headers
Chris Ball | 1 Dec 2009 01:07
Favicon
Gravatar

Re: New F11 for XO-1.5 build 48

Hi,

   > The advantages in terms of energy use must surely be tiny given
   > that we're talking about 5sec of reducing consumption by maybe
   > 0.5W.  On the other hand turning off the backlight completely
   > after a certain amount of time as a first measure to reduce power
   > consumption strikes me as an interesting alternative to the
   > current implementation...

We'll do both -- dim the backlight and suspend, and set an alarm for a
wakeup to turn off the screen altogether a few minutes later.  The
reason for leaving the backlight dimmed for a few minutes is that it
might be possible for you to continue to read the web or book page
that you're looking at while it's dimmed and we're saving the big
watts.

Dimming the screen when idle is extremely common for laptops; OS X
and gnome-power-manager do it too, so I don't think it's too much of
a distraction.  With the new versions of OHM, I do the dimming as a
fade rather than an atomic change, so that should help it be less
distracting too.  (If it still feels distracting, I could make the
fade take a little longer?  It takes around 150ms right now.)

Thanks,

- Chris.
--

-- 
Chris Ball   <cjb <at> laptop.org>
One Laptop Per Child
(Continue reading)

Permalink | Reply | 3 comments |
headers
Paul Fox | 1 Dec 2009 02:00
Favicon

Re: New F11 for XO-1.5 build 48

chris wrote:
 > Hi,
 > 
 >    > The advantages in terms of energy use must surely be tiny given
 >    > that we're talking about 5sec of reducing consumption by maybe
 >    > 0.5W.  On the other hand turning off the backlight completely
 >    > after a certain amount of time as a first measure to reduce power
 >    > consumption strikes me as an interesting alternative to the
 >    > current implementation...
 > 
 > We'll do both -- dim the backlight and suspend, and set an alarm for a
 > wakeup to turn off the screen altogether a few minutes later.  The
 > reason for leaving the backlight dimmed for a few minutes is that it
 > might be possible for you to continue to read the web or book page
 > that you're looking at while it's dimmed and we're saving the big
 > watts.

clearly we'll know more as we all use it more.  i currently find
the 20 second timeout much too short -- it's far too likely that
i'll still be looking at the screen.

 > 
 > Dimming the screen when idle is extremely common for laptops; OS X
 > and gnome-power-manager do it too, so I don't think it's too much of
 > a distraction.  With the new versions of OHM, I do the dimming as a
 > fade rather than an atomic change, so that should help it be less
 > distracting too.  (If it still feels distracting, I could make the
 > fade take a little longer?  It takes around 150ms right now.)

i also find the fade-up to be more distracting than the
(Continue reading)

Permalink | Reply | 2 comments |
headers
Philipp Kocher | 1 Dec 2009 10:14
Picon

Re: About 8.2.2

>   - It won't be signed by OLPC. You have to be on an unlocked XO, or be
> a deployment signing your own builds.

Is there a reason why 8.2.2 doesn't get signed by OLPC?
I do understand that the main target group are big deployments which can 
sign the build, but why are others excluded?

In the past even release candidates like build 800 got signed by OLPC.

Cheers Philipp
Permalink | Reply | 6 comments |
headers
John Watlington | 1 Dec 2009 10:36
Favicon

Re: New F11 for XO-1.5 build 48


I would prefer to see the backlight dimming take place on a longer
time scale (as Paul says, I'm frequently still reading when it happens)
and the suspend happening much more aggressively.

Does OHM have user configurable knobs ?

Cheers,
wad

On Nov 30, 2009, at 8:00 PM, Paul Fox wrote:

> chris wrote:
>> Hi,
>>
>>> The advantages in terms of energy use must surely be tiny given
>>> that we're talking about 5sec of reducing consumption by maybe
>>> 0.5W.  On the other hand turning off the backlight completely
>>> after a certain amount of time as a first measure to reduce power
>>> consumption strikes me as an interesting alternative to the
>>> current implementation...
>>
>> We'll do both -- dim the backlight and suspend, and set an alarm  
>> for a
>> wakeup to turn off the screen altogether a few minutes later.  The
>> reason for leaving the backlight dimmed for a few minutes is that it
>> might be possible for you to continue to read the web or book page
>> that you're looking at while it's dimmed and we're saving the big
>> watts.
>
(Continue reading)

Permalink | Reply | 1 comment |
headers
Martin Langhoff | 1 Dec 2009 11:17
Picon
Gravatar

Re: About 8.2.2

On Tue, Dec 1, 2009 at 10:14 AM, Philipp Kocher <philipp.kocher <at> gmx.net> wrote:
>>  - It won't be signed by OLPC. You have to be on an unlocked XO, or be
>> a deployment signing your own builds.
>
> Is there a reason why 8.2.2 doesn't get signed by OLPC?
> I do understand that the main target group are big deployments which can
> sign the build, but why are others excluded?

Nobody is excluded :-) OLPC wants to encourage deployments large and
small to have their own keys and sign the OSs they decide they want to
use.

I am working to make this easier for deployments with small tech teams.

Everyone else should disable the antitheft stuff (except when helping
us test it!).

m
--

-- 
 martin.langhoff <at> gmail.com
 martin <at> laptop.org -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff
Permalink | Reply | 0 comments |
headers
Ed McNierney | 1 Dec 2009 14:04
Favicon

Re: About 8.2.2

Philipp -

An OS image signed by OLPC can be booted by any XO-1.0 laptop in the world, except for those which have been
reconfigured by a deployment to only respect software signed by other security keys.  That implies a
higher level of testing and certification than an image that can be selectively adopted by specific
deployments who can do their own testing to decide whether that release is suitable for their
application.  As OLPC's deployments grow both in number of total laptops deployed and in the number of
different localities supported, it becomes increasingly burdensome / difficult to package and test One
Image to Boot Them All worldwide.

As Martin points out, we are continuing to try to move users toward either (a) using machines with
boot-image security disabled, so they can run any software, or (b) using locally-developed and
locally-maintained signature authorities to sign OS images for secure boot in local deployments.

	- Ed

On Dec 1, 2009, at 4:14 AM, Philipp Kocher wrote:

>>  - It won't be signed by OLPC. You have to be on an unlocked XO, or be
>> a deployment signing your own builds.
> 
> Is there a reason why 8.2.2 doesn't get signed by OLPC?
> I do understand that the main target group are big deployments which can 
> sign the build, but why are others excluded?
> 
> In the past even release candidates like build 800 got signed by OLPC.
> 
> Cheers Philipp
> _______________________________________________
> Devel mailing list
(Continue reading)

Permalink | Reply | 4 comments |
headers
Martin Langhoff | 1 Dec 2009 14:44
Picon
Gravatar

Re: New F11 for XO-1.5 build 48

On Mon, Nov 30, 2009 at 7:22 AM, Chris Ball <cjb <at> laptop.org> wrote:
> http://wiki.laptop.org/go/F11_for_1.5
> http://dev.laptop.org/~cjb/f11-1.5/os48

And as the wiki says - remember to use the latest OFW with it. My XO
had a bad case of psychedelic UI with q3a15, cured with q3a16...

m.
--

-- 
 martin.langhoff <at> gmail.com
 martin <at> laptop.org -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff
Permalink | Reply | 0 comments |
headers
Chris Ball | 1 Dec 2009 16:25
Favicon
Gravatar

Re: New F11 for XO-1.5 build 48

Hi,

   > I would prefer to see the backlight dimming take place on a
   > longer time scale (as Paul says, I'm frequently still reading
   > when it happens) and the suspend happening much more
   > aggressively.

Not quite sure what this means -- do you mean the dimming itself should
take longer, or that it should be happening earlier than it does?

   > Does OHM have user configurable knobs ?

Yeah.  The first two numbers in /etc/ohm/plugins.d/timeouts.ini are
the timeouts for beginning dimming and hitting suspend.  I don't have
an option for how long the dim itself takes to happen (but could add
one).  I've changed the downwards dim from 25ms * 6 to 35ms * 6 for
the next build, and removed the dim altogether when coming back up.

Thanks,

- Chris.
--

-- 
Chris Ball   <cjb <at> laptop.org>
One Laptop Per Child
Permalink | Reply | 0 comments |
headers
Franco Miceli | 1 Dec 2009 17:43
Picon

XO 1.5 - CONTENTION WINDOW

Hi,

Looking for the reasons for low throughput on the XO 1.5 I've come across with the contention window's registers.

I've found that there are like eight queues and that for all of them the CWmax and min are set as follows:

CWmin = 7 slots
CWmax = 31 slots

As I see it this would make for a decrease in performance in situations with a high number of customers connected to the same AP.

Can someone explain to me why these registers have been set to such values?
Is this a mistake or it is something done for some reason in particular?

Thanks.

Franco

_______________________________________________
Devel mailing list
Devel <at> lists.laptop.org
http://lists.laptop.org/listinfo/devel
Permalink | Reply | 8 comments |
headers
John Watlington | 1 Dec 2009 18:12
Favicon

fs-update to ext. SD


Just a reminder that it is easy to use fs-update on external SD cards as
well as the internal one.  You just first have to type:
devalias fsdisk /sd/disk <at> 1:0

Cheers,
wad
Permalink | Reply | 1 comment |

Gmane