08:02 <JensGlathe[m]> git clean -df

08:02 <JensGlathe[m]> argh, again

13:44 <Treibholz> so, let's see if this Ubuntu survives the upgrade from 24.10 to 25.04, or if I finally get the right push, to migrate to trixie...

16:21 <Treibholz[m]> my X13s did survive the upgrade and is now about 50% faster with Ubuntu 25.04 (6.14.x). I have no idea what was wrong with Ubuntu 24.10 (6.11.x), but my incomplete trixie-test-setup (6.12.x) was also 50% faster... what happened between 6.11 and 6.12?

16:23 <Treibholz[m]> and this time I will document (for myself), what and how I have to recompile stuff with different parameters, to be happy... (e.g. pipewire, to get aac for bluetooth)

16:30 <JensGlathe[m]> For me, 25.04 feels slower since a few weeks ago. Build times increased from ~38m to ~43m on the Dev Kit. Couldn't find it yet. It changed suddenly, the box was on 25.04 development channel for a while already.

16:32 <Treibholz[m]> Jens Glathe: nvme-io is still slower, than on trixie `sudo hdparm -tT /dev/nvme0n1` still takes ages.

16:33 <JensGlathe[m]> could be, yeah. But I get good tranfer rates over all, maybe request latency

16:35 <Treibholz[m]> but my benchmarks are now all 50% faster, as long as it doesn't use IO on the nvme. (python-script, which counts headers in a Maildir with 100k files and glmark2)

17:02 <tobhe_> Treibholz[m]: good to hear that it worked. we actually specifically tested the upgrade and new iso on the x13s so I would have been a little disappointed if it didn't work for you :)

17:07 <Treibholz[m]> tobhe_: Good job! :-) - but I was kind of hoping it failed, so I have to switch to Trixie... Now I have to live with all the minor annoyances of Ubuntu (compared to Debian) for even longer.... :-)

17:13 <tobhe> should have added a random ppa that changes your system python version to an incompatible one before upgrading if that was the goal :D

17:21 <Treibholz[m]> just move python into snap aswell and I will be annoyed enough, to switch to ANY other OS... :-)

18:01 <Treibholz[m]> ah, nice. the camera now doesn't need patching in pipewire anymore, to get more than 320x200.

18:15 <steev> Treibholz[m]: why not just switch to trixie? a trixie iso should "just work" as long as you have the sc8280xp-lenovo-thinkpad-x13s.dtb in the efi

18:15 <steev> you can still follow ema's instructions in the debian wiki

18:25 <Treibholz[m]> steev: I have a trixie-installation on a "spare partition", to play around and it also seems to work fine. But... (full message at <https://matrix.org/oftc/media/v1/media/download/AbG1bi_tPPdrshYbIduzslEixZN52Eg5WWYYDp-0P0RU6HpjG0aWu8DtwUijOamM-nMSv4LtvQLiR-9I7xAyEw9CeWkj7m-wAG1hdHJpeC5vcmcvQnlydEF3d25vSkJ2dXlranZMUG5hUmRP>)

18:25 <steev> fair

18:25 <steev> i do kali on mine, so i don't mind the rolling release :)

18:31 <Treibholz[m]> I'm too old for rolling releases on my daily driver. :-) If something is too old for me, I build my own backport - that works better for me.

18:43 <craftyguy> robclark: I was able to repro that vk gtk4 crash thing outside of flatpak (finally.. lol), and bisected it: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13012

18:44 <robclark> thx

18:47 <craftyguy> I'm just glad it's not some funky flatpak config issue, that thing suuuuucks to debug

19:16 <JensGlathe[m]> ok ich bring eine 14er Riege Valsartan 160 mit, das verschafft Zeit

22:52 <deathmist> hey, so I've managed to do more X1E laptop testing this time on a Lenovo Yoga Slim 7x and similar to the previous HP OmniBook X 14 while plugged in it drains battery under full multi-core loads @ 3.4 GHz (Linux) and drops its CPU frequency under Windows (WSL). is there actually any X1E device that *doesn't* do this? seems to be "expected" behavior

22:57 <deathmist> I understand that Linux probably also should be following the Windows behavior and downclocking, but this is practically just throttling to hide an issue with the battery drain when expecting to get X1E performance instead of X1P; maybe I should just wait for X2E at this point

23:05 <HdkR> deathmist: That's expected. When I tested the yoga 7x here I was able to get per core load of >8w. Since the device can only pull 60w from USB-C (even with a 100w charger!) it will occur with all cores loaded.

23:05 <HdkR> I wonder if there is any device other than the dead dev kit that allows higher current draw from the wall.

23:06 <HdkR> But yea, there's no real expected behaviour that if the core complexes are pulling enough power to drain the battery but /not/ be thermally throttling to downclock.

23:07 <HdkR> x86 devices do this behaviour as well

23:17 <roy[m]> <deathmist> "hey, so I've managed to do..." <- It might be even worse. My t14 will reduce the power it draws from the PD charger down to 20W when it gets hot. Normally it can pull ~50W. This does not happen with some magic combinations of usb-c dongles and PD chargers, so I have to use that whenever running a long cpu-intensive jobs. Really wish there was a simple barrel charger on it :-/

23:17 <deathmist> I sure hope X2E will be less power hungry to get the expected performance out of it, what's the point of picking up a device with it if it'll just downclock to perform like a lower SKU in practice to not kill battery even when plugged into a wall outlet? :D

23:18 <HdkR> I mean, Linux doesn't downclock at least

23:19 <HdkR> The device may just drop out under long sustained workloads since the battery ran flat :P

23:20 <deathmist> I'd rather just keep the battery at 80% or so while plugged in and performing while expected, bypass charging anyone?

23:21 <HdkR> Well, yes. That would be ideal. but if you're pulling 8w*12 cores you need a 100w power brick to sustain just that...Which the Yoga at least can't slurp

23:22 <HdkR> And that's assuming you're not using anything else on the system :D

23:23 <HdkR> Higher end SOCs than the X1E-78-100 with boost behaviour are likely even worse off

23:37 <deathmist> meh, thing's too power hungry for its own good. wonder why it's being put into all these laptops with "only" 65W max PSU which clearly will never be able to use the full potential without draining a battery at the same time..

23:39 <HdkR> ¯\_(ツ)_/¯