marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | Not ready for end users / self contained install yet. Soon. | General project discussion | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
chengsun has joined #asahi
chengsun_ has quit [Ping timeout: 480 seconds]
gooseduckoftc has quit [Ping timeout: 480 seconds]
sheepgoose has joined #asahi
sheepgoose has quit [Ping timeout: 480 seconds]
popo_ has joined #asahi
popo__ has quit [Ping timeout: 480 seconds]
<kit_ty_kate> Glanzmann: thanks, I started with the uboot install video you sent me.
<kit_ty_kate> It works great. I think the part when you talk about how 1tr doesn’t work for you the first time can be removed
<kit_ty_kate> at least i didn’t have any issues
<kit_ty_kate> it breaks the pace of the video a bit
<kit_ty_kate> for the m1-d-i video I have to find a free usb stick. I’m not sure i have any free on hand atm :/
<kit_ty_kate> Glanzmann: do you not have a script to install whatever distributions from 1tr or something like that?
qeeg has quit [Read error: Connection reset by peer]
qeeg has joined #asahi
<Glanzmann> kit_ty_kate: Yes, I do.
<Glanzmann> kit_ty_kate: I see. For me the part where I don't end up in 1tr, happens everytime. Om air and mini. Not a single time that it did not happen to me.
<Glanzmann> kit_ty_kate: Can you send me the output of 'diskutil list'?
qeeg has quit [Read error: Connection reset by peer]
PhilippvK has quit [Ping timeout: 480 seconds]
kov has quit [Quit: Coyote finally caught me]
phiologe has joined #asahi
kov has joined #asahi
user982492 has joined #asahi
mrbatmanface[m] is now known as fezhead[m]
marvin24_ has joined #asahi
marvin24 has quit [Ping timeout: 480 seconds]
jleightcap has joined #asahi
eroux has joined #asahi
jleightcap has quit [Remote host closed the connection]
jleightcap has joined #asahi
jleightcap has quit [Remote host closed the connection]
jleightcap has joined #asahi
jleightcap has quit [Remote host closed the connection]
sheepgoose has joined #asahi
chadmed has quit [Remote host closed the connection]
refi64 has quit [Quit: The Lounge - https://thelounge.chat]
refi64 has joined #asahi
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
chadmed has joined #asahi
<Glanzmann> kov / tpw_rules: I think the '-s 1' is causing the issue: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1005729
mrkajetanp has quit [Ping timeout: 480 seconds]
the_lanetly_052__ has joined #asahi
jleightcap has joined #asahi
sheepgoose has quit [Read error: Connection reset by peer]
sheepgoose has joined #asahi
the_lanetly_052__ has quit [Ping timeout: 480 seconds]
sheepgoose has quit [Quit: Konversation terminated!]
the_lanetly_052__ has joined #asahi
bisko has joined #asahi
eroux has quit [Quit: Textual IRC Client: www.textualapp.com]
myon98 has quit [Quit: Bouncer maintainance...]
mrkajetanp has joined #asahi
myon98 has joined #asahi
off^ has joined #asahi
eroux has joined #asahi
MajorBiscuit has joined #asahi
jochum_ has joined #asahi
jochum has quit [Ping timeout: 480 seconds]
<kit_ty_kate> Glanzmann: done in DM
leo60228- has quit [Ping timeout: 480 seconds]
leo60228 has joined #asahi
balrog has quit [Quit: Bye]
<Glanzmann> marcan wrote something about cpufreq + alpha. What does alpha mean? Alpha user release?
<Shiz> a/w 12
<jn> Glanzmann: in which context?
<j`ey> "Tomorrow, CPUfreq to (maybe) wrap up the PM week+𝛼."
<j`ey> which I read to mean week+a little bit more since it wasnt actually a week
<Glanzmann> I see, got it. Yep. :-)
<jn> j`ey: makes sense
balrog has joined #asahi
<Glanzmann> Anyway, I look forward to the cpufreq driver.
refi64 has quit [Quit: The Lounge - https://thelounge.chat]
<kit_ty_kate> Glanzmann: i think it would be great if there was a script that directly combined alxsh, your uboot install script and your debian installer perl script
refi64 has joined #asahi
<kit_ty_kate> all three worked pretty well but it would be even easier to simply have the three combined in one
<j`ey> you have to run them separately
<j`ey> alxsh requires a reboot
<kit_ty_kate> yes ofc, but you can basically modify step2.sh in the first script and directly adding the two others sequantially
<Glanzmann> kit_ty_kate: I agree. But I think marcan will do that at some point, I also thought about modifying the installer to install debian, but I assume marcan will do that soon. At the moment I want to stay as close as possible to what marcan has in his mind.
yuyichao has quit [Ping timeout: 480 seconds]
yuyichao has joined #asahi
<kit_ty_kate> unless he said otherwise i think having other experiments done in parallel, even if redundant, is always good. It allows to iron out the kinks if any is discovered and discover new ways of doing things sometimes
<kit_ty_kate> but in any case I think I’ll give a shot to doing that myself, it seems rather fun to do ^^
axboe has joined #asahi
off^ has quit [Remote host closed the connection]
axboe has quit [Quit: reboot]
<Glanzmann> kit_ty_kate: Do that. Modifying the installer is rather easy now that marcan did all the havy lifting, if you like python.
<Glanzmann> kit_ty_kate: I though about modifying the installer, but a) I hate python, b) I assume marcan will do it anyway at some point, and than adding Debian is super easy.
axboe has joined #asahi
user982492 has joined #asahi
the_lanetly_052___ has joined #asahi
the_lanetly_052__ has quit [Ping timeout: 480 seconds]
nafod has quit [Quit: Ping timeout (120 seconds)]
refi64 has quit [Quit: Ping timeout (120 seconds)]
nafod has joined #asahi
refi64 has joined #asahi
refi64 has quit [Quit: Ping timeout (120 seconds)]
refi64 has joined #asahi
balrog has quit [Remote host closed the connection]
balrog has joined #asahi
<axboe> fwiw, getting about 5-5.5h of battery life out of normal use in X with the cpufreq patches
<axboe> sw based X still taking it's toll on battery, but the laptop is noticeably cooler now (which isn't really surprising, as I was running the P cores at full tilt before...)
<mps> hmm, 5.5h is not much, but probably depends on the load
MajorBiscuit has quit [Ping timeout: 480 seconds]
<axboe> mps: it's not, but I do have cinnamon using 50% cpu all the time
<maz> 5.5h is about what I get on my X1 with all the corp crap running...
<axboe> mps: might be interesting to just check if it's not running X, but I need x ;)
<axboe> maz: yeah it's about on par with my x1 already, and it's better than the x1 when actual work (eg compiles) are done
<axboe> and it's cooler than the x1, and silent
<axboe> so already a win, I'd say
<maz> stop it. I've been trying not to add an extra piece of HW to the already huge zoo...
<mps> axboe: I didn't tested keeping it long on battery, but with xorg (and awesome wm) few days ago it drained about 10% for one hour (and that is without cpufreq patch)
<j`ey> maz: if/when mac minis with 10cores+ come out.. and 64G of ram, you'll have to :P
<axboe> mps: is that running fixed at 2ghz?
<axboe> mps: maybe cinnamon sucks for other reasons and that's part of the issue
<mps> also, my son somethimes runs it with macos and compiling a lot of dart/flutter code and it drain about 50-60% for about his 'work shift' which is about 12 hours
<axboe> maz: you know you want to...
<axboe> mps: that's awesome, pretty sure I can get to a full work day on this thing too once it's all sorted
<mps> axboe: yes, I run at at default (i.e. full) speed
<axboe> mps: default m1n1, or patched to 3.2?
<mps> axboe: default one
<axboe> mps: ok, so that's not full speed then, it's 2ghz iirc
the_lanetly_052___ has quit [Ping timeout: 480 seconds]
<mps> axboe: yes
<axboe> let's try awesome...
axboe has quit [Quit: leaving]
<mps> axboe: but if cinnmon have 50% load then it is something strange with it
<mps> as400[m]: `uptime` => 18:52:06 up 4:50, 0 users, load average: 0.02, 0.05, 0.08
thinkofher has quit [Remote host closed the connection]
<mps> currently, an I have firefox with 20-30 tabs, matrix client, few mosh sessions, telegram running
<mps> s/an/and/
<mps> as400[m]: sorry, I mean axboe
thinkofher has joined #asahi
axboe has joined #asahi
thinkofher has quit [Remote host closed the connection]
<axboe> awesome does seem to idle at almost nothing
<mps> axboe: I'm 'nitpicking' when using software. I want less possible features and options, just enough to work and to be useful
<mps> for example I don't run udev, pulseaudio or pipewire, udisk automouner and even dbus don't run always
<axboe> mps: that's actually why I just stuck with cinnamon, don't really need much
<axboe> and it's idling fine on x86, but it's doing a lot of jit stuff on arm which is odd
<axboe> mps: you run way less than me ;)
<j`ey> .. jit?
<mps> well, I have 5-6 old arm32 chromebook which is quite fine and responsive workstation
<maz> j`ey: if that happens, I may have to relegate the D05 to be a pure test box. until then...
<sven> at that point they may as well just call it mac pro
<mps> also my old mt8173 (mediatek elm) chromebook runs fine on battery for about 8 hours or sometimes even 12
<sven> and that one’s definitely going to hurt my wallet
<axboe> j`ey: yep, llvmpipe
<j`ey> axboe: oh right
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
axboe has quit [Quit: Lost terminal]
<jannau> axboe: my best guess is that blinking cursors or similar are rendered at 30 or 60 Hz with full frame buffer updates with simpledrm
axboe has joined #asahi
<jannau> should improve when we get the proper drm driver in shape
<axboe> oh certainly, guess I just either need to stick with cinnamon or replace it with something less intensive
nabaiste^ has joined #asahi
<jannau> kde/plasam/wayland is also mostly idle as long as there no screen updates
<axboe> I might just need to turn off some effects or something
<Glanzmann> axboe: One the air with x, fvwm2, firefox, xterm I can last 6-8 hours.
<axboe> Glanzmann: sweet
<tpw_rules> xfce and xorg does not use much cpu even with a lot of terminal activity
<axboe> fwiw, 4h -> 5.5h here for me with cpufreq
<axboe> but sounds like my main issue is cinnamon and that it should be a lot better with something else
<axboe> I do quite a bit of compiles, so need cpufreq so it won't take forever
<axboe> let's try xfce4... brb
axboe has quit [Quit: leaving]
joske has joined #asahi
<joske> axboe: xfce idles at 0% even with compositing enabled. Marco with compositing idles at 50% too
<joske> With compositing disabled also idles at 0%
<joske> Cinnamon can not run without compositor
<joske> Have not tried other desktops
nabaiste^ has quit [Ping timeout: 480 seconds]
<Glanzmann> btw. I use 'lightdm' as login manager on Debian and enable auto-login: https://pbot.rmdir.de/EkloZoaylqWI0EXnxzR4gQ
axboe has joined #asahi
<axboe> xfce4 definitely more idle, but still seem to get a pretty low battery life
<j`ey> how much bigger is the m1 max/pro battery than the m1?
<axboe> not sure what it is, looking at frequency it's ~6% at 600MHz, ~93% Idle
<axboe> so unless the idle loop is doing something stupid, it's not the cpus eating the power
<kettenis> display is probably the biggest consumer
<axboe> most likely
<axboe> not super bright though
<Glanzmann> axboe: I sometimes notice that my macbook air is considerable hotter than under macos.
<Glanzmann> mine is super bright.
<axboe> Glanzmann: I've never run osx on this one except to do the setup, so not sure what it's like there
<axboe> but cpufreq made it definitely cooler on the lap
user982492 has joined #asahi
<kov> fwiw gnome idles at 0% pretty much
<kov> gnome-shell 41.3 on debian testing
tomtastic_ has joined #asahi
<axboe> j`ey: says 54.7Wh
tomtastic has quit [Ping timeout: 480 seconds]
nabaiste^ has joined #asahi
<axboe> huh, cpu_capacity missing from 5..7
<Glanzmann> That is how it looks for me: https://tg.st/u/screenshot-air-2022-02-14-20_13_13.png
Gaspare has joined #asahi
axboe has quit [Quit: leaving]
___nick___ has joined #asahi
___nick___ has quit []
___nick___ has joined #asahi
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
axboe has joined #asahi
<axboe> was digging a bit, and seems like cpufreq isn't correctly set on cpus 5-7
<axboe> so they are most likely still running at full tilt, which would help explain why battery isn't as good as it should be
___nick___ has quit [Ping timeout: 480 seconds]
<axboe> the fact that cpu_capacity is also missing for those 3 cpus makes me suspicious that something is wrong in the definition or bringup of them
Gaspare has quit [Read error: Connection reset by peer]
<jannau> axboe: looks ok here on system without disabled cpu cores
<axboe> jannau: I dropped that m1n1 hack to set all P cores to 15 in cpufreq
<axboe> p0..p2 seem to scale fine, p3..p5 do not
<axboe> the ratio between them are pretty close to max-freq/boot-freq
<axboe> never worked on embedded like stuff, so just stabbing in the dark here and finding clues :)
<tpw_rules> https://github.com/AsahiLinux/m1n1/commit/6f525fc0f18652bf2804725dfa0563a30d8ff06f does this need to be fixed in a different place?
nabaiste^ has quit [Ping timeout: 480 seconds]
<kettenis> note that the cpufreq code is an old diff that is effectively abandoned because the maintainer of the relevant subsystem wan to see a different approach
<axboe> kettenis: yeah I know
<axboe> kettenis: not spending time on that, just trying to verify if there's something wrong with how I set it up
<axboe> or if m1n1 is setting up things wrong for the t600x
<jannau> axboe: I suspect the disabled cores on your break the association with clusters
<kettenis> I wouldn't be surprised if the mapping of CPU IDs onto PMGR registers is messed up for the modelsthat have some of the cores disabled
<axboe> jannau: could be, yes
<axboe> so...
<axboe> since I have 6 p cores
<axboe> are the clusters [ 0 1 2 ] [ 3 4 5], or [ 0 1 2 3 ] [ 4 5 6 7 ]
<axboe> only [4 5] in the last one
<jannau> axboe: m1n1 has no code to fix up the cpufreq related nodes for disabled cores
<axboe> that might explain it, if so
<kettenis> the cluster ID is encoded in the MPIDR register
<kettenis> MPIDR_EL1
<axboe> I might've just messed up the definitions
axboe has quit [Quit: leaving]
<jannau> `dtc -I fs -O dts -o - /proc/device-tree` will output the device-tree as seen by linux
nabaiste^ has joined #asahi
<jannau> I suspect you will end up with 3 cpu@1010x and 3 cpu@1020x nodes, i.e. 3 cores per cluster
<jannau> you probably have remove the disabled cores from the cpu map. the numbers in live cpu map reference the phandle property in the cpu nodes
axboe has joined #asahi
<axboe> ok I think the layout is right, it must be something with the freq driver then
<kov> https://bugs.webkit.org/show_bug.cgi?id=236564 landed, gnome is pretty much ready for 16k page size ;D
<tpw_rules> kov: do you know what the release cycle is like?
<kov> that's tricky to answer, but I'll probably poke the maintainers of webkitgtk to backport that to the stable branch
<tpw_rules> that would be nice
<tpw_rules> now we need to figure out how to get the libuwind folks' attention
<axboe> jannau: yep, 3 per cluster is what I see
axboe has quit [Quit: Lost terminal]
axboe has joined #asahi
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
axboe has quit [Quit: Lost terminal]
VinDuv has quit [Remote host closed the connection]
VinDuv has joined #asahi