Neat concept but the locking restrictions seem a GPL violation. In spirit at least?
http://gadgets.boingboing.net/2007/10/08/splashtop-instanton.html Being an expansion of embedded Linux integrated at the motherboard leve. Sadly however it seems to have some embedded crippling locks. WtF is the obsession with making "suck free" anything an impossibility? "The conservation of stupidity demands that everything have something stupid in or about it" -- Oren Beck 816.729.3645
boot acceleration is a neglected science; no reason why a box with a 3ghz cpu takes any time at all to start up. freeze-to-swap instead of shutdown is a candidate; so what if you have to reestablish your network connections when you restart. On Feb 12, 2008 6:57 PM, Oren Beck <orenbeck@gmail.com> wrote:
http://gadgets.boingboing.net/2007/10/08/splashtop-instanton.html
-- "The ultraleft underground armed revolutionary groups that burned down our offices and forced our staff to leave the villages at gunpoint now disappeared. We could finally concentrate on the production of fish." -- M. Yunus
On Tuesday 12 February 2008, David Nicol wrote:
boot acceleration is a neglected science; no reason why a box with a 3ghz cpu takes any time at all to start up.
Mostly I/O bound, isn't it? Hard drives haven't got much faster for years.
On Feb 12, 2008 7:14 PM, Luke -Jr <luke@dashjr.org> wrote:
On Tuesday 12 February 2008, David Nicol wrote:
boot acceleration is a neglected science; no reason why a box with a 3ghz cpu takes any time at all to start up.
Mostly I/O bound, isn't it? Hard drives haven't got much faster for years.
okay, but why doesn't, for instance gnome manager swap everything out instead of having to re-initialize itself when I reboot my etch? Because it would be more trouble than its worth, that's why.
On Tuesday 12 February 2008, David Nicol wrote:
On Feb 12, 2008 7:14 PM, Luke -Jr <luke@dashjr.org> wrote:
On Tuesday 12 February 2008, David Nicol wrote:
boot acceleration is a neglected science; no reason why a box with a
3ghz
cpu takes any time at all to start up.
Mostly I/O bound, isn't it? Hard drives haven't got much faster for years.
okay, but why doesn't, for instance gnome manager swap everything out instead of having to re-initialize itself when I reboot my etch? Because it would be more trouble than its worth, that's why.
Because you didn't tell it to. Linux has support for this.
On Tuesday 12 February 2008 19:46:26 David Nicol wrote:
okay, but why doesn't, for instance gnome manager swap everything out instead of having to re-initialize itself when I reboot my etch? Because it would be more trouble than its worth, that's why.
Frankly, Gnome has always seemed to me the antithesis of swiftness and efficiency.
On Tue, Feb 12, 2008 at 8:57 PM, Jonathan Hutchins <hutchins@tarcanfel.org> wrote:
On Tuesday 12 February 2008 19:46:26 David Nicol wrote:
okay, but why doesn't, for instance gnome manager swap everything out instead of having to re-initialize itself when I reboot my etch? Because it would be more trouble than its worth, that's why.
Frankly, Gnome has always seemed to me the antithesis of swiftness and efficiency.
Frankly, suspend to disk would be super slow with a 2GB of RAM. Most of that is likely disk cache anyways, which probably isn't so useful anymore anyways. Bootcharts often present a good idea of where your time is being spent during boot, but there are substantial reasons why booting should take time. As mentioned, disk I/O hasn't really taken off the way memory bandwidth has. Additionally, DHCP is a very slow system that could take _seconds_ to work. Mounting filesystems usually involve some time; reiserFS for example takes like 30 seconds, of mostly disk I/O. The video card usually has some timing delays to safely mode change, in case your CRT is poorly made. The way around these problems are exactly what these guys are doing -- an extremely pared down system with known attributes. If you're a quality developer and feel that GNOME is slow, perhaps profiling GNOME startup could reveal some easy performance gains. Dave Jones did a similar experiment with X.org via strace. Justin Dugger
participants (5)
-
David Nicol -
Jonathan Hutchins -
Justin Dugger -
Luke -Jr -
Oren Beck