Re: Performance difference
On Friday December 23 2022 12:38:17 Patrick Nagel wrote: >(probably 200-300ms of these "real" numbers is me reacting slowly to the >window popping up and pressing ALT-F4...) A thought: download the AppImage version of a larger application, like Firefox on both systems, and do ``` > time /path/to/firefox*.AppImage --version ``` This will unpack the entire appimage contents to a temporary directory and launch the application, excluding any KDE-specifics and also network activity but above all the human reaction component. Should be a good test of your disk subsystem. NB: the `--version` trick works for KDE apps too of course: ``` > time dolphin5 --version dolphin 20.07.70 1.074 user_cpu 0.257 kernel_cpu 0:04.02 total_time 32.8%CPU {46064M 898F 4110R 29529I 12O 840w 359c} > time dolphin5 --version dolphin 20.07.70 0.759 user_cpu 0.138 kernel_cpu 0:00.87 total_time 101.1%CPU {46068M 0F 4400R 3649I 4O 446w 167c} > time kate5 --version kate 19.08.3 0.920 user_cpu 0.249 kernel_cpu 0:02.73 total_time 42.4%CPU {45640M 380F 4320R 11687I 4O 775w 220c} > time kate5 --version kate 19.08.3 0.739 user_cpu 0.159 kernel_cpu 0:00.86 total_time 102.3%CPU {45488M 0F 4431R 3639I 4O 498w 185c} ``` Start up delays on my much slower system (that's an Intel N3150 CPU!) sometimes *feel* like 20 seconds but clearly they're much less, so yeah, you may have an issue somewhere. Have you checked your syslog and/or checked the SMART health status of the drive in your slower system? Does the delay depend on how many other applications you're running (i.e. how much free RAM you have)? R.
Re: Performance difference
Hi there, On Friday, 23 December 2022 01:40:18 CET Dave Close wrote: [...] > The only difference I've found thus far is that the slower one does not > have a swap partition. Any suggestions for what else I can check, and > hopefully fix, will be appreciated. [...] You could try swapping SSDs and see whether these delays still happen on the "other" hardware. I would suspect some kind of misconfiguration on the KDE system. 20 second delays could be some kind of network timeout for example. On my Kubuntu 22.04 box I don't see noticeable delays when starting applications: > $ time dolphin > > real0m0.665s > user0m0.240s > sys 0m0.077s > $ time kcalc > > real0m0.547s > user0m0.116s > sys 0m0.034s (probably 200-300ms of these "real" numbers is me reacting slowly to the window popping up and pressing ALT-F4...) Cheers, Patrick.
Re: Performance difference
On Thursday December 22 2022 16:40:18 Dave Close wrote: >The only difference I've found thus far is that the slower one does not >have a swap partition. Any suggestions for what else I can check, and >hopefully fix, will be appreciated. Have you tried timing the performance difference of the 2nd time you start that command? KDE apps tend to have a lot of libraries to load with lots of symbols to retrieve (I have the impression that number is only going up with every new C++ standard and Qt version). Provided you have enough RAM those libraries will be in some kind of cache, allowing for a much faster start-up. I see this myself, on my lowly, ageing beater which runs off a ZFS pool on an SSHD. I think the SSD part of that drive is probably toast by now so I no longer have the benefits of that, so loading a command for the first time clearly requires a lot of "disk scraping". I assume your 2 laptops have SSDs (of comparable performance) so that filesystem fragmentation is mostly irrelevant? BTW, you do use the same filesystem (and filesystem settings) on both? R.
Performance difference
I have two laptops with nearly identical hardware specs, both running Fedora 36 with KDE and completely up-to-date on all software. Both start in systemd's multi-user.target, KDE starts with a startx command, Both are running Xorg ($XDG_SESSION_TYPE=""). In non-graphics mode, their performance seems identical, quite snappy. However, they do not perform similarly when KDE is running. Starting an application or opening a new window involves a noticeable delay, sometimes as long as 20 seconds, even if nothing else is happening at that time. The only difference I've found thus far is that the slower one does not have a swap partition. Any suggestions for what else I can check, and hopefully fix, will be appreciated. Here's some details on the two machines. faster oneslower one CPU (4 cores)AMD Ryzen 3 3200U withAMD Ryzen 3 3250U with Radeon Vega Mobile GfxRadeon Graphics Swap zram0, sda3 zram0 fast.com speed 290 Mbps 290 Mbps bogomips 5190.47 5189.93 MemTotal 3427804 kB3404868 kB MemFree381772 kB 483060 kB MemAvailable 944388 kB1007680 kB Buffers 4 kB 56 kB Cached 751824 kB 702128 kB SwapCached 1776 kB 2180 kB Active 546260 kB 684744 kB Inactive 1270360 kB1112104 kB Active(anon)99512 kB 189148 kB Inactive(anon) 971252 kB 907516 kB Active(file) 446748 kB 495596 kB Inactive(file) 299108 kB 204588 kB Unevictable 14676 kB176 kB Mlocked 14676 kB176 kB SwapTotal12068856 kB3404796 kB SwapFree 11531256 kB2906364 kB Dirty2108 kB 0 kB Writeback 0 kB 4 kB AnonPages 102 kB1092708 kB Mapped 93480 kB 93088 kB Shmem2280 kB 2000 kB KReclaimable51176 kB 49952 kB Slab 16 kB 176112 kB SReclaimable51176 kB 49952 kB SUnreclaim 113268 kB 126160 kB KernelStack 15904 kB 13064 kB PageTables 37420 kB 33732 kB CommitLimit 13782756 kB5107228 kB Committed_AS 7497304 kB6170592 kB VmallocUsed520416 kB 527988 kB Percpu 15488 kB 14464 kB DirectMap4k351100 kB 363212 kB DirectMap2M 3254272 kB3219456 kB -- Dave Close, Compata, Irvine CA +1 714 434 7359 d...@compata.com dhcl...@alumni.caltech.edu "Any idiot can face a crisis; it's the day-to-day living that wears you out." -- Anton Chekhov