2025-08-18.log

gordon1minute: josch: fixed https://source.mnt.re/bugs/bugs/-/issues/?show=eyJpaWQiOiI0MCIsImZ1bGxfcGF0aCI6ImJ1Z3MvYnVncyIsImlkIjo2NjIxfQ%3D%3D00:08
gordon1oops, i made a patch backwards00:09
gordon1fixed00:10
* switchy_ -> switchy00:21
- switchy (QUIT: Remote host closed the connection) (~switchy@115.130.195.183)00:24
+ switchy (~switchy@mechboards/switchy)00:28
- rodion_goritskov (QUIT: Server closed connection) (~rodion_go@mail.goritskov.com)00:30
+ rodion_goritskov (~rodion_go@mail.goritskov.com)00:30
joschare "the new-ish displays" used for all dsi reforms and only there?00:32
+ L29Ah (~L29Ah@wikipedia/L29Ah)00:34
- mjw (QUIT: Ping timeout: 244 seconds) (~mjw@gnu.wildebeest.org)01:42
- chrcav (QUIT: Server closed connection) (~chrcav@user/chrcav)01:52
+ chrcav (~chrcav@user/chrcav)01:53
+ mjw (~mjw@gnu.wildebeest.org)01:59
- mjw (QUIT: Ping timeout: 244 seconds) (~mjw@gnu.wildebeest.org)02:05
+ emdevt (~emdevt@2001:7e8:fa17:7901:889:cfa1:9c37:6514)02:20
- paperManu (QUIT: Ping timeout: 252 seconds) (~paperManu@72.10.129.226)02:53
- colinsane (QUIT: Ping timeout: 252 seconds) (~colinunin@97-113-70-115.tukw.qwest.net)03:11
- skipwich (QUIT: Remote host closed the connection) (~skipwich@user/skipwich)05:01
+ skipwich (~skipwich@user/skipwich)05:01
- potatoespotatoes (QUIT: Ping timeout: 276 seconds) (~quassel@user/potatoespotatoes)05:12
- burley (QUIT: Server closed connection) (fc32d5b940@user/burley)07:11
+ burley (fc32d5b940@user/burley)07:11
- emdevt (QUIT: Remote host closed the connection) (~emdevt@2001:7e8:fa17:7901:889:cfa1:9c37:6514)07:32
+ LainIwakura (~LainIwaku@user/LainIwakura)08:58
- LainIwakura (QUIT: Ping timeout: 250 seconds) (~LainIwaku@user/LainIwakura)09:08
+ LainIwakura (~LainIwaku@user/LainIwakura)09:44
+ mjw (~mjw@gnu.wildebeest.org)09:54
+ chomwitt (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)10:04
- LainIwakura (QUIT: Quit: Client closed) (~LainIwaku@user/LainIwakura)10:08
gordon1oh wow stuff is visible on the display now even with echo 1 >/sys/class/backlight/backlight/brightness10:24
joschgordon1: i just wonder if the pwm freq change can be applied to all rk3588 dsi or whether they were actually sold with different panels or something10:27
gordon1i couldn't test anything but the latest one unfortunately, but i bet the answer lays in the datasheet10:28
gordon1also there is very faint, almost inaudible hum now emanating from the laptop (you need to hold your breath to hear it) and it goes away with 100% brightness10:29
gordon1probably some microphonics10:30
gordon1maybe 10000000 would work better then 20000000, but i picked value right in the middle to be sure10:30
joschOB10:31
joschwhoops, sorry10:31
gordon1s/0000000/000000/g10:32
- chomwitt (QUIT: Ping timeout: 272 seconds) (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)10:38
+ LainIwakura (~LainIwaku@user/LainIwakura)10:47
+ andreas-e (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890d.rev.sfr.net)11:21
- LainIwakura (QUIT: Quit: Client closed) (~LainIwaku@user/LainIwakura)11:27
minutejosch: there are a number of panels out there :D11:42
- mjw (QUIT: Ping timeout: 255 seconds) (~mjw@gnu.wildebeest.org)12:11
* Guest2572 -> mjw12:38
- henesy (QUIT: Server closed connection) (d7619ffbc2@2a03:6000:1812:100::143)12:52
+ henesy (d7619ffbc2@2a03:6000:1812:100::143)12:52
+ paperManu (~paperManu@72.10.129.226)13:14
gordon1have somebody seen something like this? http://0x0.st/KsHS.txt14:43
gordon1after that mpv playback stops and never restarts14:43
gordon1but everything else works fine14:43
gordon1but i have to reboot to recover14:44
gordon1would like to know if it can be recovered without reboot14:45
+ mark_ (~mjw@gnu.wildebeest.org)14:46
minutegordon1: strange. are you using hw video decoding?14:47
gordon1no, software only, didn't sort out v4l-request patch for ffmpeg yet14:47
minuteah also i had some issues with fullscreen14:48
minutewith vulkan gpu-api14:48
+ chomwitt (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)14:49
minutegordon1: you can try different gpu-api settings, like opengl, vulkan, sdl etc, there's also a "new one" but i forgot the name14:49
minutegordon1: ah yeah gpu-next14:49
minutegordon1: otherwise it can only be audio i gues14:50
gordon1sure, will try that14:50
gordon1however that can take a while to verify14:50
minuteand also make sure hwdec is off14:50
minutein .config/mpv/mpv.conf14:50
minutei.e. not set to auto14:50
minutei'm watching tv shows on rk3588 reform every evening with mpv so it should in general be fine :D14:51
gordon1i mean i could but i doubt it can use something that doesn't exist in ffmpeg14:51
minuteand i think i use gpu-next at the moment on gnome14:51
gordon1actually something weird is happening now14:52
minutewhat i found interesting is that vlc isn't able to play back h265 for me (something is too slow) and mpv does it just fine14:52
gordon1reform2-lpc hung http://0x0.st/KsH_.txt14:52
minutegordon1: that's common14:52
minutegordon1: it's not a hang. it's an error by the driver when the value from LPC read over SPI doesn't seem valid14:53
minuteso that you don't get 0% battery reported etc14:53
gordon1it says "Device or resource busy" when i try to read from sysfs14:53
minuteyeah14:53
minuteit's the error that i return in the driver14:53
gordon1right14:53
gordon1is that related to battery status screen flickering in keyboard?14:54
minuteis lpc itself responsive, i.e. can you see battery status on oled?14:54
minutegordon1: not directly related, as the keyboard gets the info over UART14:54
+ sadplan (sadplan@tilde.club)14:54
* sadplan -> sad_plan14:54
gordon1you know what, you nailed it14:54
minutegordon1: i haven't seen that issue in a while though btw, how long do you have to leave the battery screen open to see this issue?14:54
gordon1it is the sound issue14:54
minuteah yeah. sound is often connected to DMA14:55
gordon1i mean with mpv not playing14:55
minuteah yeah and videos not playing is often connected to audio pipeline being stuck14:55
gordon1but pipewire restart does not solve it14:55
minutewhat's the default audio sink at the moment?14:55
minutemaybe it's pro audio but you don't have hdmi audio connected?14:55
gordon1what is pro audio?14:56
minutehdmi shows up as "pro audio" for me (misnomer)14:56
minutei.e. in pavucontrol -> output devices / configuration -> one of the two "built-in audio"14:57
gordon1no, in pipewire i have only the sink named as rk3588-wm896014:57
minuteok14:57
gordon1there is hdmi1 but it is below in priority14:58
minuteare you on debian or something else like gentoo?14:58
gordon1gentoo14:58
minuteah ok. i don't know how to install packages there and stuff. did audio ever work?14:58
gordon1right now no14:59
gordon1not even speaker-test14:59
gordon1i mean i can maybe unload and load the module (when i recompile it as module)14:59
minuteis the wm8960 configured?14:59
minutei.e. in alsamixer14:59
minutenormally reform-hw-setup does that14:59
gordon1yes14:59
minutei.e. setting the right knobs so you can hear something15:00
gordon1i mean i did it by hand and saved the settings15:00
gordon1the sound worked fine before the pl330 message15:00
minuteah, then my question was somehow not clear15:00
minutewhen i asked "did audio ever work?"15:00
gordon1yes, it did15:00
minuteok, that's important for me to know :D ok15:01
gordon1but now it doesn't not only in mpv but everywhere else15:01
minutethen you have to reboot i guess15:01
minutenot sure if dma can be reloaded15:01
gordon1i mean that's like admitting defeat :D15:01
minuteok, then here my knowledge ends, because i haven't debugged pl330 yet15:01
gordon1i'll try to reload wm8960 module next time15:01
minutei can tell you that i didn't have this issue yet at least15:01
gordon1i mean it very well might be not the pl330 fault15:02
gordon1also there is a small chance that i misconfigured the kernel15:02
andreas-eHello, minute! I take the opportunity to give my performance impression following the discussion on Mastodon.15:05
andreas-eI do a lot of C compiling (and C++, in the context of preparing packages for the Guix distribution). And there the MNT Reform Classic is not really usable as the main machine.15:06
minuteandreas-e: on rk3588 i assume, yes?15:07
andreas-eJust to give an example, the other day I have compiled mit-kerberos. This takes about 2 minutes on my 2-core laptop from 2016 (Intel(R) Core(TM) i5-6300U CPU @ 2.40GHz). And about one minute on a 4-core laptop that was given away by a company that did not need it anymore (11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz).15:08
andreas-eAnd 7.5 minutes on the old im8 module that came with the Reform.15:09
andreas-eIndeed I was a bit shocked how slow the processor was when the machine arrived at the time...15:09
andreas-eNow I understand that with the rk3588 I should gain a factor of about 4 or 5? That would bring it roughly to the level of the 2016 laptop, so I could not quite bring myself to invest into that.15:10
andreas-eOn the other hand, I like to have the machine around for debugging build problems on aarch64, or to confirm that there are none. So the machine is useful, but I would not yet want to drop the x86_64 ones.15:12
gordon1so there is a difference between my config and one from source.mnt.re, mine has continuous dma turned off, maybe this is related15:18
- bkeys (QUIT: Ping timeout: 252 seconds) (~Thunderbi@98.19.131.193)15:27
+ murphnj (~murph@user/murphnj)15:34
minuteandreas-e: well, that's only CPU perf. but rk3588 also has better memory bandwidth of course15:40
minuteandreas-e: can i time this mit-kerberos compile on one of my debian rk3588 reforms easily?15:40
minuteandreas-e: i.e. could i clone, automake/make this and that would be comparable? https://github.com/krb5/krb515:41
andreas-eI do not know what is the bottleneck; I usually assume it is raw CPU performance.15:41
minuteandreas-e: compilation is oftentimes also i/o bound15:42
minutejust for fun i'm trying autoreconf and ./configure and then... time make -j8 probably?15:43
andreas-eWe use the tarball from here: https://web.mit.edu/kerberos/dist/krb5/1.22/krb5-1.22.tar.gz15:43
minuteah15:44
andreas-eSo there it would just be ./configure; make -jXXX; make check -jXXX15:44
andreas-eI needed to look up our package recipe to see whether we did anything special, but not really; hm, a few configure flags.15:45
minuteif the configure flags turn on or off features that would be compiled, it could make a difference15:45
minutei'll just run it without any flags though15:45
gordon1can i ask why is that and should I expect ethernet not to work with dwmac builtin? https://source.mnt.re/reform/reform-tools/-/blob/main/bin/reform-hw-setup?ref_type=heads#L59-6715:46
minutegordon1: this is no longer needed15:46
andreas-eminute: It does not look like making a difference, but it is of course difficult to be sure.15:46
gordon1okay, good that i didn't take it15:46
minutegordon1: i found a dts fix for this but i couldn't test it on all devices out there15:46
minutegordon1: but it shouldn't be needed anymore15:46
gordon1i need to test the ethernet but then i need to stand up from the couch...15:47
gordon1ah wouldn't it be nice to have PoE charging capability?15:48
minutereal1m5.224s15:50
minuteuser3m8.771s15:50
minutesys0m30.208s15:50
minuteandreas-e: it took 1 minute and 5 seconds to compile here15:51
minuteandreas-e: and there were a bunch of times where some cpu cores were idling15:51
minuteandreas-e: (this is only running make -j8 though, not make check)15:52
gordon1yeah ethernet works without this hack15:52
gordon1can confirm for mb3.0 reform2-dsi15:52
andreas-eThanks! Could you try "make check -j 8" also? I do not know for this package, but the tests can take longer than the building.15:52
minutegordon1: yeah @ PoE but it's a snake pit apparently, to get it right15:53
gordon1oh well, a man can dream...15:53
gordon1it's not call for action or anything15:53
andreas-eOr the other way round, I will just "make" on my end.15:53
minuteinteresting, it takes a lot of time on t_gss_sample.py but there's not much cpu activity during that time15:55
minuteok:15:56
minutereal1m29.225s15:56
minuteuser1m36.651s15:56
minutesys0m29.004s15:56
minuteandreas-e: so the build took 1:05 mins and the test took 1:29 mins15:58
andreas-eOkay! So very close to my old laptop. I will just do the builds to see.15:58
minuteandreas-e: if you're running the test on that old laptop, maybe you could also run the "7z b" benchmark? it would be helpful to see how well 7z b numbers compare16:02
minuteandreas-e: btw i still use a i9-9900 desktop at the mnt hq for most of my work, esp. heavy cad work, but it "feels" maybe only 2x as fast as the rk3588 in practice, which is very subjective of course16:03
minute(even if benchmarks suggest it should be faster than that)16:03
minutebut right now i'm at home with a cold and do all my work on rk3588 pocket reform ^^16:04
andreas-e2-core x86 with make -j2: 1m19 for build; "make check" fails due to some missing packages, I will just do the build.16:04
andreas-e4-core x86 with make -j4: 31s for build16:08
minuteandreas-e: which x86?16:12
minuteandreas-e: ah, 2 core is the i5-6300u right16:12
andreas-eThe two I mentioned above, yes.16:12
minuteandreas-e: ok, sound like the i5-6300u is slightly slower at building this than the rk3588 and the i7-1165G7 has around 2x the speed of the rk3588 for this task16:13
andreas-eOld MNT Reform: 4m55s16:14
andreas-eSo the rk3588 is about 4 times faster here as well, which if I remember well corresponds to the comparison on the MNT website.16:15
minuteyeah, not bad16:15
minutea bit more than 4x but yeah16:15
minutealso you get 8x the ram :D16:16
andreas-eWell, this is just a random test; it all depends also on the package, how parallel the build is and so on.16:16
minuteyeah. but it is kind of in the expected range. 16:16
andreas-eWhich number do you want to see for "7z b"?16:16
andreas-e2-core x86: Tot:             354   2500   870016:17
minuteandreas-e: yeah tot is the most relevant, but also Avr16:17
minuteandreas-e: to compare, https://source.mnt.re/reform/reform/-/blob/master/reform2-benchmarks/benchmark-7zb-rk3588.txt16:17
minuteandreas-e: here you can see that the rk3588 can be a lot faster than the 2-core intel. 16:18
andreas-eAvr:             318   2912   9265  |              389   2089   813616:18
andreas-eTot:             354   2500   870016:18
minutebut only if the task makes good use of multiple cores16:18
minutesingle core speed seems similar, the rk3588 being slightly faster on avg, and multicore perf is more than double of the 2-core (it would be sad if not, with 8 cores)16:19
andreas-eNot bad indeed! That is why I was a bit surprised about the i.mx8, I thought that double the core would somehow compensate lower performance per core.16:20
andreas-eMy old Reform:16:21
andreas-eAvr:             303    781   2369  |              393   1322   520116:21
andreas-eTot:             348   1052   378516:21
andreas-eSo here the factor to the rk3588 is more than 5!16:21
minuteandreas-e: so avg multicore compression perf of rk3588 is almost 10x of imx8m. which makes sense16:22
minutemore memory bandwidth, more advances cores (esp imx8mq has only in-order cores!)16:22
minuteit has only cortex-a53 which are one of the simplest and smallest 64 bit arm cores16:22
andreas-eThanks for discussing all these benchmarks! And it is good to see progress.16:23
minuteand rk3588 has 4x cortex-a76 which have much better single core perf because they are out-of-order and more complex16:23
andreas-eAnd actually one thing I like a lot about MNT is that you continue improving things, and that you do not just sell one-shot devices.16:23
minute(and 4x cortex-a55 for lower energy tasks)16:23
minuteandreas-e: yeah, the idea is that one can really upgrade over time, otherwise the modularity is not very helpful (except for repairs of course) https://mntre.com/about.html16:24
minutecurrently the most performant chip i have on my radar that we can realistically get is the SM8750/QCS8750 which gets close to i9-9900 perf (with fewer cores). but not sure when it will be working+supported enough in mainline. also rk3688 is supposedly quite a beast, but it hasn't been shown yet nor benchmarked16:26
andreas-eAnd so far everything is passively cooled, right? So that is a factor as well. One cannot have everything at the same time...16:27
andreas-eminute: All the best with your cold, I wish you a fast recovery!16:30
- L29Ah (PART: !!unknown attribute: msg!!) (~L29Ah@wikipedia/L29Ah)16:41
+ L29Ah (~L29Ah@wikipedia/L29Ah)16:42
minuteandreas-e: thanks!16:48
minuteandreas-e: (yes, everything is passive cooled)16:48
- L29Ah (PART: !!unknown attribute: msg!!) (~L29Ah@wikipedia/L29Ah)16:51
+ L29Ah (~L29Ah@wikipedia/L29Ah)16:52
- wiedi (QUIT: Ping timeout: 244 seconds) (~wiedi@ip5f581498.dynamic.kabel-deutschland.de)17:06
+ wiedi (~wiedi@ip5f581498.dynamic.kabel-deutschland.de)17:08
- Zaba (QUIT: Server closed connection) (80b9b4b35e@2a03:6000:1812:100::116)17:25
+ Zaba (80b9b4b35e@2a03:6000:1812:100::116)17:25
+ bkeys (~Thunderbi@66.110.201.50)17:28
- bkeys (QUIT: Ping timeout: 244 seconds) (~Thunderbi@66.110.201.50)17:41
+ AnimaInvicta (~AnimaInvi@88-120-179-216.subs.proxad.net)18:33
- dominicm (QUIT: Server closed connection) (45ee0d70e9@2a03:6000:1812:100::3e6)18:58
+ dominicm (45ee0d70e9@2a03:6000:1812:100::3e6)18:58
- chomwitt (QUIT: Ping timeout: 272 seconds) (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)19:28
* f_ -> funderscore19:57
* funderscore -> f_19:58
+ reform26423 (~rene@p200300d0ff230b007fc2100dbdd09829.dip0.t-ipconnect.de)19:59
+ akira (~akira@5.42.207.8)20:02
* reform26423 -> rwa_20:18
- andreas-e (QUIT: Quit: Leaving) (~Andreas@2a02-8434-b6a3-e901-facc-8e87-8e54-890d.rev.sfr.net)20:19
- rwa_ (QUIT: Quit: Leaving) (~rene@p200300d0ff230b007fc2100dbdd09829.dip0.t-ipconnect.de)20:26
- irimi1 (QUIT: Server closed connection) (47a15995ae@2a03:6000:1812:100::12aa)20:35
+ irimi1 (47a15995ae@2a03:6000:1812:100::12aa)20:35
- ndufresne (QUIT: Quit: The Lounge - https://thelounge.chat) (~ndufresne@apple.collaboradmins.com)20:41
+ chomwitt (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)20:51
- mjw (QUIT: Killed (osmium.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c)20:52
* mark_ -> mjw20:52
+ Guest560 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c)20:52
- akira (QUIT: Read error: Connection reset by peer) (~akira@5.42.207.8)20:53
+ akira (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)20:54
- akira (QUIT: Read error: Connection reset by peer) (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)20:57
+ akira (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)20:58
+ ndufresne (~ndufresne@apple.collaboradmins.com)21:02
- akira (QUIT: Ping timeout: 260 seconds) (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)21:03
+ akira (~akira@5.42.207.7)21:03
- kfx (QUIT: Ping timeout: 260 seconds) (~kfx@wopr.sciops.net)21:40
- chomwitt (QUIT: Ping timeout: 260 seconds) (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)21:45
+ kfx (~kfx@wopr.sciops.net)21:45
+ enwu8 (~enwu@user/enwu)22:14
- enwu (QUIT: Ping timeout: 252 seconds) (~enwu@user/enwu)22:19
* enwu8 -> enwu22:19
- murphnj (QUIT: *.net *.split) (~murph@user/murphnj)22:20
- abortretryfail (QUIT: *.net *.split) (~arf@146.ip-149-56-132.net)22:20
- manis (QUIT: *.net *.split) (01a66df340@185.72.67.185)22:20
- Asmadeus (QUIT: *.net *.split) (~asmadeus@user/meow/Asmadeus)22:20
- blast007 (QUIT: *.net *.split) (~blast@user/blast007)22:20
- b0 (QUIT: *.net *.split) (~b0@user/b0)22:20
- elektron (QUIT: *.net *.split) (~elektron@apoc.halo.nu)22:20
- AnimaInvicta (PART: !!unknown attribute: msg!!) (~AnimaInvi@88-120-179-216.subs.proxad.net)22:24
+ murphnj (~murph@user/murphnj)22:26
+ abortretryfail (~arf@146.ip-149-56-132.net)22:26
+ manis (01a66df340@185.72.67.185)22:26
+ Asmadeus (~asmadeus@user/meow/Asmadeus)22:26
+ blast007 (~blast@user/blast007)22:26
+ b0 (~b0@user/b0)22:26
+ elektron (~elektron@apoc.halo.nu)22:26
- murphnj (QUIT: Remote host closed the connection) (~murph@user/murphnj)22:37
+ murphnj (~murph@user/murphnj)22:38
+ MyNetAz (~MyNetAz@95.216.142.242)22:41
- akira (QUIT: Read error: Connection reset by peer) (~akira@5.42.207.7)23:05
+ akira (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)23:05
- akira (QUIT: Read error: Connection reset by peer) (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)23:14
+ akira (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)23:15
- romi (QUIT: Server closed connection) (bd30729973@user/romi)23:19
+ romi (bd30729973@user/romi)23:19
- akira (QUIT: Ping timeout: 256 seconds) (~akira@2a02:590:501:c00:c86:5031:7f83:e7b1)23:20
+ akira (~akira@5.42.207.7)23:21
- akira (QUIT: Client Quit) (~akira@5.42.207.7)23:21
+ chomwitt (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)23:34
- chomwitt (QUIT: Ping timeout: 256 seconds) (~chomwitt@2a02:85f:9a74:ce00:1ac0:4dff:fedb:a3f1)23:59

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!