performance decrease with newer kernels

colegu · 2024-04-23 10:13:48

hi all,

i'm running 6.9.0-rc3 , all packages are up to date. yet, over time i'm experiencing lower and lower performance with newer kernels and userland

systemd-analyze gives

Startup finished in 9.116s (firmware) + 3.945s (loader) + 1.300s (kernel) + 16.903s (userspace) = 31.266s 
graphical.target reached after 16.903s in userspace.


kern6.7rc3-v5 
Startup finished in 8.100s (firmware) + 1.552s (loader) + 1.187s (kernel) + 11.947s (userspace) = 22.788s 
graphical.target reached after 11.947s in userspace.

kern6.6.1.1
Startup finished in 8.211s (firmware) + 2.430s (loader) + 1.195s (kernel) + 10.722s (userspace) = 22.560s 
graphical.target reached after 10.722s in userspace.

sure, the worst performance decrease is due to plasma 6 upgrade (kde 5 loaded almost instantly, plasma 6 takes at least 10 seconds)
anyone notices the same?

V1del · 2024-04-23 10:32:02

For plasma try disabling the splash screen. It's on record for causing somewhat of an unnecessary delay (... that should've been present on 5 as well, as I disabled it there for the same reason). Generally speaking I can't say I've really noted anything, and can you actually see this anywhere else other in those 200ms during boot?

Lone_Wolf · 2024-04-23 10:38:49

The loader and firmware parts look to have increased, but kernel only see a minor one . graphical target on the other hand is long even for the 6.1.1 kernel .

To investigate loader & firmware timings we'd need to look at the initramfs / mkinitcpio, kernel timings journal/dmesg and for graphical target systemd services need to be investigated.

To start please post lspci -k , dmesg/journal (run with root rights) and the output of systemd-analyze blame .

colegu · 2024-04-23 11:40:32

V1del wrote:

For plasma try disabling the splash screen. It's on record for causing somewhat of an unnecessary delay (... that should've been present on 5 as well, as I disabled it there for the same reason). Generally speaking I can't say I've really noted anything, and can you actually see this anywhere else other in those 200ms during boot?

kde 5 showed the splash screen for half second only, and then the actual desktop appeared. in plasma 6 it takes quite some time for the kde gears to spin, so i might disable it, and see if it really loads faster

Lone_Wolf wrote:

The loader and firmware parts look to have increased, but kernel only see a minor one . graphical target on the other hand is long even for the 6.1.1 kernel .
To investigate loader & firmware timings we'd need to look at the initramfs / mkinitcpio, kernel timings journal/dmesg and for graphical target systemd services need to be investigated.
To start please post lspci -k , dmesg/journal (run with root rights) and the output of systemd-analyze blame .

yeah, i'm suspecting mostly an userland performance downgrade rather than a kernel issue. but i'm guessing they are a little bit related still. for example hibernate suspend to ram worked like a charm in 6.7.3, but now hibernate sometimes take 2 minutes, and this on a nvme disk, and with only 1gb of ram used , and more often than not the hibernate process just crashes.. but i digress

systemd-analyze blame does not bring any culprits aside networkdmanager which waits for a password in kwallet to start properly

14.055s NetworkManager-wait-online.service
  928ms NetworkManager.service
  673ms systemd-journal-flush.service
  640ms tor.service
  500ms dev-nvme0n1p3.device
  296ms monitorix.service
  240ms systemd-udev-trigger.service
  212ms upower.service

the firmware part i dont think i can speed up. usually i try to avoid having an initramfs in the first place (i always compile my kernel, tuned for my particular system) but i have to load fw for my cyrus logic sound card, otherwise the sound is crap. i'm not sure whether i still need to do this with newer kernels but i might disable the initram and see if that will work

graphical target is the slowest part, and maybe that's also related to plasma 6/x11 performance mostly. games performance is also impacted but maybe this is more related to the nvidia driver/plasma/x11 combo rather than the linux kernel, but the fact of the matter is that i'm trying to speed up the boot process as much as possible, since hibernate and suspend to ram often crash..

thanx for your replies!

seth · 2024-04-23 20:18:39

but now hibernate sometimes take 2 minutes, and this on a nvme disk, and with only 1gb of ram used smile, and more often than not the hibernate process just crashes… hibernate and suspend to ram often crash

Did you confirm the behavior is (still) different w/ eg. the LTS kernel?
Did you test the behavior on the multi-user.target (2nd link below)?
Do you have backtraces or generally logs for the "crashed" hibernation attempts?

colegu · 2024-04-24 14:37:06

unfortunately the log shows nothing. i had to wait 1 min, and then i pressed the power button to force a reboot. nothing fancy in the logs.. (yes, happens the same with LTS kernels,same config), latest kernel that worked well with suspends to disk/ram was 6.7.3....

i haven't tried suspends without the UI. I use my laptop for gaming and random browsing. I don't do CLI coding anymore, I need the desktop lol

Apr 24 17:29:45 njordr systemd-logind[2804]: The system will hibernate now!
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.5752] manager: sleep: sleep requested (sleeping: no  enabled: yes)
Apr 24 17:29:45 njordr asusd[2798]: INFO: RogPlatform reloading panel_od
Apr 24 17:29:45 njordr asusd[2798]: INFO: RogPlatform reloading panel_od
Apr 24 17:29:10 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:78: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:10 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:23: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:28:09 njordr kmix[3676]: org.kde.kmix: No such icon "audio-card-pci"
Apr 24 17:28:09 njordr kmix[3676]: org.kde.kmix: No such icon "audio-card-pci"
Apr 24 17:28:09 njordr kmix[3676]: org.kde.kmix: No such icon "mixer-front"
Apr 24 17:28:09 njordr kmix[3676]: adding 4 mixers
Apr 24 17:28:09 njordr kmix[3676]: mixer "PulseAudio::Playback_Devices:1" "Playback Devices"
Apr 24 17:28:09 njordr kmix[3676]: PulseAudio adding 1 devices
Apr 24 17:28:09 njordr kmix[3676]: adding "alsa_output.pci-0000_00_1f.3.analog-surround-40" "Built-in Audio Analog Surround 4.0"
Apr 24 17:28:09 njordr kmix[3676]: mixer "PulseAudio::Capture_Devices:1" "Capture Devices"
Apr 24 17:28:09 njordr kmix[3676]: PulseAudio adding 1 devices
Apr 24 17:28:09 njordr kmix[3676]: adding "alsa_input.pci-0000_00_1f.3.analog-stereo" "Built-in Audio Analog Stereo"
Apr 24 17:28:09 njordr kmix[3676]: mixer "PulseAudio::Playback_Streams:1" "Playback Streams"
Apr 24 17:28:09 njordr kmix[3676]: PulseAudio adding 1 devices
Apr 24 17:28:09 njordr kmix[3676]: mixer "PulseAudio::Capture_Streams:1" "Capture Streams"
Apr 24 17:28:09 njordr kmix[3676]: PulseAudio adding 0 devices
Apr 24 17:28:09 njordr kmix[3676]: org.kde.kmix: No such icon "audio-card-pci"
Apr 24 17:28:09 njordr kmix[3676]: org.kde.kmix: No such icon "audio-card-pci"
Apr 24 17:28:12 njordr systemd[3329]: Started Chromium - Web Browser.
Apr 24 17:28:13 njordr plasmashell[30770]: [30770:30770:0424/172813.279283:ERROR:chrome_browser_cloud_management_controller.cc(161)] Cloud management controller initialization aborted as CBCM is not enabled.>
Apr 24 17:28:30 njordr plasmashell[30811]: [30811:30811:0424/172830.993544:ERROR:gl_utils.cc(424)] [.WebGL-0x2dec0dc6d500]GL Driver Message (OpenGL, Performance, GL_CLOSE_PATH_NV, High): GPU stall due to Rea>
Apr 24 17:29:06 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:100: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:06 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:78: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:06 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:23: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:07 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:100: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:07 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:78: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:07 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:23: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:08 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:100: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:08 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:78: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:08 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:23: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:10 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:100: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:10 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:78: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:10 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:23: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:19 njordr plasmashell[3438]: file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardlayout/contents/ui/main.qml:100: TypeError: Value is undefined and could not be converted to an object
Apr 24 17:29:45 njordr asusd[2798]: INFO: RogPlatform reloading panel_od
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.5754] device (enp57s0): state change: unavailable -> unmanaged (reason 'sleeping', sys-iface-state: 'managed')
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.5872] device (p2p-dev-wlan0): state change: disconnected -> unmanaged (reason 'sleeping', sys-iface-state: 'managed')
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.5880] manager: NetworkManager state is now ASLEEP
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.5888] device (wlan0): state change: activated -> deactivating (reason 'sleeping', sys-iface-state: 'managed')
Apr 24 17:29:45 njordr kded6[3419]: org.kde.plasma.nm.kded: Unhandled active connection state change:  3
Apr 24 17:29:45 njordr systemd[1]: Starting Network Manager Script Dispatcher Service...
Apr 24 17:29:45 njordr systemd[1]: Started Network Manager Script Dispatcher Service.
Apr 24 17:29:45 njordr kernel: wlan0: deauthenticating from 54:e1:05:7d:3c:99 by local choice (Reason: 3=DEAUTH_LEAVING)
Apr 24 17:29:45 njordr wpa_supplicant[3058]: wlan0: CTRL-EVENT-DISCONNECTED bssid=54:e1:05:7d:3c:99 reason=3 locally_generated=1
Apr 24 17:29:45 njordr wpa_supplicant[3058]: wlan0: CTRL-EVENT-DSCP-POLICY clear_all
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.6796] device (wlan0): supplicant interface state: completed -> disconnected
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.6799] device (wlan0): state change: deactivating -> disconnected (reason 'sleeping', sys-iface-state: 'managed')
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.6802] dhcp4 (wlan0): canceled DHCP transaction
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.6802] dhcp4 (wlan0): activation: beginning transaction (timeout in 45 seconds)
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.6802] dhcp4 (wlan0): state changed no lease
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.6933] device (wlan0): set-hw-addr: set MAC address to 1E:56:66:A0:CE:47 (scanning)
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.7309] device (wlan0): state change: disconnected -> unmanaged (reason 'sleeping', sys-iface-state: 'managed')
Apr 24 17:29:45 njordr NetworkManager[2796]: <info>  [1713968985.7610] device (wlan0): set-hw-addr: reset MAC address to F4:C8:8A:78:A0:F5 (unmanage)
Apr 24 17:29:45 njordr wpa_supplicant[3058]: p2p-dev-wlan0: CTRL-EVENT-DSCP-POLICY clear_all
Apr 24 17:29:45 njordr wpa_supplicant[3058]: p2p-dev-wlan0: CTRL-EVENT-DSCP-POLICY clear_all
Apr 24 17:29:45 njordr wpa_supplicant[3058]: nl80211: deinit ifname=p2p-dev-wlan0 disabled_11b_rates=0
Apr 24 17:29:45 njordr wpa_supplicant[3058]: wlan0: CTRL-EVENT-DSCP-POLICY clear_all
Apr 24 17:29:45 njordr wpa_supplicant[3058]: wlan0: CTRL-EVENT-DSCP-POLICY clear_all
Apr 24 17:29:45 njordr wpa_supplicant[3058]: nl80211: deinit ifname=wlan0 disabled_11b_rates=0
Apr 24 17:29:46 njordr systemd[1]: Reached target Sleep.
Apr 24 17:29:46 njordr systemd[1]: Starting System Hibernate...
Apr 24 17:29:46 njordr systemd-sleep[31626]: Performing sleep operation 'hibernate'...
Apr 24 17:29:46 njordr kernel: PM: hibernation: hibernation entry
Apr 24 17:29:48 njordr ntpd[3127]: Deleting interface #8 wlan0, 192.168.1.139#123, interface stats: received=144, sent=144, dropped=0, active_time=5581 secs
Apr 24 17:29:48 njordr ntpd[3127]: 193.226.12.21 local addr 192.168.1.139 -> <null>
Apr 24 17:29:48 njordr ntpd[3127]: 162.159.200.123 local addr 192.168.1.139 -> <null>
Apr 24 17:29:48 njordr ntpd[3127]: 109.99.63.159 local addr 192.168.1.139 -> <null>
Apr 24 17:29:48 njordr ntpd[3127]: 109.102.183.146 local addr 192.168.1.139 -> <null>
Apr 24 17:29:48 njordr ntpd[3127]: Deleting interface #9 wlan0, fe80::e313:f8c6:2625:68a5%3#123, interface stats: received=0, sent=0, dropped=0, active_time=5581 secs
Apr 24 17:29:48 njordr kernel: Filesystems sync: 0.011 seconds
-- Boot ae5b26ab6e444b7d8c069a04321a6807 --
Apr 24 17:31:12 njordr kernel: Linux version 6.8.7 (bogdan@njordr) (gcc (GCC) 13.2.1 20230801, GNU ld (GNU Binutils) 2.42.0) #1 SMP PREEMPT Tue Apr 23 14:53:53 EEST 2024

seth · 2024-04-24 16:14:42

i pressed the power button to force a reboot

Don't.
https://wiki.archlinux.org/title/Keyboa … el_(SysRq)

i haven't tried suspends without the UI. I use my laptop for…

The point is to narrow down the cause.

And itr

latest kernel that worked well with suspends to disk/ram was 6.7.3

does it still?
(You can downgrade the kernel in isolation, don't forget the headers and/or any OOT module, eg. nvidia or virtualbox, you might use)

colegu · 2024-04-27 15:23:00

well i don't use my system in CLI, so even if it doesn't occur in CLI it still does not matter for me

just had another crash during a normal shutdown... i'll just go back to 6.7.3 at this point and see if that crashes the same...

Apr 27 18:18:35 njordr systemd[3348]: Stopped dbus-:1.2-org.kde.Shutdown@0.service.
Apr 27 18:18:35 njordr systemd[3348]: Removed slice Slice /app/dbus-:1.2-org.kde.Shutdown.
Apr 27 18:18:35 njordr kernel: BUG: kernel NULL pointer dereference, address: 0000000000000022
Apr 27 18:18:35 njordr kernel: #PF: supervisor read access in kernel mode
Apr 27 18:18:35 njordr kernel: #PF: error_code(0x0000) - not-present page
-- Boot 282373eaa75545499849e12a8999b4b1 --

loqs · 2024-04-27 15:25:22

colegu wrote:

well i don't use my system in CLI, so even if it doesn't occur in CLI it still does not matter for me

It would be a data point to help diagnose the cause of the issue.
Edit:

Apr 27 18:18:35 njordr kernel: BUG: kernel NULL pointer dereference, address: 0000000000000022
Apr 27 18:18:35 njordr kernel: #PF: supervisor read access in kernel mode
Apr 27 18:18:35 njordr kernel: #PF: error_code(0x0000) - not-present page

Was there anything in the journal relating to this beyond those three lines?

Last edited by loqs (2024-04-27 15:27:32)

colegu · 2024-04-28 07:17:47

loqs wrote:

colegu wrote:
well i don't use my system in CLI, so even if it doesn't occur in CLI it still does not matter for me
It would be a data point to help diagnose the cause of the issue.
Edit:
Apr 27 18:18:35 njordr kernel: BUG: kernel NULL pointer dereference, address: 0000000000000022
Apr 27 18:18:35 njordr kernel: #PF: supervisor read access in kernel mode
Apr 27 18:18:35 njordr kernel: #PF: error_code(0x0000) - not-present page
Was there anything in the journal relating to this beyond those three lines?

I understand that. but I don't use the system that way. I have to remove all nvidia related stuff just to be able to log into the CLI, and this is quite a hassle

unfortunately no, those are the only lines of code printed before the system completely crashes. most probably an unrelated bug.
just testet kernel 6.7.3 now, and the first hibernate crashed (no logs nothing), booting again with 6.7.3, after plasma 6 loaded, i opened chrome and konsole and system froze. again, no logs.
i suspect the culprit in this case is not the kernel, rather the plasma6/x11 setup, kde 5 was so stable for me, upgrading to plasma 6 has been a really unpleasant experience since. same thing happened with kmail2/akonadi bullshit back in the day. but it was easier to uninstall kmail entirely than it is now to mask all plasma6 related stuff..

so definitely it's not a kernel issue, rather a userland one.

seth · 2024-04-28 07:31:56

I understand that. but I don't use the system that way. I have to remove all nvidia related stuff just to be able to log into the CLI

No?
See the 2nd link below.

so definitely it's not a kernel issue, rather a userland one.

The above test could confirm that.
You can also try to first log out and then shutdown/hibernate from there.

wrt the nullptr reference, you might be hitting the nvidia 550xx bug, https://bbs.archlinux.org/viewtopic.php?id=293400

colegu · 2024-04-29 08:01:27

seth wrote:

I understand that. but I don't use the system that way. I have to remove all nvidia related stuff just to be able to log into the CLI
No?
See the 2nd link below.
so definitely it's not a kernel issue, rather a userland one.
The above test could confirm that.
You can also try to first log out and then shutdown/hibernate from there.
wrt the nullptr reference, you might be hitting the nvidia 550xx bug, https://bbs.archlinux.org/viewtopic.php?id=293400

if i don't completely uninstall nvidia related stuff i'm not able to boot straight into cli, that's why i'm saying that for me this is not a valid option.

the second option i can also try, thanks for the tip!

anout nvidia, i've already downgraded my current version since the latest drivers are not working. i'm unable to open anything with prime-run...

seth · 2024-04-29 08:03:50

if i don't completely uninstall nvidia related stuff i'm not able to boot straight into cli

That sounds completely nonsensical, can you please elaborate how exactly the nvidia module prevents you from booting the multi-user.target?

the latest drivers are not working. i'm unable to open anything with prime-run...

Likewise please don't paraphrase, https://bbs.archlinux.org/viewtopic.php?id=57855

colegu · 2024-04-29 10:19:21

seth wrote:

if i don't completely uninstall nvidia related stuff i'm not able to boot straight into cli
That sounds completely nonsensical, can you please elaborate how exactly the nvidia module prevents you from booting the multi-user.target?
the latest drivers are not working. i'm unable to open anything with prime-run...
Likewise please don't paraphrase, https://bbs.archlinux.org/viewtopic.php?id=57855

instead of the login prompt there is a blank screen in which i can't do nothing.

if i completely uninstall it than it's fine. or i can completely disable framebuffer and have a 640x480 resolution on a 16" screen which is fun as hell ))
but rather than fiddling with this i'm just using X and leave things as they are. don't have the time anymore to troubleshoot things that i'm not using anyways

seth · 2024-04-29 15:49:37

Make sure to enable https://wiki.archlinux.org/title/NVIDIA … de_setting - use the "nvidia_drm.modeset=1" kernel parameter (modprobe.conf won't do!), stay away from fbdev=1 for now (the reports about this are still inconclusive) and then if you still don't get a proper framebuffer, please post your complete system journal for the boot:

sudo journalctl -b | curl -F 'file=@-' 0x0.st

Also try https://aur.archlinux.org/packages/nvidia-535xx-dkms to rule out it's the 550xx cgroupfs bug

colegu · 2024-05-02 07:30:10

i'm not trying to make nvidia work with cli. i'm trying to solve the performance degradation that happens with never kernels and package versions (like plasma 6). as well as the random kernels freezes that occur during hibernate/halt/shutdown..

again, even if i completely uninstall nvidia, and remove X altogether, even if these things will not occur anymore that won't mean anything, because i'm trying to fix things according to my usecase, not some artificial setup that has completely no relevance for me. in fact, installing nvidia drivers in a 100% cli environment has zero relevance. why would you need 3d accelerated graphics in runlevel 3?

seth · 2024-05-02 12:49:09

In order to fix a problem you've to understand the problem you've to isolate the problem you've to eliminate variables.
Also the fact that the framebuffer console doesn't work is a MASSIVE RED FLAG in and by itself and you want to address that just to make sure it doesn't relate to your immediate concern.

I don't know whether you're just trolling or seriously /that/ dense, but if you don't want to identify the problem you don't want to fix it.
Live with it.
Die with it.
Install windows.
Sell the computer and watch scripted reality docu soaps on TV 24/7.

I don't care.

loqs · 2024-05-02 20:28:37

colegu wrote:

again, even if i completely uninstall nvidia, and remove X altogether, even if these things will not occur anymore that won't mean anything, because i'm trying to fix things according to my usecase, not some artificial setup that has completely no relevance for me. in fact, installing nvidia drivers in a 100% cli environment has zero relevance. why would you need 3d accelerated graphics in runlevel 3?

Who supports your use case? The upstream kernel developers do not as nvidia taints the kernel. Do you expect nvidia to support your use case and provide the fix?

colegu · 2024-05-03 16:05:39

loqs wrote:

colegu wrote:
again, even if i completely uninstall nvidia, and remove X altogether, even if these things will not occur anymore that won't mean anything, because i'm trying to fix things according to my usecase, not some artificial setup that has completely no relevance for me. in fact, installing nvidia drivers in a 100% cli environment has zero relevance. why would you need 3d accelerated graphics in runlevel 3?
Who supports your use case? The upstream kernel developers do not as nvidia taints the kernel. Do you expect nvidia to support your use case and provide the fix?

I'm not sure this is an nvidia issue. since i only use the nvidia card when i play games, otherwise x11 starts with the integrated video card. but yes, no one is supposed to 'support' my case.

seth wrote:

In order to fix a problem you've to understand the problem you've to isolate the problem you've to eliminate variables.
Also the fact that the framebuffer console doesn't work is a MASSIVE RED FLAG in and by itself and you want to address that just to make sure it doesn't relate to your immediate concern.
I don't know whether you're just trolling or seriously /that/ dense, but if you don't want to identify the problem you don't want to fix it.
Live with it.
Die with it.
Install windows.
Sell the computer and watch scripted reality docu soaps on TV 24/7.
I don't care.

before updating to plasma6 i've had the same issues. the moment i am loading nvidia drivers i can't use the virtual terminal. i did not mind this behaviour since kde5 worked like a charm. sadly, this is not the case anymore.
why should i invest time in solving something that simply does not bother me.

why should i install windows? will it solve my performance issues?

loqs · 2024-05-03 21:12:02

Compare the progress you are making with https://bbs.archlinux.org/viewtopic.php?id=295199. There is no point in performing the bisection though if you are unwilling to ever disable nvdia. I use nvidia but when I report an upstream kernel issue I disable it.

colegu · 2024-05-04 17:19:53

thank, i'll follow the link you mentioned.

Arch Linux

#1 2024-04-23 10:13:48

performance decrease with newer kernels

#2 2024-04-23 10:32:02

Re: performance decrease with newer kernels

#3 2024-04-23 10:38:49

Re: performance decrease with newer kernels

#4 2024-04-23 11:40:32

Re: performance decrease with newer kernels

#5 2024-04-23 20:18:39

Re: performance decrease with newer kernels

#6 2024-04-24 14:37:06

Re: performance decrease with newer kernels

#7 2024-04-24 16:14:42

Re: performance decrease with newer kernels

#8 2024-04-27 15:23:00

Re: performance decrease with newer kernels

#9 2024-04-27 15:25:22

Re: performance decrease with newer kernels

#10 2024-04-28 07:17:47

Re: performance decrease with newer kernels

#11 2024-04-28 07:31:56

Re: performance decrease with newer kernels

#12 2024-04-29 08:01:27

Re: performance decrease with newer kernels

#13 2024-04-29 08:03:50

Re: performance decrease with newer kernels

#14 2024-04-29 10:19:21

Re: performance decrease with newer kernels

#15 2024-04-29 15:49:37

Re: performance decrease with newer kernels

#16 2024-05-02 07:30:10

Re: performance decrease with newer kernels

#17 2024-05-02 12:49:09

Re: performance decrease with newer kernels

#18 2024-05-02 20:28:37

Re: performance decrease with newer kernels

#19 2024-05-03 16:05:39

Re: performance decrease with newer kernels

#20 2024-05-03 21:12:02

Re: performance decrease with newer kernels

#21 2024-05-04 17:19:53

Re: performance decrease with newer kernels

Board footer