Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On Wed, 2007-01-24 at 17:34 +0100, ext Frantisek Dufka wrote: Karl Bellve wrote: Igor Stoppa wrote: Power saving is not so important when the _energy_ wasted is so little and for an event which is supposed to happen very seldom (you don't turn off your device, don't you?). Every time it crashes! :-) When using the Nokia at night, while watching TV in a dark room, the nokia can be absurdly bright during its reboot process. So bright, I have to face the nokia down on the couch while it reboots. So, the bright light is over whelming when you have the nokia set to its dimmest level because your eyes have adjusted to the dark. Exactly. That is my second (and maybe bigger) concern. Reboot or crash at night is not a pleasant thing. While it is supposed to happen very seldom in reality it is not such rare event. And the backligh really does not consume significant power? Check what i wrote, i wrote _energy_ and i'm gonna restate it: the energy wasted by the backlighing durinig a device boot sequence is not so much, when compared to other things that affect runtime power saving. Even 100 mA over less than 60 seconds are really nothing. Otoh 2 mA over few days are _a lot_ I thought it can be half of power consumption on such devices. It was like that on ipaq 3870 I had. I remeber something like 80mA when display off ~130mA minimum brightness ~200mA full brightness, ~300mA full brightness + 100%CPU (not sure about exact values, it is long time ago, it was available as file in /proc in Familiar linux). On N770 I cannot check easily since some parts of power management is ... guess what? There is this myth about hidden power management ;-) Nothinig prevents you from creating a sysfs entry wich returns tahvo_get_backlight_level() or even a proc one, if you really are into that sort of thing. But anyway is there some real reason while the default cannot be much lower until device boots to the point where user specified setting is honored? Frantisek No, just nobody so far had complained so much and from am effectiveness standpoint, it's irrelevant to power saving. -- Cheers, Igor Igor Stoppa [EMAIL PROTECTED] (Nokia M - OSSO /Helsinki Finland) ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
Igor Stoppa wrote: I thought it can be half of power consumption on such devices. It was like that on ipaq 3870 I had. I remeber something like 80mA when display off ~130mA minimum brightness ~200mA full brightness, ~300mA full brightness + 100%CPU (not sure about exact values, it is long time ago, it was available as file in /proc in Familiar linux). On N770 I cannot check easily since some parts of power management is ... guess what? There is this myth about hidden power management ;-) Nothinig prevents you from creating a sysfs entry wich returns tahvo_get_backlight_level() or even a proc one, if you really are into that sort of thing. There is such entry - /sys/devices/platform/omapfb/panel/backlight_level I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. This is easy to get on laptops and it was possible on iPAQ but is currently kept very close to Nokia's chest. Yes one can measure it but it would need some tools, reading some register is much easier. Would be also good guide for programmers implementing proper idle state. I would be interested in difference betwen SD/MMC vs internal flash I/O (i.e. how does rootfs on MMC affect battery life) but other things are equally interesting to know (wlan transmit/receive, sound, DSP on, DSP vs main CPU doing similar stuff ...). Frantisek ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On Thu, 2007-01-25 at 13:08 +0100, ext Frantisek Dufka wrote: Igor Stoppa wrote: I thought it can be half of power consumption on such devices. It was like that on ipaq 3870 I had. I remeber something like 80mA when display off ~130mA minimum brightness ~200mA full brightness, ~300mA full brightness + 100%CPU (not sure about exact values, it is long time ago, it was available as file in /proc in Familiar linux). On N770 I cannot check easily since some parts of power management is ... guess what? There is this myth about hidden power management ;-) Nothinig prevents you from creating a sysfs entry wich returns tahvo_get_backlight_level() or even a proc one, if you really are into that sort of thing. There is such entry - /sys/devices/platform/omapfb/panel/backlight_level I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. This is easy to get on laptops and it was possible on iPAQ but is currently kept very close to Nokia's chest. I think you won't see this information made available anytime soon for a very good reason: there is a significant risk that somebody uses this knowledge to write malicious software that literally blows the device by doing nasty thinigs with the battery and Nokia would be liable if it was releasing such information. This is also behind the clause about reverse engineering. One has to prove that we made it hard for him to get such information, so that we are not liable when he blows hiimself up with a modified 770/n800. Yes one can measure it but it would need some tools, reading some register is much easier. Would be also good guide for programmers implementing proper idle state. We are tryinig to figure out something similar, without risk. But your statement is mostly for kernel side. Applications can use the methods already described. I would be interested in difference betwen SD/MMC vs internal flash I/O (i.e. how does rootfs on MMC affect battery life) but other things are equally interesting to know (wlan transmit/receive, sound, DSP on, DSP vs main CPU doing similar stuff ...). It requires a little hw hacking, but you can measure them quite easily with a good enough power meter (precise to mA) and some wiring. -- Cheers, Igor Igor Stoppa [EMAIL PROTECTED] (Nokia M - OSSO /Helsinki Finland) ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On Thu, Jan 25, 2007 at 12:38:18PM +, Igor Stoppa wrote: On Thu, 2007-01-25 at 13:08 +0100, ext Frantisek Dufka wrote: I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. This is easy to get on laptops and it was possible on iPAQ but is currently kept very close to Nokia's chest. I think you won't see this information made available anytime soon for a very good reason: there is a significant risk that somebody uses this knowledge to write malicious software that literally blows the device by doing nasty thinigs with the battery and Nokia would be liable if it was releasing such information. This is also behind the clause about reverse engineering. Why is this not a problem for laptop manufacturers? One has to prove that we made it hard for him to get such information, so that we are not liable when he blows hiimself up with a modified 770/n800. Marius Gedminas -- Gates' Law: Every 18 months, the speed of software halves. signature.asc Description: Digital signature ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On 1/25/07, Igor Stoppa [EMAIL PROTECTED] wrote: Here we are really talking of being able to charge the battery outside its safety ranges. Given the right knowlege it is possible cause do phisical damage to the device and the user. No, we're talking about finding out how much the battery is being used. This is obviously of benefit to end users without accurate multimeters. On Thu, 2007-01-25 at 13:08 +0100, ext Frantisek Dufka wrote: I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. End-users being able to know if something's (albeit not exactly what) is driving the battery consumption through the roof would be of immeasurable benefit, especially since the N800 is pushed as an always on device. Cheers, Andrew -- Andrew Flegg -- mailto:[EMAIL PROTECTED] | http://www.bleb.org/ ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
Andrew Flegg wrote: On 1/25/07, Igor Stoppa [EMAIL PROTECTED] wrote: Here we are really talking of being able to charge the battery outside its safety ranges. Given the right knowlege it is possible cause do phisical damage to the device and the user. No, we're talking about finding out how much the battery is being used. This is obviously of benefit to end users without accurate multimeters. Seems like the design of the device is so unfortunate that they simply cannot give us that. While my imagination is not good enough to know how it could be, I do believe it. Let's close this long topic :-) Igor also mentioned this: We are tryinig to figure out something similar, without risk. So let's hope we will see something useful in future. Frantisek ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On 1/25/07, Andrew Flegg [EMAIL PROTECTED] wrote: On 1/25/07, Igor Stoppa [EMAIL PROTECTED] wrote: Here we are really talking of being able to charge the battery outside its safety ranges. Given the right knowlege it is possible cause do phisical damage to the device and the user. No, we're talking about finding out how much the battery is being used. This is obviously of benefit to end users without accurate multimeters. On Thu, 2007-01-25 at 13:08 +0100, ext Frantisek Dufka wrote: I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. End-users being able to know if something's (albeit not exactly what) is driving the battery consumption through the roof would be of immeasurable benefit, especially since the N800 is pushed as an always on device. Something like a CPU meter but for wattage consumed at any given point would be great. Even just the raw battery voltage would be nice, although having the estimate on battery life is at least an improvement in IT2007. As an aside, you need a fairly sophisticated meter to measure the current-- it's not like you can pick up a $30 meter and expect to get good readings. The 770 and n800 consume current in large bursts and if the internal resistance of the meter is too high (i.e. when set to read mA) the device won't run. I had to use a 4-wire ratio-resistance method with a resistor (piece of thin wire actually) around 50/1000 of a ohm, hooked to a desktop 6.5 digit DVM. If anyone has some tests for power consumption they needd I can run them, within limits of course. Larry ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On Thursday 25 January 2007 04:38, Igor Stoppa wrote: On Thu, 2007-01-25 at 13:08 +0100, ext Frantisek Dufka wrote: Igor Stoppa wrote: I thought it can be half of power consumption on such devices. It was like that on ipaq 3870 I had. I remeber something like 80mA when display off ~130mA minimum brightness ~200mA full brightness, ~300mA full brightness + 100%CPU (not sure about exact values, it is long time ago, it was available as file in /proc in Familiar linux). On N770 I cannot check easily since some parts of power management is ... guess what? There is this myth about hidden power management ;-) Nothinig prevents you from creating a sysfs entry wich returns tahvo_get_backlight_level() or even a proc one, if you really are into that sort of thing. There is such entry - /sys/devices/platform/omapfb/panel/backlight_level I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. This is easy to get on laptops and it was possible on iPAQ but is currently kept very close to Nokia's chest. I think you won't see this information made available anytime soon for a very good reason: there is a significant risk that somebody uses this knowledge to write malicious software that literally blows the device by doing nasty thinigs with the battery and Nokia would be liable if it was releasing such information. This is also behind the clause about reverse engineering. One has to prove that we made it hard for him to get such information, so that we are not liable when he blows hiimself up with a modified 770/n800. I'm sorry but if the above were a valid reason then I would highly recommend that Nokia shut down it's doors. No amount of effort on Nokia's part will halt malicious lawsuits. Period. Further the above is invalid if not only because you are using OSS as your base and it is governed by a legally tested license, but also because people are asking for access to public information not anything proprietary and bound by NDA's. If you are modifying OSS and choose intentionally to refuse to divulge modifications to this software please let me know. James Yes one can measure it but it would need some tools, reading some register is much easier. Would be also good guide for programmers implementing proper idle state. We are tryinig to figure out something similar, without risk. But your statement is mostly for kernel side. Applications can use the methods already described. I would be interested in difference betwen SD/MMC vs internal flash I/O (i.e. how does rootfs on MMC affect battery life) but other things are equally interesting to know (wlan transmit/receive, sound, DSP on, DSP vs main CPU doing similar stuff ...). It requires a little hw hacking, but you can measure them quite easily with a good enough power meter (precise to mA) and some wiring. ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On Thu, 2007-01-25 at 14:48 -0800, ext James Sparenberg wrote: On Thursday 25 January 2007 04:38, Igor Stoppa wrote: On Thu, 2007-01-25 at 13:08 +0100, ext Frantisek Dufka wrote: Igor Stoppa wrote: I thought it can be half of power consumption on such devices. It was like that on ipaq 3870 I had. I remeber something like 80mA when display off ~130mA minimum brightness ~200mA full brightness, ~300mA full brightness + 100%CPU (not sure about exact values, it is long time ago, it was available as file in /proc in Familiar linux). On N770 I cannot check easily since some parts of power management is ... guess what? There is this myth about hidden power management ;-) Nothinig prevents you from creating a sysfs entry wich returns tahvo_get_backlight_level() or even a proc one, if you really are into that sort of thing. There is such entry - /sys/devices/platform/omapfb/panel/backlight_level I meant way to get discharge current from battery circuit to know what is actual consumption of the device and how big is the influence of various parts. This is easy to get on laptops and it was possible on iPAQ but is currently kept very close to Nokia's chest. I think you won't see this information made available anytime soon for a very good reason: there is a significant risk that somebody uses this knowledge to write malicious software that literally blows the device by doing nasty thinigs with the battery and Nokia would be liable if it was releasing such information. This is also behind the clause about reverse engineering. One has to prove that we made it hard for him to get such information, so that we are not liable when he blows hiimself up with a modified 770/n800. I'm sorry but if the above were a valid reason then I would highly recommend that Nokia shut down it's doors. No amount of effort on Nokia's part will halt malicious lawsuits. Period. Advice registered. Period. Further the above is invalid if not only because you are using OSS as your base and it is governed by a legally tested license, but also because people are asking for access to public information not anything proprietary and bound by NDA's. If you are modifying OSS and choose intentionally to refuse to divulge modifications to this software please let me know. If only you would have spent some time reading (this thread, for example) before writing ... but one cannot expect such courtesy from everybody. James Yes one can measure it but it would need some tools, reading some register is much easier. Would be also good guide for programmers implementing proper idle state. We are tryinig to figure out something similar, without risk. But your statement is mostly for kernel side. Applications can use the methods already described. I would be interested in difference betwen SD/MMC vs internal flash I/O (i.e. how does rootfs on MMC affect battery life) but other things are equally interesting to know (wlan transmit/receive, sound, DSP on, DSP vs main CPU doing similar stuff ...). It requires a little hw hacking, but you can measure them quite easily with a good enough power meter (precise to mA) and some wiring. ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users -- Cheers, Igor Igor Stoppa [EMAIL PROTECTED] (Nokia M - OSSO /Helsinki Finland) ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
Igor Stoppa wrote: Power saving is not so important when the _energy_ wasted is so little and for an event which is supposed to happen very seldom (you don't turn off your device, don't you?). Every time it crashes! :-) When using the Nokia at night, while watching TV in a dark room, the nokia can be absurdly bright during its reboot process. So bright, I have to face the nokia down on the couch while it reboots. So, the bright light is over whelming when you have the nokia set to its dimmest level because your eyes have adjusted to the dark. begin:vcard fn:Karl Bellve n:Bellve;Karl org:Biomedical Imaging Group;Department of Physiology adr:2 Biotech, Suite 114;;373 Plantation Street;Worcester;Massachusetts;01655;United States of America email;internet:[EMAIL PROTECTED] title:Assistant Professor tel;work:508-856-6514 tel;fax:508-856-6514 x-mozilla-html:TRUE version:2.1 end:vcard ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
Karl Bellve wrote: Igor Stoppa wrote: Power saving is not so important when the _energy_ wasted is so little and for an event which is supposed to happen very seldom (you don't turn off your device, don't you?). Every time it crashes! :-) When using the Nokia at night, while watching TV in a dark room, the nokia can be absurdly bright during its reboot process. So bright, I have to face the nokia down on the couch while it reboots. So, the bright light is over whelming when you have the nokia set to its dimmest level because your eyes have adjusted to the dark. Exactly. That is my second (and maybe bigger) concern. Reboot or crash at night is not a pleasant thing. While it is supposed to happen very seldom in reality it is not such rare event. And the backligh really does not consume significant power? I thought it can be half of power consumption on such devices. It was like that on ipaq 3870 I had. I remeber something like 80mA when display off ~130mA minimum brightness ~200mA full brightness, ~300mA full brightness + 100%CPU (not sure about exact values, it is long time ago, it was available as file in /proc in Familiar linux). On N770 I cannot check easily since some parts of power management is ... guess what? There is this myth about hidden power management ;-) But anyway is there some real reason while the default cannot be much lower until device boots to the point where user specified setting is honored? Frantisek ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
OK, reported as https://maemo.org/bugzilla/show_bug.cgi?id=965 so it is not forgotten. Frantisek ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users
Re: brightness Re: [maemo-users] Re: Storage/Sleep Issues..
On Wednesday 24 January 2007 06:52, Karl Bellve wrote: Igor Stoppa wrote: Power saving is not so important when the _energy_ wasted is so little and for an event which is supposed to happen very seldom (you don't turn off your device, don't you?). Every time it crashes! :-) When using the Nokia at night, while watching TV in a dark room, the nokia can be absurdly bright during its reboot process. So bright, I have to face the nokia down on the couch while it reboots. So, the bright light is over whelming when you have the nokia set to its dimmest level because your eyes have adjusted to the dark. Good this means I'm not the only one who has blinded himself doing the exact same thing. ;) However it makes a great nightlight when needing to get up and go to the restroom. James ___ maemo-users mailing list maemo-users@maemo.org https://maemo.org/mailman/listinfo/maemo-users