Sorry if you've seen me here before...

What I'm trying to do, currently: exclude certain processes from swap to test whether they keep the desktop lag-free during high memory pressure.

From /proc/$$/smaps_rollup I identified what processes are using swap and what sounds GUI-related. From those, plasmashell was the largest, with kwin_wayland in second.

I'm creating a Cgroup override in using systemctl --user edit plasma-plasmashell.service and systemctl --user edit plasma-kwin_wayland.service.

[Service]
MemorySwapMax=0
MemoryZSwapMax=0

This seems to work with plasmashell just fine, with its entire 250M residential set staying on RAM. However, I checked kwin_wayland and it has about 50M in swap -- which makes it impossible for me to rule my hypothesis out.

Using this script to check Swap usage:

spoiler

#!/bin/bash

get_swap() {
  pid=$1
  output=$(sudo cat "/proc/$pid/smaps_rollup" 2>/dev/null | grep -m 1 Swap: | awk '{ print $2 }')
  if [ "$output" = "" ]; then
        echo 0
  else
        echo "$output"
  fi
}

ps ax -o pid,cmd --no-headers | while IFS= read -r line; do
        pid=$(echo "$line" | awk '{ print $1 }')
        cmd=$(echo "$line" | awk '{ $1 = ""; print $0 }')
        swapkb=$(get_swap "$pid")
        if [ "$swapkb" = "0" ]; then
                continue
        fi
        echo "$swapkb $pid ${cmd:0:30}"
done | sort -n

~> ./ps_swap.sh | grep -E 'plasma|kwin'
2208 3117  /usr/bin/kwin_wayland_wrapper
6972 3041  /usr/bin/startplasma-wayland
19752 3121  /usr/bin/kwin_wayland --wayla
cat /etc/sysctl.conf 
vm.swappiness=10
vm.vfs_cache_pressure=150

ZSwap is enabled.

systemctl --user show plasma-kwin_wayland.service

spoiler

Id=plasma-kwin_wayland.service
Names=plasma-kwin_wayland.service
Requires=basic.target session.slice dbus.socket
PartOf=graphical-session.target
RequiredBy=plasma-workspace-wayland.target
BoundBy=plasma-workspace-wayland.target
Conflicts=shutdown.target
Before=drkonqi-coredump-launcher.socket plasma-workspace-wayland.target shutdown.target plasma-core.target plasma-ksmserver.service plasma-kwallet-pam.service plasma-kded6.service
After=basic.target dbus.socket session.slice home.mount -.mount
WantsMountsFor=/home/spk
Description=KDE Wayland Compositor
LoadState=loaded
ActiveState=active
FreezerState=running
SubState=running
FragmentPath=/usr/lib/systemd/user/plasma-kwin_wayland.service
DropInPaths=/etc/systemd/user/plasma-kwin_wayland.service.d/noswap.conf
UnitFileState=static
UnitFilePreset=disabled
StateChangeTimestamp=Mon 2026-11-16 04:34:07 EST
StateChangeTimestampMonotonic=50088438
InactiveExitTimestamp=Mon 2026-11-16 04:34:07 EST
InactiveExitTimestampMonotonic=50071047
ActiveEnterTimestamp=Mon 2026-11-16 04:34:07 EST
ActiveEnterTimestampMonotonic=50088438
ActiveExitTimestampMonotonic=0
InactiveEnterTimestampMonotonic=0
CanStart=yes
CanStop=yes
CanReload=no
CanIsolate=no
CanFreeze=yes
CanLiveMount=no
StopWhenUnneeded=no
RefuseManualStart=no
RefuseManualStop=no
AllowIsolate=no
DefaultDependencies=yes
SurviveFinalKillSignal=no
OnSuccessJobMode=fail
OnFailureJobMode=replace
IgnoreOnIsolate=no
NeedDaemonReload=no
JobTimeoutUSec=infinity
JobRunningTimeoutUSec=infinity
JobTimeoutAction=none
ConditionResult=yes
AssertResult=yes
ConditionTimestamp=Mon 2026-11-16 04:34:07 EST
ConditionTimestampMonotonic=50070246
AssertTimestamp=Mon 2026-11-16 04:34:07 EST
AssertTimestampMonotonic=50070269
Transient=no
Perpetual=no
StartLimitIntervalUSec=10s
StartLimitBurst=5
StartLimitAction=none
FailureAction=none
SuccessAction=none
InvocationID=a65c739cb449408da1b812bbc3828796
CollectMode=inactive
DebugInvocation=no
Type=dbus
ExitType=main
Restart=no
RestartMode=normal
NotifyAccess=none
RestartUSec=100ms
RestartSteps=0
RestartMaxDelayUSec=infinity
RestartUSecNext=100ms
TimeoutStartUSec=1min 30s
TimeoutStopUSec=1min 30s
TimeoutAbortUSec=1min 30s
TimeoutStartFailureMode=terminate
TimeoutStopFailureMode=terminate
RuntimeMaxUSec=infinity
RuntimeRandomizedExtraUSec=0
WatchdogUSec=0
WatchdogTimestampMonotonic=0
RootDirectoryStartOnly=no
RemainAfterExit=no
GuessMainPID=yes
MainPID=3134
ControlPID=0
BusName=org.kde.KWinWrapper
FileDescriptorStoreMax=0
NFileDescriptorStore=0
FileDescriptorStorePreserve=restart
StatusErrno=0
Result=success
ReloadResult=success
CleanResult=success
LiveMountResult=success
UID=[not set]
GID=[not set]
NRestarts=0
OOMPolicy=stop
ReloadSignal=1
ExecMainStartTimestamp=Mon 2026-11-16 04:34:07 EST
ExecMainStartTimestampMonotonic=50070784
ExecMainExitTimestampMonotonic=0
ExecMainHandoffTimestamp=Mon 2026-11-16 04:34:07 EST
ExecMainHandoffTimestampMonotonic=50074908
ExecMainPID=3134
ExecMainCode=0
ExecMainStatus=0
ExecStart={ path=/usr/bin/kwin_wayland_wrapper ; argv[]=/usr/bin/kwin_wayland_wrapper --xwayland ; ignore_errors=no ; start_time=[Mon 2026-11-16 04:34:07 EST] ; stop_time=[n/a] ; pid=3134 ; code=(null) ; status=0/0 }
ExecStartEx={ path=/usr/bin/kwin_wayland_wrapper ; argv[]=/usr/bin/kwin_wayland_wrapper --xwayland ; flags= ; start_time=[Mon 2026-11-16 04:34:07 EST] ; stop_time=[n/a] ; pid=3134 ; code=(null) ; status=0/0 }
Slice=session.slice
ControlGroup=/user.slice/user-1000.slice/user@1000.service/session.slice/plasma-kwin_wayland.service
ControlGroupId=9462
MemoryCurrent=[not set]
MemoryPeak=[not set]
MemorySwapCurrent=[not set]
MemorySwapPeak=[not set]
MemoryZSwapCurrent=[not set]
MemoryAvailable=16312848384
EffectiveMemoryMax=16312848384
EffectiveMemoryHigh=16312848384
CPUUsageNSec=148705090000
TasksCurrent=44
EffectiveTasksMax=18590
IPIngressBytes=[no data]
IPIngressPackets=[no data]
IPEgressBytes=[no data]
IPEgressPackets=[no data]
IOReadBytes=[not set]
IOReadOperations=[not set]
IOWriteBytes=[not set]
IOWriteOperations=[not set]
OOMKills=0
ManagedOOMKills=0
Delegate=no
CPUWeight=[not set]
StartupCPUWeight=[not set]
CPUQuotaPerSecUSec=infinity
CPUQuotaPeriodUSec=infinity
IOAccounting=no
IOWeight=[not set]
StartupIOWeight=[not set]
MemoryAccounting=yes
MemoryMin=0
MemoryLow=0
StartupMemoryLow=0
MemoryHigh=infinity
StartupMemoryHigh=infinity
MemoryMax=infinity
StartupMemoryMax=infinity
MemorySwapMax=0
StartupMemorySwapMax=infinity
MemoryZSwapMax=0
StartupMemoryZSwapMax=infinity
MemoryZSwapWriteback=yes
DevicePolicy=auto
TasksAccounting=yes
TasksMax=18590
IPAccounting=no
ManagedOOMSwap=auto
ManagedOOMMemoryPressure=auto
ManagedOOMMemoryPressureLimit=0
ManagedOOMMemoryPressureDurationUSec=[not set]
ManagedOOMPreference=none
MemoryPressureWatch=auto
MemoryPressureThresholdUSec=200ms
CPUPressureWatch=auto
CPUPressureThresholdUSec=200ms
IOPressureWatch=auto
IOPressureThresholdUSec=200ms
CoredumpReceive=no
UMask=0022
LimitCPU=infinity
LimitCPUSoft=infinity
LimitFSIZE=infinity
LimitFSIZESoft=infinity
LimitDATA=infinity
LimitDATASoft=infinity
LimitSTACK=infinity
LimitSTACKSoft=8388608
LimitCORE=infinity
LimitCORESoft=infinity
LimitRSS=infinity
LimitRSSSoft=infinity
LimitNOFILE=524288
LimitNOFILESoft=1024
LimitAS=infinity
LimitASSoft=infinity
LimitNPROC=61969
LimitNPROCSoft=61969
LimitMEMLOCK=8388608
LimitMEMLOCKSoft=8388608
LimitLOCKS=infinity
LimitLOCKSSoft=infinity
LimitSIGPENDING=61969
LimitSIGPENDINGSoft=61969
LimitMSGQUEUE=819200
LimitMSGQUEUESoft=819200
LimitNICE=0
LimitNICESoft=0
LimitRTPRIO=0
LimitRTPRIOSoft=0
LimitRTTIME=infinity
LimitRTTIMESoft=infinity
WorkingDirectory=!/home/spk
RootEphemeral=no
OOMScoreAdjust=200
CoredumpFilter=0x33
Nice=0
IOSchedulingClass=2
IOSchedulingPriority=4
CPUSchedulingPolicy=0
CPUSchedulingPriority=0
CPUAffinityFromNUMA=no
NUMAPolicy=n/a
TimerSlackNSec=50000
CPUSchedulingResetOnFork=no
NonBlocking=no
StandardInput=null
StandardOutput=journal
StandardError=inherit
TTYReset=no
TTYVHangup=no
TTYVTDisallocate=no
SyslogPriority=30
SyslogLevelPrefix=yes
SyslogLevel=6
SyslogFacility=3
LogLevelMax=-1
LogRateLimitIntervalUSec=0
LogRateLimitBurst=0
SecureBits=0
CapabilityBoundingSet=cap_chown cap_dac_override cap_dac_read_search cap_fowner cap_fsetid cap_kill cap_setgid cap_setuid cap_setpcap cap_linux_immutable cap_net_bind_service cap_net_broadcast cap_net_admin cap_net_raw cap_ipc_lock cap_ipc_owner cap_sys_module cap_sys_rawio cap_sys_chroot cap_sys_ptrace cap_sys_pacct cap_sys_admin cap_sys_boot cap_sys_nice cap_sys_resource cap_sys_time cap_sys_tty_config cap_mknod cap_lease cap_audit_write cap_audit_control cap_setfcap cap_mac_override cap_mac_admin cap_syslog cap_wake_alarm cap_block_suspend cap_audit_read cap_perfmon cap_bpf cap_checkpoint_restore
DynamicUser=no
SetLoginEnvironment=no
RemoveIPC=no
PrivateTmp=no
PrivateTmpEx=no
PrivateDevices=no
ProtectClock=no
ProtectKernelTunables=no
ProtectKernelModules=no
ProtectKernelLogs=no
ProtectControlGroups=no
ProtectControlGroupsEx=no
PrivateNetwork=no
PrivateUsers=no
PrivateUsersEx=no
PrivateMounts=no
PrivateIPC=no
PrivatePIDs=no
ProtectHome=no
ProtectSystem=no
SameProcessGroup=no
UtmpMode=init
IgnoreSIGPIPE=yes
NoNewPrivileges=no
SystemCallErrorNumber=2147483646
LockPersonality=no
RuntimeDirectoryPreserve=no
RuntimeDirectoryMode=0755
StateDirectoryMode=0755
StateDirectoryAccounting=no
StateDirectoryQuota=[not set]
CacheDirectoryMode=0755
CacheDirectoryAccounting=no
CacheDirectoryQuota=[not set]
LogsDirectoryMode=0755
LogsDirectoryAccounting=no
LogsDirectoryQuota=[not set]
ConfigurationDirectoryMode=0755
TimeoutCleanUSec=infinity
MemoryDenyWriteExecute=no
RestrictRealtime=no
RestrictSUIDSGID=no
RestrictNamespaces=no
DelegateNamespaces=18446744073709551615
MountAPIVFS=no
BindLogSockets=no
KeyringMode=inherit
ProtectProc=default
ProcSubset=all
ProtectHostname=no
PrivateBPF=no
MemoryKSM=no
MemoryTHP=inherit
RootImagePolicy=root=verity+signed+encrypted+encryptedwithintegrity+unprotected+absent:usr=verity+signed+encrypted+encryptedwithintegrity+unprotected+absent:home=encrypted+encryptedwithintegrity+unprotected+absent:srv=encrypted+encryptedwithintegrity+unprotected+absent:tmp=encrypted+encryptedwithintegrity+unprotected+absent:var=encrypted+encryptedwithintegrity+unprotected+absent
MountImagePolicy=root=verity+signed+encrypted+encryptedwithintegrity+unprotected+absent:usr=verity+signed+encrypted+encryptedwithintegrity+unprotected+absent:home=encrypted+encryptedwithintegrity+unprotected+absent:srv=encrypted+encryptedwithintegrity+unprotected+absent:tmp=encrypted+encryptedwithintegrity+unprotected+absent:var=encrypted+encryptedwithintegrity+unprotected+absent
ExtensionImagePolicy=root=verity+signed+encrypted+encryptedwithintegrity+unprotected+absent:usr=verity+signed+encrypted+encryptedwithintegrity+unprotected+absent:home=encrypted+encryptedwithintegrity+unprotected+absent:srv=encrypted+encryptedwithintegrity+unprotected+absent:tmp=encrypted+encryptedwithintegrity+unprotected+absent:var=encrypted+encryptedwithintegrity+unprotected+absent
KillMode=control-group
KillSignal=15
RestartKillSignal=15
FinalKillSignal=9
SendSIGKILL=yes
SendSIGHUP=no
WatchdogSignal=6

OS: openSUSE Tumbleweed x86_64
Host: 83HM (Yoga Slim 7 15ILL9)
Kernel: Linux 7.3.1-1-default
Display (M27Q P): 2560x1440 in 27", 144 Hz [External]
Desktop Environment: KDE Plasma 6.8.2
Window Manager: KWin (Wayland)
CPU: Intel(R) Core(TM) Ultra 7 256V (8) @ 4.80 GHz
GPU: Intel Core Ultra 200V Series Processors Arc Graphics 130V/140V GPU @ 1.95 GHz [Integrated]
Memory: 5.34 GiB / 15.19 GiB (35%)
Swap: 0 B / 15.20 GiB (0%)
 

edit: I decreased swappiness and increased vfs cache pressure to reduce the amount of swapping in general.

I'm noticing that a very intended behavior of Zswap is to write pages to disk when they don't cleanly compress (or for some other reason). How can I configure it to act kind of like Zram but only writeback when the system is actually out of RAM?

grep -R . /sys/kernel/debug/zswap/
/sys/kernel/debug/zswap/stored_incompressible_pages:8224
/sys/kernel/debug/zswap/stored_pages:218538
/sys/kernel/debug/zswap/pool_total_size:229318656
/sys/kernel/debug/zswap/written_back_pages:251371
/sys/kernel/debug/zswap/decompress_fail:0
/sys/kernel/debug/zswap/reject_compress_poor:0
/sys/kernel/debug/zswap/reject_compress_fail:0
/sys/kernel/debug/zswap/reject_kmemcache_fail:0
/sys/kernel/debug/zswap/reject_alloc_fail:0
/sys/kernel/debug/zswap/reject_reclaim_fail:0
/sys/kernel/debug/zswap/pool_limit_hit:0

My parameters: zswap.enabled=1 zswap.compressor=zstd zswap.shrinker_enabled=0 zswap.max_pool_percent=20. Swappiness is at default value.

 

edit for this wondering, the reason is:

Also something called GTT slab exists?

original post: This is bullshit. I wanted to play Terraria last night but I looked at my RAM usage and it was sitting uncomfortably high at 13 GB (of my 15). So, I close Terraria, and it drops about 2 GB. I close Firefox, Thunderbird, and Zed, and that drops about 2.5 GB. Then I close basically everything and it's still super high! Each time this happens, I have to restart my computer in order to get the RAM usage lower.

Whenever I look this up it's usually two camps:

  • KDE is just bloated.
  • Unused RAM is wasted RAM.

To address both of these:

I disabled a few things in KDE like file indexing (which was my home directory). Cool, so that saved like 100 MB. KDE is supposed to idle at around 3GB anyways, so what now?

The "unused RAM" belief is like a folktale at this point. It's not true anymore, and buffered/cache memory is shown separately. Plus, I can drop those caches whenever I want... Which I did. File inode cache only really became a problem for me when I was doing Minecraft Datapack development and I was creating hundreds of files on a regular basis as part of hot reload. Then I'd have to drop the entire cache because it wasn't being free'd...

However, the RAM I'm complaining about is actually not that, because it steps on my available RAM meter. Having a game freeze up because I was out of memory is a pain in the ass.

I even tried installing Zram, which only made matters worse since the active memory allocations seemed to be robbing the CPU of time it would be running game code, just so it could compress file pages. (My system has also just had major lag problems with swap, in general, during high memory pressure, which I should probably report as a bug). My swappiness was probably way too high, but regardless I didn't see major improvements from that and it only seemed like a bandaid to the real problem. Also, why should I need to use swap when I have 16 GB of RAM and only a few programs open at a time?

This whole thing is pretty annoying. Whenever I have like 11 hours of uptime my memory is unusually high. I also can't seem to track it down to any particular application...

System: OpenSUSE Tumbleweed

 

Bought from goHardDrive. I've already had a drive fail and require replacement, and that was with pretty much no use. It just failed on a smart test. The same drive, maybe, has already failed again. I'm going to nag em because of the warranty, but it's pretty worrying. Like, I'm sure failing a scrub isn't the end of the world, but it's also not a good thing, right?

 

I currently use Thunderbird to organize my tasks. I have them pinned to a calendar by their due-date and in a side panel too. However, a major limitation of Thunderbird as a whole is the fact that I cannot attach a start date to a task without it appearing several times on my calendar for each day that it can be worked on. This is pretty annoying and I only want it to appear on the calendar for the date it's due. I also want my task pane to sort of show the number of days it's due, whether it can be started yet, and if not, how many days until it can be.

Afaik not many task organizers can do something this complex. I've used KOrganizer and its forks in the past but there was too much friction in using it that I switched to Thunderbird.

 

EDIT: Ok of course as soon as I post this I get a trident. But, I definitely had to change my strategy up. I found a river biome at night and used a rinse-and-repeat strategy of generated as many drowned, killing any trident holders, then going really high up with an elytra and despawning them. Still, I had to kill like 7 more trident holders before getting one, which only reinforces how unlucky I was (<5% worst chances). That is---10% of people have killed as many drowned as I have without getting a trident.

Original:

Not a support thread. I've killed like 15 drowned holding tridents with a Fire Aspect II, Smite V, Looting III sword and gotten zero tridents. The wiki claims 11.5% is the success rate with my enchantments, but plugging that into a binomial distribution puts my case in the unlucky 15%. I refuse to cave in and build a useless farm for just this item. (I've found 4 trident-holding drowned in the last hour, so the number is higher).

It also doesn't help at all that drowned do not spawn in our Realm, unless you find some underwater ruins. It might be due to some chunk loading elsewhere from a Zombie Piglin farm (I blame my friend), but surely that's more than 15 chunks away? Ugh!

Fucking Mojang...

Cool pic provided as a reward for reading that drivel

 

The enhanced version of ClassiCube seems to handle the low amounts of VRAM better than the Classic version, which is nearly unplayable. It seems like that’s where the actual development is happening.

You can change controls, use resource packs, join multiplayer, chat in public, save worlds, place liquids, custom blocks, change skins, go into third person, fly and speed hack, change graphical settings, and play custom plugins on servers. Cons are that download speeds on servers can be low (resource packs are 2012 speeds) and you can sometimes run out of sysRAM or VRAM while joining a custom world once leaving a lobby.

The game also freezes up at times during loading. This could be helped by using a second thread to actually do this work and give the user a more interactive progress screen for any behind the scenes stuff, which is currently opaque. Besides that, it’s really impressive, and much more than what I expected from the old 3DS with only 128 MB RAM and 6 MB VRAM.

view more: next ›