HowTo Speed Up Boot & Increase Responsiveness (Desktop)

Message

julian67 · #16 Post by **julian67** » 2008-09-29 02:00

I think it's a large bookmarks file which kills start up times on browsers. My Epiphany and Iceweasel browsers have a similar bookmark list, somewhere between 1500 and 2000 entries. I no longer have any idea how Iceweasel handles bookmarks but I know it has to either load them or load a database containing them. I also noticed when I tried Icecat I was impressed with how fast it loaded. The I imported my bookmarks into it and it was the same as Iceweasel....

Probably the extensions don't help but I don't have many.

Epiphany is the browser I use all the time, it definitely benefits from preload and starts near enough instantly but it was quite slow to load when i switched it off. I rarely use Iceweasel and I guess it doesn't get the benefit of preload. From a cold start it's 15 seconds minimum. I've read reviews where people found it took 30 seconds + to start.

I'm not complaining too much because the only fast to start browsers when you have 1000s of bookmarks are the text based ones. I use elinks with the same bookmark set and I guess it doesn't load bookmarks into memory, it waits for you to press 's' and then it parses the text file containing them.

Anyway I have enough RAM to allow preload to allocate a chunk of it at boot to all the stuff I always use and it works very nicely

At the moment I'm only using laptops and while this one has a Core Duo and plenty of RAM it's still bottlenecked for disk intensive operations by a typically slow laptop HDD. It's great for CPU intensive or RAM hungry tasks but when it comes to read/write performance my old self built desktop (in storage atm) with Athlon CPU and cheapest motherboard I could find 5 years ago is much superior. I have a lot of external USB storage and I always notice how much faster a 3.5" ATA drive in an enclose is than the 2.5" SATA drive in the laptop.

didi · #17 Post by **didi** » 2008-09-29 16:07

julian67 wrote:There's no negative consequence unless you have applications that depend on access times to work. This is unlikely on a server and on your desktop is not even worth worrying about.

I've read some stuff and indeed it has no use most of the time. So I thought, what the hell ... let's try it and if I feel uncomfortable about it for some reason, it's easily reversed.
And I have to say that the boot process seems to have speed up considerably

http://tinyurl.com/wurfw from Lou's guide has quite some interesting info about atime too.

MeanDean wrote:Actually, there was a discussion not long ago about it and I think the agreement was that it is truly a relic. I think it was concluded that there was one (or maybe two) software packages that expected atime to be used. They were obscure packages if I remember correctly. I have used noatime for about a year now - no problems here.

Good to have conformation from another experienced user

And it made me do the obvious thing (which I hadn't), search this forum about it and already found some interesting stuff. Thanks!

Among those was this post from eyelid about mlocate (which didn't get answered):

instead of re-reading all the contents of all directories each time the database is updated, mlocate keeps timestamp information in
its database and can know if the contents of a directory changed without reading them again. This makes updates much faster and less
demanding on the hard drive. This feature is only found in mlocate.

Since mlocate is the replacement of locate (on my system) I like to make sure noatime doesn't affect mlocate's behavior. It seems to look for write times. Will that still be updated regardless of noatime and hence keep mlocate working properly?

On to reading more about this stuff.
LOL, I feel like a kid in a candy shop

julian67 · #18 Post by **julian67** » 2008-09-29 16:21

I use mlocate and use the locate command a lot. No problems at all.

If you look at man mlocate.db you can read

[quote]Directory time is the maximum of st_ctime and st_mtime of the directory. updatedb(8) uses the original data if the
directory time in the database and in the file system match exactly. Directory time equal to 0 always causes resâ€

vmclark · #19 Post by **vmclark** » 2008-09-29 20:41

Thanks julian67 for this tutorial. I'll give it a go. I think outside swappiness, everything else is default on my system.

anticapitalista wrote:IMO, what really does speed up boot time is the services loaded or not at boot.
This is why Arch seems so fast at first as there are hardly any services running
and slows down when you add more apps that need services started.
So, if you know which services to disable, you gain a lot of boot speed.

I found this out also. Installed Arch , was amazed at the startup speed...until I added needed
services like Gnome. It ended up that Ubuntu was faster than Arch!

craigevil · #20 Post by **craigevil** » 2008-09-29 22:54

Yes it always funny watching people in irc boast how fast their distro boots. Until you take a look at it and very few needed services like wifi, network, etc are started at boot and the fact that they boot into a terminal. Try comparing apples to apples and not apples to pumpkins.

Even on my Eee PC 701 with 1 gb of ram running Lenny I get to LXDE in less than 30 seconds from cold boot.

Some other tweaking ideas:
http://sidux.com/PNphpBB2-viewtopic-t-3 ... aking.html
http://forums.debian.net/viewtopic.php?t=14129

MeanDean · #21 Post by **MeanDean** » 2008-09-30 17:45

anyone use elevator=deadline and seen a difference?

jongi · #22 Post by **jongi** » 2008-10-03 23:37

I've got this long wait during my boot up stage

Code: Select all

Oct  1 22:52:47 localhost kernel: [    9.124564] udev: renamed network interface wmaster0 to eth1
Oct  1 22:52:47 localhost kernel: [    9.997042] hda_intel: azx_get_response timeout, switching to polling mode: last cmd=0x011f0900
Oct  1 22:52:47 localhost kernel: [   40.258927] Adding 793760k swap on /dev/sda6.  Priority:-1 extents:1 across:793760k
Oct  1 22:52:47 localhost kernel: [   40.835790] EXT3 FS on sda1, internal journal

Is this saying that:
1. the hda_intel process started 9.99s and failed at 40.25s, or
2. hda_intel process failed at 9.99s and adding the swap completed at 40.25s?

julian67 · #23 Post by **julian67** » 2008-10-04 00:00

jongi wrote:I've got this long wait dur.....

Your support request is about reading logs. This is Docs, Howtos, Tips & Tricks. You should post a support request in an appropriate place.

MeanDean wrote:anyone use elevator=deadline and seen a difference?

I've been trying it and don't like it at all. Everything seems fine until I transfer a big file/archive to an external USB drive. It seems to take priority over everything else and it can be very boring to wait for several GB to transfer onto a drive with slow write speeds before being able to do anything else. I 'm going back to the default cfq....it looks really impressive now

MeanDean · #24 Post by **MeanDean** » 2008-10-04 00:23

julian67 wrote: I 'm going back to the default cfq....it looks really impressive now

I havent noticed any difference....yet...

julian67 · #25 Post by **julian67** » 2008-10-04 05:08

just went back to default scheduler.....all is good again. Using deadline I had horrible freezes in games, whole desktop freezes on large file transfers via usb, audio stuttering, desktop freezes during updatedb cronjob...very unsatisfactory. It was pretty snappy if you have nothing going on in the background but who runs a PC with nothing going on in the background???? It's what automation is for.

So I've rebooted with default (cfq). I have rtorrent in a disconnected screen session, external media being used, mail client, feed reader, web browser, midnight commander and moc running in hidden tilda sessions, movie player accessing the external media. Everything is very responsive including 3d games. cfq is fine with me

Ook · #26 Post by **Ook** » 2008-10-04 19:20

Is there any gain in mounting the CD in fstab with noatime?

julian67 · #27 Post by **julian67** » 2008-10-04 19:36

Ook wrote:Is there any gain in mounting the CD in fstab with noatime?

No. Don't change anything in fstab related to your optical drives. nd don't change anything at all unless you have some idea what the change means.

The atime, noatime and nodiratime flags are about recording access/write times. Those times are written to the file/directory (or maybe to the journal, I'm not sure but it doesn't matter for the purpose of this question) that is accessed. A CD is read-only, there is no way anything can be written back to it. The atime, noatime, nodiratime flags have no relevance. If you use DVD-RAM or use CDRWs with packet writing then maybe there is some relevance but try not to worry about it

You're not going to improve aspects of system performance which depend on the hard drive and RAM by tinkering with optical drive parameters.

Bulkley · #28 Post by **Bulkley** » 2008-10-04 20:14

julian67 wrote:The atime, noatime and nodiratime flags are about recording access/write times. Those times are written to the file/directory (or maybe to the journal, . . .

Will noatime use be a problem when the power fails?

julian67 · #29 Post by **julian67** » 2008-10-04 20:26

No more than anything else......

atime, noatime and relatime are about access/write timestamps. That's timestamps. That's not your valuable data.

The parameter that deals with writing cached data is the VM writeback time. You can easily google to learn about this and make up your own mind if the system defaults are appropriate for you. If you have a laptop, or a desktop or server with UPS, then a longer writeback interval might be much better. If you live somewhere with unreliable power and you don't use a UPS or a laptop you really don't want to increase this.

If you suffer power failures you need to get a UPS.

Ook · #30 Post by **Ook** » 2008-10-04 23:59

julian67 wrote:
Ook wrote:Is there any gain in mounting the CD in fstab with noatime?
. . . Those times are written to the file/directory (or maybe to the journal, I'm not sure but it doesn't matter for the purpose of this question) that is accessed. . . .

That is what I was looking for. All is implemented and working well. Thank you.

gerry · #31 Post by **gerry** » 2008-11-03 12:50

Historical note about need for atime: when Unix started, computers were still using magnetic core storage. The "read" operation was destructive, so to retain the data it had to be immediately re-written. Don't think we need that these days.....

Gerry

birdywa · #32 Post by **birdywa** » 2008-11-18 01:45

Should I use noatime on my swap partition?

julian67 · #33 Post by **julian67** » 2008-11-18 02:01

birdywa wrote:Should I use noatime on my swap partition?

you can try

no.

You can find references to people doing it but I'm not sure why they bother. Nothing is stored in swap in normal use so there are no directory or file access times to be written there. What advantage anyone expects by using noatime on their swap partition is a mystery to me.

Swynndla · #34 Post by **Swynndla** » 2008-11-20 03:54

My boot time took 1m31s.
Then I installed insserv (and preload and changed vm.swappiness & vm.vfs_cache_pressure) following the instructions given in this nice tutorial.
Boot time now is 1m29s.
So there's wasn't a lot in it for me.

julian67 · #35 Post by **julian67** » 2008-11-20 05:20

Swynndla wrote:My boot time took 1m31s.
Then I installed insserv (and preload and changed vm.swappiness & vm.vfs_cache_pressure) following the instructions given in this nice tutorial.
Boot time now is 1m29s.
So there's wasn't a lot in it for me.

Of those only the insserv change is aimed at improving boot time. As stated earlier using preload will probably lengthen boot time. It's worth actually reading the howto and the references at the end of it before making changes to your system.