tuxd3v | there are a amdgpu regression on kernel 5.10 | 01:22 |
---|---|---|
tuxd3v | https://gitlab.freedesktop.org/drm/amd/-/issues/1407 | 01:22 |
tuxd3v | basically: | 01:23 |
tuxd3v | /sys/class/drm/card0/device/pp_dpm_mclk | 01:23 |
tuxd3v | is set to maximum frequency even at idle | 01:23 |
tuxd3v | which in turn overheat the graphics card.. | 01:23 |
tuxd3v | the issua was observed in Navi14 | 01:24 |
tuxd3v | but maybe present also in other cards.. | 01:24 |
tuxd3v | so better to check first | 01:24 |
mason | This is why I'm always happy for others to bleed on the edge for me. | 01:37 |
mason | I don't want to discover something like that on my computer. | 01:37 |
tuxd3v | mason, its always nice to have someone on the "edge of the sword", they find bugs and report back, on the other hand, its very time consuming and frustation amounts when problems arise.. | 01:41 |
tuxd3v | :) | 01:41 |
mason | Yes indeed. | 01:42 |
tuxd3v | I still didn't understand why AMD chosen the most powerful performance profile, when the card is in iddle.. | 01:44 |
tuxd3v | and the problem is because the card is in iddle, it doesn't activate the fans :D | 01:45 |
tuxd3v | so only when the card is almost burning is when emergency situation activate the fans :D | 01:45 |
tuxd3v | probably we will see sooner a 5.10.2 point release :D | 01:46 |
clort | everning tuxd3v - how you been? | 01:52 |
tuxd3v | clort, hello, I am fine | 01:53 |
tuxd3v | I have being testing chimaera i386 setup | 01:53 |
tuxd3v | :) | 01:53 |
tuxd3v | and you, how are you? | 01:53 |
clort | good, i am finishing the microSD card IOPS tests | 01:54 |
tuxd3v | also I am around nanopi r2s trying to apply a patch for the uarts.. | 01:54 |
tuxd3v | clort, have you managed to recover your data? | 01:54 |
tuxd3v | you had a sdcard issue | 01:55 |
clort | yes | 01:55 |
clort | all good, stable, backedup | 01:55 |
tuxd3v | nice :) | 01:55 |
clort | i haven't figured out what nvidia is doing with all those partitions yet | 01:55 |
clort | and why extlinux.conf selected boot doesn't boot | 01:55 |
clort | maybe nvidia forces looking for the dtb in the DTB partition | 01:55 |
tuxd3v | my take is that somehow nvidia signed the kernel :( | 01:56 |
clort | i'll let you know if i find out anything i can confirm | 01:56 |
clort | ahh | 01:56 |
tuxd3v | I don't know, its just a guess | 01:56 |
tuxd3v | maybe | 01:56 |
tuxd3v | the bootloader is in the SPI flash | 01:57 |
clort | it is something to investigate yes. i have seen in nvidia forums other people replacing kernel | 01:57 |
clort | so i just need to study and lear | 01:57 |
clort | n | 01:57 |
tuxd3v | and they succeded just replacing it? | 01:57 |
clort | (and clean my keyboard) | 01:57 |
clort | i do't know, i saw they are 'shipping product' with their own kernel but don't see how they did it | 01:58 |
tuxd3v | humm | 01:58 |
tuxd3v | check if they have that crazy amount of partitions like with oficial release | 01:58 |
clort | they are just posting in forum, no detailed blogging of how they did it. but i will report if i find aything i can cofirm | 01:59 |
tuxd3v | Android images usually have tons of partitions, Nvidia seems to like that | 01:59 |
clort | hehe | 01:59 |
tuxd3v | :) | 01:59 |
tuxd3v | We will eventually get there | 02:04 |
gnarface | tuxd3v: thanks for the warning about the 5.10 amdgpu regression | 02:18 |
gnarface | tuxd3v: currently not observing that on 5.9 in ceres | 02:18 |
tuxd3v | gnarface, you welcome | 02:18 |
tuxd3v | i believe in a matter of some 2 weeks or so, devuan will get it , and so do devuan | 02:19 |
tuxd3v | I believe a 5.10.2 point release will have to be issued | 02:19 |
mason | Oh, right, it'll percolate down. I'd forgotten that. | 02:20 |
* gnarface also removing linux-image-amd64 and linux-headers-amd64 so he doesn't get it accidentally | 02:20 | |
mason | I only ever think in terms of stable nowadays. | 02:21 |
mason | gnarface: Rather than doing that, you can apt hold those. | 02:21 |
gnarface | noted | 02:21 |
AbsolutZer0 | Hi I'm a noob running kubuntu but I'd rather not be, is Steam on Devuan possible? Or does it take a lot of tweaking? | 06:54 |
AbsolutZer0 | I don't know why you people just sit here without answering a simple question, fuck it I'll find out for myself. | 07:04 |
tarzeau | some sleep, different timezone, some are not interested to answer :) | 07:21 |
rwp | Empirical evidence that the patience level was exactly 10 minutes. | 07:28 |
tarzeau | </telepathy> | 07:45 |
KREYREEN | heads-up https://github.com/Debian/README.Debian/issues/9 i think that devuan has the same issue | 09:46 |
clort | with btrfs, mount -o ssd enables SSD-optimized allocation schemes | 17:00 |
mason | tarzeau: I think I answered the fellow in the dream I was having at the time. | 18:17 |
tarzeau | mason: hehe | 18:48 |
* sammi` hoho | 19:04 | |
anjan | Hi, Im on devuan testing. I am having suspend issues. This is expected because I always have suspend issues on 5.9.0 (even when I was running arch). On arch, I was able to fix it by installing linux-lts ( currently 5.4.76). Is there anyway I can install the 5.4 kernel? The 4.19 kernel is too old for my liking | 19:53 |
mason | anjan: https://wiki.debian.org/BuildADebianKernelPackage | 19:58 |
anjan | ah thanks | 19:59 |
anjan | is there anyway to get alerted when a new lts kernel is available? | 19:59 |
mason | LKML? I'm not sure there's a good method if you're rolling your own. | 19:59 |
anjan | maybe I should look into fixing suspend on the new kernel | 20:00 |
mason | If you can do that, it'd be a good option. You'll still have rebuilds unless it's a fix you can implement in configuration somehow. | 20:01 |
anjan | maybe Ill cope with 4.19 | 20:01 |
anjan | but it's so ancient ;_; | 20:01 |
anjan | Ive tried all the other fixes on the archwiki turns out | 20:01 |
mason | It's only ancient if there's distinct functionality you need that it lacks. | 20:02 |
anjan | thats true | 20:02 |
mason | And there might actually be, but I don't have any way to know that. :) | 20:02 |
anjan | I think the newer kernels had better vfio performance | 20:02 |
mason | Hrm. I'd suggest that a machine that has a workload that cares about that might not be in the same set of machines as one that needs to suspend. | 20:03 |
anjan | mason: it's a desktop that I use for pci passthrough sometimes | 20:03 |
mason | Hrm, so that's a valid desire then. | 20:04 |
mason | Dunno. If 4.19 ends up slowing down your VM noticeably it might take more looking, but I'd just run the stable release if it's not something you can notice. | 20:05 |
anjan | yup sounds good mason | 20:05 |
anjan | thanks | 20:05 |
mason | I might also not have the perspective you want. I like old things, and if they're slow, all the better. :P | 20:06 |
anjan | I think I just have to learn to stop about version numbers in programs | 20:06 |
anjan | Ive been using arch for toooooo long | 20:06 |
mason | You know about Artix, yes? | 20:07 |
anjan | I dont like artix | 20:07 |
anjan | I never liked the AUR | 20:07 |
mason | kk | 20:07 |
anjan | I used and loved void | 20:07 |
anjan | but then there was a break in pci passthrough | 20:08 |
anjan | and it took forever to fix cause ovmf requires openssl and void uses libressl | 20:08 |
mason | Oh, I thought you could resolve that with source builds. But I've never done more than initial test installs, so I could be confused. | 20:08 |
anjan | mason: ya, thats the workaround | 20:09 |
anjan | but then python5.8 came and it broke it agian | 20:09 |
anjan | and I was so upset, I needed something stable. So.... here we are | 20:09 |
mason | Mm, Python. <sarcasm> | 20:09 |
anjan | I just want a distro that fits my needs ;_l | 20:10 |
anjan | Im seriously considering gentoo | 20:10 |
mason | Look at LiGurOS too if you're looking in that direction. | 20:10 |
mason | Then again, Devuan is what I most like/want, and I bet it'd work out for you. I mean, here we are, after all. | 20:11 |
anjan | as long as devuan is more stable than everything else, Im happy with it | 20:11 |
anjan | I tried making a package for devuan last night....It was not fun though | 20:12 |
mason | APT packaging isn't the most trivial thing in the world. | 20:12 |
mason | Depending on what you're doing, there's also stuff like GNU Stow. I've used that a lot in the past for one-off installations where I just want to be able to easily upgrade from time to time. | 20:12 |
anjan | I maintain a number of packages for alpine and void | 20:12 |
anjan | ah ive used gnu stow | 20:15 |
anjan | mason: alright I found a fix for my suspend issue: just open tty2 and type in loginctl suspend | 21:35 |
anjan | dont type in loginctl suspend on the tty with x open | 21:37 |
anjan | Im using the latest kernel | 21:43 |
clort | i dunno this | 21:45 |
mason | anjan: oh, weird... that doesn't sound like a kernel issue at all. | 23:32 |
mason | anjan: I wouldn't probably recommend using loginctl, anyway. It's part of systemd. You can install pm-utils and say pm-suspend and similar. | 23:33 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!