Weekly Edition
Return to the Press page
| |||||||||||||
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek)
InformationWeek
covers an OSCON talk by
Dirk Hohndel.
"Dirk Hohndel, chief open source and Linux technologist for Intel, addressed OSCON 2009, an annual convention of open source developers, and said it was a re-architecting of the Linux start up process that gave the netbook one of its most desirable characteristics.
"We know, after all, that we're an instant gratification society," he told about 1,000 attendees gathered at the San Jose Convention Center Wednesday. "If it's more than 15 seconds No one wants to wait until they can do something."
(Log in to post comments)
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 0:18 UTC (Fri) by fuhchee (subscriber, #40059) [Link]
I don't understand the newsworthiness of this, after reading the InformationWeek article. Does it refer to work other than Arjan et al.'s fast boot work from late 2008? As for the "designing netbooks for fast starts" headline, is there anything to that as opposed to "designing linux software for netbooks" - a somewhat different thing?
Arjan's fast boot demo at LPC Posted Jul 24, 2009 15:15 UTC (Fri) by dmarti (subscriber, #11625) [Link] LWN article and long comment thread about the fast boot session at last year's Linux Plumbers Conference: LPC: Booting Linux in five seconds
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 3:07 UTC (Fri) by mjg59 (subscriber, #23239) [Link]
I'm slightly disappointed that the Intel portability story seems to have changed from power management to boot speed. There's still a great deal of low hanging fruit in the power management field, and I'd have thought that users would be more interested in an extra 25 minutes of battery life and working suspend/resume than it taking 3 seconds longer to boot - especially when that benefits the rest of the Linux market as well.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 5:16 UTC (Fri) by arjan (subscriber, #36785) [Link]
we're still working on power management... mostly around usb/bluetooth nowadays. But to be honest, there is not that much on the table anymore (assuming you have a good bios, and have USB devices that autosuspend properly). If you have ideas on where there is more on the table do talk to us... we can help throw resources at things...
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 13:14 UTC (Fri) by mjg59 (subscriber, #23239) [Link]
The runtime PM story still isn't very compelling. We've got infrastructure for dealing with USB, but not so much on other hardware - the absence of any support in the SCSI layer means we're not able to go anywhere with powering down host controllers and also limits the ability to do any kind of decent autosuspend of USB mass storage. Working with Jesse this week, we've managed to shave something like a Watt off graphics in return for 150 lines or so of code. We're seeing vendors starting to engage in aggressive memory underclocking through SMM code and we don't have any kind of answer to that yet. Hibernation's still an utter disaster, though admittedly that's a rather bigger problem.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 16:35 UTC (Fri) by jsbarnes (guest, #4096) [Link]
I've been meaning to look at Rafael's new runtime PM support code; my quick glance yesterday wasn't nearly enough (though what I saw looked good).
The main issue I see with power management is that it seems to be mostly driver and subsystem level work (though mostly the former I'm afraid). And it's often not totally trivial; saving & restoring device state for suspend/resume has already proven difficult. Doing it at runtime should give us more test coverage there, but doesn't make the problem easier. Also, doing things like the downclocking we're working on in graphics is great, but it does add more complexity (timer driven state machines) to each driver or subsystem...
I hope that as more vendors start shipping Linux, we'll see people focus on a whole set of drivers at a time (the set of drivers for a given platform). But I expect the experience for users running on platforms where the OEM hasn't made such an effort will always lag.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 18:40 UTC (Fri) by jwb (guest, #15467) [Link]
Powering down unused AHCI controller devices still doesn't work. There was a patch years ago from Intel but it never worked right, was never merged, and nobody seems to have followed up on it. On laptops that have a docking port or swappable SATA optical drive, completely shutting down the unused ports results in a measurable power saving.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 22:03 UTC (Fri) by arjan (subscriber, #36785) [Link]
the patch is in the kernel; it's a sysfs tunable though.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 25, 2009 0:25 UTC (Sat) by jwb (guest, #15467) [Link]
Hrmm I must have overlooked it, because I thought that "power_off" was going to become an acceptable argument to link_power_management_policy, and in my copy of libata-scsi.c only min_power, medium_power, and max_performance are defined. The only place in ahci.c I see touching SCTL.DET gets called on system suspend.
Anyway, I should probably take it to the libata mailing list instead of here :)
I was going to mention PCIe ASPM but I saw that it's been enabled in Rawhide, which is great news.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 25, 2009 5:19 UTC (Sat) by mjg59 (subscriber, #23239) [Link]
There's two different parts of this. The first is downclocking the SATA link when it's idle - this is speced as ASPM and is implemented, but is primarily handled by the hardware itself. The second is recognising that the link has been idle for a sufficient length of time and powering down the entire AHCI controller, which is more complex and involves the OS paying attention to what's going on. We don't have support for that yet.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 5:47 UTC (Fri) by trasz (guest, #45786) [Link]
For people who need to use notebook without waiting until it boots, the boot time doesn't matter at all - they just put it to sleep instead of powering it off.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 7:28 UTC (Fri) by dgm (subscriber, #49227) [Link]
That's because they still take 30-40 seconds to boot.
5 seconds, on the other hand, is faster that waking up from sleep (in my Aspire One, at least).
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 7:57 UTC (Fri) by trasz (guest, #45786) [Link]
Except that after boot, one still needs to log in and launch required applications. Also, 5 seconds is about 3 seconds longer than resuming from sleep on a macbook.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 13:12 UTC (Fri) by dgm (subscriber, #49227) [Link]
I don't usually have many running applications in the netbook, usually only the browser. And I don't have a macbook, neither plan to buy one, thanks.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 8:09 UTC (Fri) by sf_alpha (guest, #40328) [Link]
Sleep is for short time and however drain battery power. Hibernation/Poweroff is only option to keep battery power. Current Hibernation implements still involves kernel booting process, boot time still need improvements.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 13:51 UTC (Fri) by zlynx (subscriber, #2285) [Link]
My Macbook Pro can sleep for several days, almost a week, before running out of power.
I think Apple has set the standard to meet in power management.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 19:43 UTC (Fri) by dufkaf (subscriber, #10358) [Link] My Nokia internet tablet lasts weeks sitting completely idle (not suspended or hibernated, just idle with applications started). It can also last many days idle with wi-fi enabled (i.e. connected to AP, it is possible to ssh into the tablet). Fast boot is nice but good power management is much better. How useful is fast boot or quick suspend/resume if you e.g. cannot receive skype call when device is not on?
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 19:59 UTC (Fri) by zlynx (subscriber, #2285) [Link]
That's really good. That's what a good ARM design can give us for netbooks too.
The Nokia probably turns off the screen and then the processor and most of the chipset essentially goes into suspend until some device interrupt wakes things up.
Although I doubt its performance is even up to Atom netbook standards because really low-power RAM that can do that isn't very fast.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 24, 2009 21:05 UTC (Fri) by jwb (guest, #15467) [Link]
N800 performance isn't anywhere near an Atom, to be certain, but it's good enough for Skype calls and SD video playback.
OSCON 2009: How Intel Designed Netbooks For Fast Starts (InformationWeek) Posted Jul 28, 2009 9:09 UTC (Tue) by pharm (guest, #22305) [Link]
The N8x0 devices are 200MHz in-order ARM cores: Not fast. The next Maemo devices will be built
around the Kirkwood cores, which are an OoO ARM implementation which can be clocked at up to 1.2 GHz. That's enough for most purposes I should think (although whether the Maemo devices will be clocked that high I don't know.)
(Idle power on the Sheevaplug which uses a 1.2GHz Kirkwood ARM core is just over 2W & I bet a lot
|
|||||||||||||