah | 09:06 |
rafael2k | lemme finish compile libcamera first | 09:06 |
freemangordon | you can import vcf files | 09:06 |
rafael2k | good idea | 09:09 |
norayr | rafael2k: i know what is the issue with your pine keyboard. | 10:33 |
norayr | i was not able to charge mine without the pin64 charger. | 10:33 |
norayr | you need something that gives 5.4V to charge it. | 10:33 |
rafael2k | norayr: really? oh my... | 10:47 |
rafael2k | can you confirm which one in the website? | 10:50 |
norayr | rafael2k: the pine charger makes it really clear. it has 5 points of charge. the leftmost usb is of 5.3V, others of 5.2V, and it doesn't take charge if connected to the port with 5.2V. | 10:51 |
norayr | rafael2k: https://www.pine64.org/pinepowerdesktop/ this one. | 10:52 |
norayr | but i believe if you find something which may give you 5.3V it'll start charging. | 10:53 |
norayr | (and i want you to use it, because currently the support for pine keyboard in maemo is not great. it works much better in pmos) | 10:53 |
rafael2k | Did you buy this charger? | 10:55 |
Wizzup | freemangordon: still need me to check? | 10:55 |
rafael2k | I can buy it... it seems nice indeed | 10:55 |
rafael2k | norayr: why the support of the pp kbd is not great? | 10:59 |
rafael2k | did you try mobian? we just the same patches they use. | 11:00 |
rafael2k | *we use | 11:00 |
norayr | no, didn't try mobian for long time. let's say hyphen character is not possible to print from the keyboard. that's right alt with 7 but these shortcuts don't work. | 11:06 |
Guest224 | something about PP keyboard on maemo forum: https://talk.maemo.org/showthread.php?t=101337&page=3 | 11:20 |
rafael2k | norayr: are the keycodes being generated correctly? | 11:24 |
rafael2k | try with xev | 11:24 |
Guest224 | It would be nice that "pine"-button do same that Sym(blue arrow+ctrl) in N900. | 11:31 |
norayr | rafael2k: i don't know what to compare to. under phosh xev won't work. | 11:37 |
norayr | i don't remember what was the sym button in n900. | 11:45 |
Guest224 | sym button gives symbol-menu on screen. | 11:46 |
Guest224 | on scandic-layout it was usefull to get characters that didnt found on keyboard. | 11:48 |
Guest224 | or all layouts. | 11:49 |
Guest224 | or pine-button can be just blue arrow-button and with ctrl it work like with N900. | 11:57 |
Guest224 | in Meamo 5. | 11:57 |
rafael2k | norayr: just test in Mobian if the key combinations you mention does work | 11:59 |
Guest224 | Good maemo propaganda: https://leste.maemo.org/Propaganda , but should Leste be more ready for average user, before share..so they don't get disappointed and never test again? | 12:05 |
rafael2k | norayr: not asking to compare, just run in Maemo and check if the keycodes are correct | 12:05 |
rafael2k | really cool the propaganda! | 12:07 |
rafael2k | I like all, especiallu the "Niet" one, and also the other with Chico Buarque as the Maemo user | 12:08 |
rafael2k | love him | 12:08 |
buZz | hmm, how -should- i make those debian/changelog files? it keeps complaining about one i handtyped | 12:10 |
buZz | :P | 12:10 |
Wizzup | copy paste another one, edit it | 12:11 |
Wizzup | date -R for the date | 12:11 |
buZz | Guest224: the Propaganda is mostly just to accompany Wizzup on openfest.org next weekend | 12:11 |
buZz | Wizzup: there is no other one | 12:11 |
Wizzup | buZz: any other project | 12:11 |
buZz | oh right, ok | 12:11 |
buZz | i still need a .desktop file i guess | 12:12 |
buZz | this is the changelog error ; | 12:13 |
buZz | dpkg-genbuildinfo: warning: debian/changelog(l4): found end of file where expected more change data or trailer | 12:13 |
buZz | eh > W: devilutionx: empty-binary-package | 12:14 |
buZz | meh :( | 12:15 |
rafael2k | be aware of tabs vs spaces... at least in rules, this is critical | 12:15 |
buZz | rafael2k: if you have any useful tips about Packaging, i want to expand https://leste.maemo.org/Packaging a bit | 12:15 |
Wizzup | buZz: fwiw I -always- copy fomr some other pkg and just adjust words/date | 12:16 |
buZz | alright | 12:16 |
buZz | meh, seems it did build but somehow lost the binaries , lol | 12:17 |
buZz | guess i'll look later, today is otherjobday , -likely- i'll be sitting in a car doing not much for ~6-8 hours today so maybe i'll continue remote ;) | 12:18 |
buZz | *throws usb powerbanks and a 12V usbadapter in bag* | 12:18 |
buZz | Wizzup: maybe that ebay vendor got lost in their warehouse :P | 12:19 |
rafael2k | buZz: I think Debian docs are pretty enough tbh | 12:19 |
rafael2k | but the more docs, the better, always | 12:19 |
rafael2k | : ) | 12:19 |
Wizzup | buZz: give it another day, we're not in a rush | 12:20 |
buZz | rafael2k: yeah if you know any good ones? i can add em if you want | 12:20 |
buZz | Wizzup: yes we arrrre :P lol | 12:20 |
buZz | :P | 12:20 |
Wizzup | everything I ordered from ebay is already waiting for me at the office | 12:20 |
Wizzup | it's like 1 day delivery or so | 12:20 |
buZz | nice, all the phones? | 12:20 |
Wizzup | the few I ordered yes | 12:20 |
Wizzup | atrix and some | 12:20 |
buZz | cool :) | 12:20 |
rafael2k | I'm still without the ringing sound since the maemo-ringtones is wrong (or profile-data is wrong) | 12:21 |
buZz | rafael2k: aw :( even if you just set it by hand directly? | 12:21 |
rafael2k | buZz: what do you mean "by hand" here? | 12:21 |
buZz | > sphone -c options | 12:21 |
Wizzup | buZz: no, he's saying the default does not work | 12:21 |
rafael2k | I just want the distro fixed, of course I can make it work in different ways... | 12:22 |
buZz | ah hmhm | 12:22 |
rafael2k | renaming the .aac, changing in the config... | 12:22 |
Wizzup | iirc someone submitted a pr for a config change no? | 12:23 |
Wizzup | maybe my memory doesn't serve me well | 12:23 |
rafael2k | if someone send a PR, I'm not aware | 12:23 |
Wizzup | ok | 12:23 |
rafael2k | *sent | 12:23 |
* Wizzup bbiab, packing :) | 12:23 | |
* buZz waves at Wizzup | 12:23 | |
rafael2k | if nobody does it, I'll just explode the maemo-ringtones package and edit it by hand, and re-pack it with ar | 12:24 |
buZz | rafael2k: would be cool, i still want to try 'resetup from new image' so i can hit those default configs again some day soonish | 12:24 |
Wizzup | rafael2k: I think the proper thing is to fix profilex | 12:25 |
buZz | rafael2k: i found this btw ; http://space.nurdspace.nl/~buzz/maemo/Nokia_Ringtones_Opendir_mirror/ | 12:25 |
buZz | (and mirrored it) | 12:25 |
rafael2k | buZz: I plan to do the same | 12:25 |
buZz | yeah its not uncommon for 'apt update' from a old install to behave somewhat different from a 'fresh' install | 12:25 |
buZz | oh right, i still want a second simcard, maybe i should order it already so i can bring up my second d4 as 'permanent' 3rd internetconnection | 12:26 |
rafael2k | buZz: tks! lets pack all the N900 in that package and be happy! I think I have it here anyway, but will download again just in case | 12:26 |
buZz | :D | 12:26 |
buZz | rafael2k: you're welcome to it, space.nurdspace isnt superfast but our internet is free/sponsored | 12:27 |
rafael2k | sim card is good, sms alert is working, my phone just made some noise now | 12:27 |
buZz | so no issues in hogging the line :) | 12:27 |
rafael2k | cool, will do it! | 12:27 |
buZz | rafael2k: i think many of the /N900 stuff there is already in (that? a different?) package | 12:28 |
buZz | one of the 'issues' i ran into was that all the ringtones included lacked metadata inside the .aac , so (non-default) selection looks like ; http://space.nurdspace.nl/~buzz/maemo/2022-08-31-010412_960x540_scrot.png | 12:29 |
buZz | minor i guess, its not the normal method people would select a ringtone (by browsing to the files) | 12:29 |
rafael2k | I'll check this | 12:30 |
rafael2k | are you sure the metadata was embedded in the file? | 12:30 |
buZz | tiny quality of life fixes ftw :P | 12:30 |
buZz | no, i am not sure | 12:30 |
buZz | hmmm > Raw ADTS AAC files have no native tagging format. Some applications can force ID3 tags on them with unpredictable results/readability. | 12:31 |
buZz | yeah not sure what nokia did/does | 12:31 |
rafael2k | indeed... may be an out-of-band metadata file would be ok too | 12:32 |
rafael2k | but I agree having the metadata is cool | 12:32 |
Guest224 | now new news after april 2022 at https://maemo-leste.github.io/ ..should there be at least news that you have worked hard for basic phone thinks to get work. | 12:34 |
Guest224 | now = no | 12:34 |
Wizzup | it's being worked on | 12:34 |
buZz | Wizzup: do you need some material collected by 'us' as suggested topics for a new news post? | 12:37 |
Guest224 | there don't have to be all news at once, you can leave something to next month :) | 12:38 |
buZz | Guest224: yeah we could use a more trickled newsflow | 12:39 |
buZz | maybe we should start some 'leste news' channel and do weekly state-of-the-maemo reports :P | 12:40 |
buZz | hehe | 12:40 |
Guest224 | just any news would be good that https://maemo-leste.github.io/ show that project is alive...half year is long time these days. | 12:41 |
Wizzup | buZz: you can send me a list | 12:43 |
Wizzup | I do have a bunch | 12:43 |
Wizzup | buZz: I think we're probably not doing enough for weekly reports, but maybe | 12:43 |
buZz | Guest224: we're aware :P | 12:43 |
Wizzup | Guest224: it will happen this week | 12:43 |
Wizzup | brb again | 12:43 |
buZz | Wizzup: even just a irc summary can be cool, we do discuss a lot here | 12:43 |
buZz | or maybe a summary of commits to stuff | 12:43 |
buZz | i'll think about it a bit | 12:44 |
buZz | hmm, default q/calendar 'alarm' is just a single beep and vibration motor once | 12:45 |
buZz | thats not very attention grabbing :P | 12:45 |
Wizzup | buZz: not for me, it's a proper alarm | 12:58 |
Wizzup | if you put the phone upside down during the alarm it will suspend the alarm | 12:58 |
buZz | oo, that might have happened then | 12:59 |
buZz | is it smart enough for 'pocket survival' then? | 12:59 |
buZz | i mean, if phone is in your bag/pocket, how will it prevent auto-suspending | 12:59 |
Wizzup | try it :) | 13:00 |
buZz | willdo ;) | 13:07 |
buZz | hmmm, only just noticed, the video of droid4 is kinda 'blue' , wonder if i can use a calibration tool to make a better color profile for droid4 | 13:13 |
buZz | (and maybe the others too) | 13:13 |
buZz | i'll go ask if i can borrow that device soonish (datacolor spyder4 , iirc) | 13:13 |
buZz | afk, bbl | 14:31 |
buZz | i keep wanting to do /op * in here | 14:42 |
buZz | lol | 14:42 |
freemangordon | uvos: I wonder shall I dig deeper to try to find who creates/destroys pixmap on every frame or just implement some pixmap/bo cache and call it a day | 15:18 |
tmlind | sicelo: heh that gpu stuff looks amazing :) | 15:55 |
tmlind | freemangordon: i'll take a look if that one patch is needed when rebasing patches to v6.0, maybe it's just some leftover from the ddk-1.9 stuff | 15:56 |
freemangordon | tmlind: ok, thanks | 19:17 |
freemangordon | tmlind: though, I think it is more of a question if WC memory is cache synced or not | 19:18 |
norayr | buzz, i didn't understand, what do you want to calibrate. are you able to take photos with droid4 under linux? | 20:04 |
norayr | Wizzup, will we have new supported devices? | 20:06 |
norayr | (i mean did you get any new omap devices you intent to start supporting?) | 20:06 |
buZz | norayr: the color of the screen | 20:44 |
buZz | droid4 camera afaik has no driver yet? | 20:45 |
buZz | freemangordon: did oy see? d4 was faster in make -j1 than pp | 20:45 |
norayr | buzz: wow wow wow | 20:55 |
norayr | did MartijnBraam[m] see it? | 20:55 |
MartijnBraam[m] | did I see what? | 20:56 |
norayr | btw MartijnBraam[m], i am making photos with megapixels again, and i use my LUTs for emulating film, so i use ~/.config/megapixels/postprocess.sh and i get results like this: https://xn--49ap3c.xn--y9a3aq/media/uploads/2022/10/07/7c39f295-0c0c-418b-a9f8-96497ee94d6f.jpg | 20:57 |
norayr | so i remember you wanted to remove that possibility, but i really need to have custom postprocess.sh | 20:57 |
norayr | MartijnBraam[m]: buZz said that droid4 was able to compile faster than pp. | 20:57 |
MartijnBraam[m] | oh I'm definetly not removing that capability | 20:57 |
MartijnBraam[m] | that's why I implemented a whole postprocessor switching thing | 20:58 |
Wizzup | norayr: yes we plan to support more devices, we have razr and atrix, but I first want droid 3 and xyboards to work ok | 20:58 |
norayr | oh good, i misunderstood something you said probably. | 20:58 |
MartijnBraam[m] | norayr: I thought it was between the n900 and the pp :P | 20:58 |
norayr | yay yay yay, droid3, it looks amazing, waiting impatiently to test newer builds. | 20:58 |
Wizzup | norayr: no fixes were made yet | 20:58 |
Wizzup | don't get too excited | 20:58 |
norayr | theoretically pp must be faster, but i don't understand why droid4 overperformes it. | 20:59 |
MartijnBraam[m] | if you upgrade a pmos system with megapixels and postprocess.sh to the new release it will default to the old postprocess.sh script, it will show postprocess.sh and postprocessd in the settings window | 20:59 |
MartijnBraam[m] | and you can add multiple custom postprocessors to switch between by adding files in the right directories | 20:59 |
norayr | good. | 20:59 |
MartijnBraam[m] | well if you're comparing single core, the droid 4 is cortex-a9 1.2Ghz and the pp is cortex-a53 1.15Ghz, the memory bandwidth and storage bandwidth on the pp suck so I can see why it wins | 21:00 |
norayr | i also do film photography, a lot. so when i do digital, i then use my luts. and when using megapixels, i don't need to separately convert out of camera jpegs to my jpegs, i just get my jpegs out of camera. | 21:00 |
MartijnBraam[m] | I kinda want to add lut support to the postprocessd flow, have not looked into it yet | 21:01 |
MartijnBraam[m] | applying luts is annoying with libraries | 21:01 |
norayr | i think i can find, i did single core comparison by using fibbonaci calculations. | 21:01 |
norayr | minute. | 21:01 |
norayr | i have those published | 21:01 |
norayr | https://ծմակուտ.հայ/content/4cf0e201-8e9f-4564-8fc7-1d67eef94db3/ | 21:07 |
norayr | oh pp outperforms droid4 on fibonacci test here. | 21:07 |
norayr | and xperia xa2 (in 32 bit mode) just a little bit faster than both. | 21:08 |
uvos | cortex a53 should have a bit better ipc | 21:50 |
norayr | what's wrong with its ipc? | 21:50 |
uvos | nothing | 21:50 |
uvos | its ipc should be better than the a9 | 21:50 |
norayr | ah! understand. | 21:50 |
uvos | but the difference in single core performance between the a53 and the a9 is maybe not that massive. | 21:50 |
uvos | i thought the pp was clocked higher than that | 21:50 |
uvos | i gues not | 21:51 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!