Hibernate to console or lightweight WM -- a better/faster way to boot?

paraflu · 2008-02-26 00:53:34

Hello everbody

I`ve got an idea which i would like to discuss with you.
It could be a stupid idea, so don`t rip me apart.
When thought about accellerating the boot process it comes to my mind that the system always search and examine the hardware and load the needed modules. But is this really needed? There is the autodetect hook which detects only the
needed hardware, but isn`t there a faster way to do this, if i stay on an nochanged system? What has come into my mind is to use hibernate, but i don`t know if this is possibile. When i boot the system to the login prompt or when
i have been logged in into the console my system only use 8Mb ram. When i hibernate at this moment and the ram gets compressed into the hibernate file, it would only takes seconds to boot the system into the previous state.
I do not use hibernate but if anybody has experience with hibernate and can explain to me if this stupid or impossible, i would like to hear about it. There also the possibility that the hardware detection isn`t the real problem and it is more the populating of udev. I don`t know. Also if i use fluxbox or similiar lightweight wm the memory consumption do not exceed 30Mb which would also fast to hibernate. Maybe somebody use these WMs and hibernate
and has some numbers.

Mods: I didn`t know where to post this. In the linux discussion or the kernel/hardware subforum. So i dicided to post it in the linux discussion.

Thanks for your answers

fwojciec · 2008-02-26 01:02:53

I don't fully understand your question, but both hibernate and suspend are possible on both laptops and desktops and that's completely regardless of the WM you're using (you can do it from the console as well). I use suspend on both my laptop and my desktop (both with Openbox), for example.

dyscoria · 2008-02-26 01:07:33

I think compiling your own kernel would accelerate the boot process in terms of autodetecting stuff, though this is not for the faint hearted. On an unchanging system as you mentioned, once you learn to compile once, then it's simply a matter of doing the same procedure every time a new kernel comes out.

Endperform · 2008-02-26 01:44:12

dyscoria wrote:

I think compiling your own kernel would accelerate the boot process in terms of autodetecting stuff, though this is not for the faint hearted. On an unchanging system as you mentioned, once you learn to compile once, then it's simply a matter of doing the same procedure every time a new kernel comes out.

Yeah, I think you're right here. Back in the Gentoo days, I'd compile everything I needed into a kernel and didn't worry about auto detection. Eventually I got tired of re-tweaking when a major kernel revision came out and just went with the genkernel. I don't know how much speed would be gained by doing this, though.

On the topic of hibernation, I don't have much experience there, but it might be worthwhile to try, at least, although I can't see why waiting about 20-30 seconds is a big deal.

paraflu · 2008-02-26 01:51:15

Thank you for your answers.

What i really wanted to know was that if you use hibernate to get a ready-system with a lightweight wm or console , do you experience instant booting times after grub (bios and grub use time too, but i mean after that) in the range 1 or 2 seconds. Like a C64 or gameconsole. Because i think when you use a full desktop with for
example 200MB used ram and divide the data transfer rate of your harddisk like 20Mb/s i think this would take 10 seconds or something like that. I would like to know how fast this could get especially if you take an lightweight wm or console with lesser than a full desktop. How many seconds do you measure does it get faster
with lesser memory use. Can it reach even lesser than 1 second? But i see i have to read myself more thru the docs of hibernate [suspend to disk] to know more about it. Some suspend to ready-system times would be nice if you like to share them with me. To get a feeling what is possible.

Thanks

fwojciec · 2008-02-26 03:15:36

Even suspend (no hard drive reading/writing involved) takes about 5-10 seconds to resume (on my desktop). Hibernate takes longer -- I've never timed it but I'd say at least 10-15 seconds (probably more) from grub.

Last edited by fwojciec (2008-02-26 03:29:58)

Arch Linux

#1 2008-02-26 00:53:34

Hibernate to console or lightweight WM -- a better/faster way to boot?

#2 2008-02-26 01:02:53

Re: Hibernate to console or lightweight WM -- a better/faster way to boot?

#3 2008-02-26 01:07:33

Re: Hibernate to console or lightweight WM -- a better/faster way to boot?

#4 2008-02-26 01:44:12

Re: Hibernate to console or lightweight WM -- a better/faster way to boot?

#5 2008-02-26 01:51:15

Re: Hibernate to console or lightweight WM -- a better/faster way to boot?

#6 2008-02-26 03:15:36

Re: Hibernate to console or lightweight WM -- a better/faster way to boot?

Board footer