| |||||||||||||
Don't cheat on boot time.Don't cheat on boot time.Posted Nov 26, 2009 10:46 UTC (Thu) by dwmw2 (subscriber, #2063)In reply to: Don't cheat on boot time. by Tuxie Parent article: The Grumpy Editor's Moblin review If I'm boasting about how quickly my PHP application can be up and running again after a power cut, of course I'll include the boot time of the server I choose to run it on, the SQL server it depends on, etc. To exclude those and publish just an irrelevant microbenchmark of the part I want you to think about would be massively disingenuous. If boot time matters a lot, that will definitely affect the choice of operating system I run it on. It could also affect my choice of chipset/CPU — if I really care about how quickly the whole thing starts up, I'm more likely to use hardware from a company which supports open source projects like coreboot, rather than one which refuses to support open source and leaves me stuck with a slow, closed-source BIOS implementation that I can't fix for myself.
(Log in to post comments)
Don't cheat on boot time. Posted Nov 26, 2009 10:57 UTC (Thu) by Tuxie (guest, #47191) [Link]
But this review was a review of Moblin, not the laptop it was running on.
If he was reviewing a netbook with Moblin preinstalled, of course the BIOS time should be counted, but that was not the case here.
Don't cheat on boot time. Posted Nov 26, 2009 20:00 UTC (Thu) by amikins (guest, #451) [Link]
To be fair, it WAS a netbook with Moblin preinstalled. :)
Don't cheat on boot time. Posted Nov 26, 2009 21:32 UTC (Thu) by Tuxie (guest, #47191) [Link]
Ouch, you're right. If someone asks for me, I'm over there in the corner hiding for a while.
Don't cheat on boot time. Posted Nov 26, 2009 23:29 UTC (Thu) by giraffedata (subscriber, #1954) [Link] I don't think it was a review of the whole package, even though the first tests were done on a the preassembled package. It was a review of Moblin.And it doesn't matter. Readers aren't trying to judge the engineering effort; they're judging the result. If Moblin can't make boot fast because the time is spent in things Moblin can't control, then Moblin fails just as surely as if Moblin engineers just did stupid things in the code. Another way of putting it is that to the extent that things outside Moblin elongate the boot, the time it takes Moblin per se to start is irrelevant and the article might as well not mention it.
Don't cheat on boot time. Posted Dec 3, 2009 12:16 UTC (Thu) by robbe (guest, #16131) [Link]
I don't think the BIOS-to-login time is a microbenchmark. Relationships
between micro- and "real" benchmarks is normally much more complicated than in this case where the time you want to optimise is simply a sum of the BIOS and the OS part.
I'd expect the BIOS time to be below 10 seconds in this case, so trimming
FWIW, which hardware vendors support coreboot well?
Datapoint: one of my *server* VMs takes 26 seconds from poweron to login.
P.S.: Am I the only one who thought that our editor was surfing a site
|
|||||||||||||