topic title: boycott systemd
Posts: 4,164
rokytnji
Joined: 20 Feb 2009
#16
By the way, I thought this forum was for the discussion of non-Linux subjects.
The only rules in this section is to be polite, non demanding, and non insulting.

Other than that.

I am not too rigid on what can be started and discussed here.
I know the description says non linux subjects.
But believe or not. I can be flexible. Semper Gumbi.
I like a good grin like anybody else.
But they never trusted this six foot seven inch high schooler as a hall monitor so I do not
stick to the rules too tightly.

As far as opinions go.
Mostly.

They can be about as useful as a ash tray on a motorcycle. __{{emoticon}}__
Posts: 127
KrunchTime
Joined: 05 Dec 2014
#17
Thanks, that's good to know. I was wondering where one could post other miscellaneous stuff. The CrunchBang forums offer an Off Topic forum where all else goes, but politics are a no no over there.
Posts: 604
thriftee
Joined: 27 Feb 2009
#18
I timed it with Manjaro tonight. 52 seconds with OpenRc and 64 seconds with Systemd. So much for the myth that systemd is faster. In fact, it's quite a bit slower, not even close.
Posts: 1,445
skidoo
Joined: 09 Feb 2012
#19
I've never understood the fascination with, the fixation on, boot times.
Anyhow, I've seen proof-is-in-the-pudding youTube videos demonstrating that systemd can, in a given case, be noticeably faster.
Benchmarking the default configuration, per debian, is a non-ideal scenario.

systemd aside, too many of the folks whining about boot times neglect to remove unused kernel modules and eliminate any unneeded startup services.
Default configuration == intended best starting point for most users == unlikely to be the best config for your specific use.

Okay. Yours is bigger than mine. But mine runs 3 degrees cooler...
Posts: 604
thriftee
Joined: 27 Feb 2009
#20
For me, that was an easy thing to benchmark, and it happened to be what I read was supposed to be faster.

Playing a video is very dependent on video driver, IMO, especially on my ancient machines.
Posts: 127
KrunchTime
Joined: 05 Dec 2014
#21
I played around with Antergos, an Arch respin, a few weeks ago. Of course it uses systemd and boot and shutdown were wicked fast.
Posts: 325
male
Joined: 04 Nov 2011
#22
thriftee wrote:I timed it with Manjaro tonight. 52 seconds with OpenRc and 64 seconds with Systemd. So much for the myth that systemd is faster. In fact, it's quite a bit slower, not even close.
I admit to being a fan systemd ...
What version of systemd is in Manjaro? Did it come with it or was sysvinit replaced?

Why not try it with antiX and then form your opinion?
My"Old Mr." here is from 2004

Code: Select all

 inxi -SM
System:    Host: antix14i386RD Kernel: 3.15.2-antix.1-486 i686 (32 bit) Desktop: Fluxbox 1.3.5 
           Distro: antiX-14-a3-RD_386-full Killah P 11 October 2014 
Machine:   Mobo: ASUSTeK model: P4P800SE v: Rev 2.xx
           Bios: American Megatrends v: 080009 date: 01/27/2004

Code: Select all

systemd-analyze blame | head
          3.911s slim.service
          3.326s rc-local.service
          3.293s alsa-restore.service
          3.285s console-kit-log-system-start.service
          3.278s lm-sensors.service
          3.229s pppd-dns.service
          3.228s systemd-logind.service
          3.065s ifplugd.service
          2.286s keyboard-setup.service
          2.228s wicd.service

Code: Select all

systemd-analyze critical-chain
The time after the unit is active or started is printed after the"@" character.
The time the unit takes to start is printed after the"+" character.

graphical.target @16.230s
└─multi-user.target @16.230s
  └─wicd.service @14.000s +2.228s
    └─loadcpufreq.service @12.994s +1.001s
      └─slim.service @9.045s +3.911s
        └─basic.target @8.767s
          └─timers.target @8.738s
            └─systemd-tmpfiles-clean.timer @8.737s
              └─sysinit.target @8.686s
                └─console-setup.service @8.499s +171ms
                  └─kbd.service @8.091s +365ms
                    └─networking.service @7.733s +304ms
                      └─ufw.service @7.543s +187ms
                        └─local-fs.target @7.530s
                          └─media-DATEN.mount @7.378s +149ms
                            └─dev-sda6.device @7.333s

Code: Select all

apt-cache policy systemd
systemd:
  Installiert:           215-5+b1
  Installationskandidat: 215-5+b1
  Versionstabelle:
 *** 215-5+b1 0
        500 http://ftp.de.debian.org/debian/ testing/main i386 Packages
        100 /var/lib/dpkg/status
Systemd is currently at 218
Posts: 604
thriftee
Joined: 27 Feb 2009
#23
Actually, I timed the systemd version again this morning and it was better this time, at about 50 seconds, so maybe something held it up last time.

Anyway, I still don't see any advantage that makes me want it. I don't have memory or CPU to spare, so need to either get software that's already tweaked to have a minimal footprint, or easily tweakable, because I'm looking for a desktop that needs at max 70mb to run a graphical desktop, panel, menus and windowing system with wifi running before I start opening programs, or I won't have enough memory to open them.

If I find that with systemd, I'll be more open to using it, otherwise, I need to run things that will work on the old hardware I have.
Posts: 325
male
Joined: 04 Nov 2011
#24
Why do not you just use the Commands above, so that get a better understanding?

I'll never understand why in English forums always a thousand words, a state is described, instead of showing this with a command in the terminal ... __{{emoticon}}__
Posts: 604
thriftee
Joined: 27 Feb 2009
#25
15 seconds more or less on boot or shutdown is not important to me. It surprises me that people spend time analyzing it down to thousandth's of a second, and I guess someone needs to do that or everything will get as slow as microsoft, but saving the 15 seconds will not solve the big problems for me.

What is important to me is having enough available memory that I don't go into swap. If I go into swap, I either have to power it off or spend 30 minutes trying to close enough applications for the machine to recover. Neither is a good solution, and what I really am trying to do is avoid the problem, and I am doing that by tweaking the system to be very efficient because my old machine is maxxed out already.

I am less able to tweak with systemd so far. Maybe you can suggest somewhere good to read how to reduce memory requirements under systemd? That is the big problem with it for me.
Posts: 325
male
Joined: 04 Nov 2011
#26
Why do not you show inxi -F etc. ?

I can only repeat myself:
Install antiX-14-a3-RD_386-full Killah P 11 October 2014

The memory usage has m.M. mainly to do with it and not with systemd:

========= SCRAPER REMOVED AN EMBEDDED LINK HERE ===========
url was:"https://www.bitblokes.de/2012/10/desktop-vergleich-wer-ist-der-schnellste-und-braucht-am-wenigsten-speicher/"
linktext was:"https://www.bitblokes.de/2012/10/deskto ... -speicher/"
====================================
Posts: 604
thriftee
Joined: 27 Feb 2009
#27
male wrote:Why do not you show inxi -F etc. ?

I can only repeat myself:
Install antiX-14-a3-RD_386-full Killah P 11 October 2014

The memory usage has m.M. mainly to do with it and not with systemd:

========= SCRAPER REMOVED AN EMBEDDED LINK HERE ===========
url was:"https://www.bitblokes.de/2012/10/desktop-vergleich-wer-ist-der-schnellste-und-braucht-am-wenigsten-speicher/"
linktext was:"https://www.bitblokes.de/2012/10/deskto ... -speicher/"
====================================
I can't run any of those with 512 mb of memory.

Here is a comparison of his machine to mine. His machine is 18 times faster than mine and has 32 times as much memory.

========= SCRAPER REMOVED AN EMBEDDED LINK HERE ===========
url was:"http://www.cpubenchmark.net/compare.php?cmp%5B%5D=890&cmp%5B%5D=1149"
linktext was:"http://www.cpubenchmark.net/compare.php ... cmp[]=1149"
====================================


I cannot reasonably run the standard Manjaro Openbox desktop, because in addition to being slower it eats up too much memory, so after I install a system, I need to turn as many things off as I can to save memory.

Using the Manjaro OpenRC Openbox version with things turned off it took 80 seconds to boot, login, and bring up a terminal, and free says 69 mb used.

Running the standard Manjaro Systemd Openbox version took 110 seconds to bring up a terminal, and free says 126 mb used.

The antiX 14r3.x gets a kernel panic and won't boot or run. There is a thread on that, and I think its the USB not being able to handle 2.0 causing the problem.
viewtopic.php?f=6&t=5393

The antiX 13.2 runs fine if I install the proprietary nVidia 96.xx drivers after changing to 3.2.0-4 kernel.

The MX-14.x runs fine but is slow.

I was going to update this post with a benchmark and inxi output from antiX 13.2 but it looks like I need to reinstall it to get it to run.

A tuned Crunchbang 11 took 70 seconds and used 73 mb.

antiX stock 13.2 version took 87 seconds and used 61 mb. That was in 1024 x 768 because its stock.

Code: Select all

$ inxi -Fxz
System:    Host: d8100 Kernel: 3.7.10-antix.5-486-smp i686 (32 bit, gcc: 4.8.1) 
           Desktop: IceWM 1.3.7 Distro: antiX-13.2_386-full Luddite 4 November 2013
Machine:   Mobo: Dell model: Inspiron 8100 Bios: Dell version: A08 date: 12/21/2001
CPU:       Single core Intel Pentium III Mobile CPU (-UP-) cache: 512 KB flags: (pae sse) bmips: 1993.27 clocked at 1000.00 MHz 
Graphics:  Card: NVIDIA NV11 [GeForce2 Go] bus-ID: 01:00.0 
           X.Org: 1.12.4 drivers: vesa,nouveau (unloaded: fbdev) Resolution: 1024x768@61.0hz 
           GLX Renderer: Gallium 0.4 on softpipe GLX Version: 2.1 Mesa 8.0.5 Direct Rendering: Yes
Audio:     Card: ESS ES1983S Maestro-3i PCI Audio Accelerator port: dc00 bus-ID: 02:03.0 
           Sound: Advanced Linux Sound Architecture ver: k3.7.10-antix.5-486-smp
Network:   Card-1: Intersil ISL3886 [Prism Javelin/Prism Xbow] driver: prism54 bus-ID: 0d:00.0
           IF: eth1 state: down mac: <filter>
           Card-2: Intel 82557/8/9/0/1 Ethernet Pro 100 driver: e100 ver: 3.5.24-k2-NAPI port: ecc0 bus-ID: 08:04.0
           IF: eth0 state: up speed: 100 Mbps duplex: full mac: <filter>
Drives:    HDD Total Size: 46.0GB (5.6% used) 1: id: /dev/sda model: HITACHI_DK23CA size: 30.0GB 
           2: USB id: /dev/sdb model: JD_FireFly size: 16.0GB 
Partition: ID: / size: 4.9G used: 2.5G (53%) fs: ext4 ID: swap-1 size: 0.51GB used: 0.00GB (0%) fs: swap 
           ID: swap-2 size: 0.70GB used: 0.00GB (0%) fs: swap 
Sensors:   System Temperatures: cpu: 56.5C mobo: N/A 
           Fan Speeds (in rpm): cpu: N/A 
Info:      Processes: 76 Uptime: 5 min Memory: 62.6/500.7MB Runlevel: 5 Gcc sys: 4.7.2 
           Client: Shell (bash 4.2.37) inxi: 1.9.16
Last edited by thriftee on 13 Dec 2014, 20:33, edited 7 times in total.
Posts: 4,164
rokytnji
Joined: 20 Feb 2009
#28

Code: Select all

free says 259 mb used.
That is a lot unless you mean with a browser, terminal, and conky running.
My XFCE installs pull that much. With a browser open also.
Posts: 4,164
rokytnji
Joined: 20 Feb 2009
#29
Yeah, even my custom LXDE AntiX install only pulled about 100MB of ram.


========= SCRAPER REMOVED AN EMBEDDED LINK HERE ===========
url was:"http://forum.lxde.org/viewtopic.php?f=11&t=31202"
linktext was:"http://forum.lxde.org/viewtopic.php?f=11&t=31202"
====================================
Posts: 604
thriftee
Joined: 27 Feb 2009
#30
rokytnji wrote:

Code: Select all

free says 259 mb used.
That is a lot unless you mean with a browser, terminal, and conky running.
My XFCE installs pull that much. With a browser open also.
The 259 is with just the terminal, and after letting it settle. It was 262 the first time, then a little later dropped to 259. Only conky was running and its a very basic stock conky that comes with the install. Nothing was changed after the install. It needs a lot of memory and is slow probably due to that, and anyway, with the OS taking so much, there isn't much left for browsers etc.

My tweaked down OpenRC version is quite tolerable and I can open 20 browser tabs in seamonkey, but eventually if it runs low on memory it will get very slow as it starts swapping, but that is to be expected.

antiX 13.2 is one of the more efficient OS's and needs about 70 mb, similar to Manjaro OpenRC Openbox. I am running a reinstall of it now. Luckily its a fast install compared to many that take hours. I have a good install of it for that laptop but its on a different disk drive.

My point in this thread is that to be able to run an OS on a limited machine, I need to be able to tune down the memory and cpu usage, and that with systemd I can't find ways to do that as well as I can with the non-systemd ones.

PS: I goofed on the memory number for Manjaro Openbox systemd. The correct number is 126 not 269. I went back and redid the test, and it took 115 seconds and got a 126 mb number. I should have noticed the number was unreasonable. i read the number above it from the screen by mistake.