Search
Power from a different angle
Today is day three with my Neo1973. It is well know by owners of the Neo and members of the OpenMoko development community that the GTA01 suffers many power management issues. Today, I got the first taste of those issues myself when I forgot my USB cable.
The Neo runs on a "basic" mobile battery and it packs a devent punch. A sharp touch screen, GPS, GSM, Bluetooth, speakers and a backlight. In addition, the CPU and memory consume power as they're used. This means, without proper power management systems on the OpenMoko platform, the Neo's battery life is rather short - in my case, about 5 hours.
There's another problem - the power controller firmware doesn't flag "critical" use - what this means is rather than shutting down when the phone's power gets to 5% it will continue to draw power, right until the battery is 100% drained and the system dies - or as SpeedEvil says, "rests" - mercilessly. Why is this so important? Because the Neo has an advantage called "Quick Charge" - when powered on (i.e. running the Linux kernel) it is able to tap the full power it recieves in the USB charging port resulting in a faster charge. However, when the phone is NOT running Linux (dead or running the bootloader, u-boot) it draws about 1/5th of that power resulting in painfully slow charges. So a phone that has "rested" totally need to be slow charged for about an hour before it has enough power to boot into Linux and enable quick charge.
The promising part is the reasons for the total battery death are know and can and will be fixed over time. That's not so much of an important thing.
What IS important to me is that pondering on this caused a paradigm shift. Having been a GNU/Linux user for the past several years (almost a half decade now?) I've seen several Linux kernel release announcements from the exciting release of KVM to things I considered mundane - drive I/O improvements or some such.
One specific area I've always deemed unimportant - power management. Up until recently, it never mattered to me as I'm a full-time desktop user. I hate laptops and have always found them to be novelties. Now, having really looked at how dramatic this is on my Neo, I have a new appreciation for the time and energy that kernel hackers put into these things - and I realize that every patch submittited is someone else's "power management"; their issues that everyone else is oblivious to.
Taking this track along a bit more, I realized just how powerful the Free Software community is. Until today, I've never thought about power on my computer before. The more CPU speed needed, the more power it consumes. The more memory swaps, the more power. More HD? Bright screen? MORE POWER!
My friend Danijel finds a connect between being a geek and "being green" - I do not. Personally, I feel that the concept of Global Warming (implying it's our 'fault' and that we actually have the capacity to screw a planet over in under a hundred years - or more crazily, that we have the capacity to "fix" it if such things were true) to be laughable. I don't, however, think we can ignore issues of pollution, energy consumption and waste - just the opposite in fact. I think these issues are SO important that, like Freedom, they should be addressed on their merits alone, without propaganda scare tactics of an impending global catastrophy to motivate people to do something about it.
That said, until today, my eyes were closed to the sheer about of waste generated by shitty code. The FSF has been pushing that angle for a little while with the Bad Vista campaign, but this isn't an attack on Vista because it applies to us - users of software libre - just as much. How much electricity are we wasting because of crappily optimized code?
Think about this for a moment... There are datacenters FULL of servers on redundant power systems. Each of these servers is running some sort of AMP stack, spinning hard drives and swapping bits around in memory. If Apache itself has a bug that causes just a bit of unneeded movement it has a cascading effect running 24 hours a day on hundreds of machines. How much energy is being consumed because a developer didn't think about being efficient while writing his code?
Free Software users and developers have an added kind of power, the power to consume less. The amount of waste generated by that hypothetical Apache bug is decently large, but imagine now if the bug was in Linux itself - now it's affecting desktop users and laptop users as well as a lot of servers. This isn't merely a case of "bloat" because even on very slim, "minimal" systems it's possible to be doing more than is strictly needed.
We all need to expand our horizons sometimes and I'm glad for having the chance to have done it here with the Neo - it makes me feel powerful. :) Pun intended, unless it's tacky. Then I won't own up to it.
The Neo runs on a "basic" mobile battery and it packs a devent punch. A sharp touch screen, GPS, GSM, Bluetooth, speakers and a backlight. In addition, the CPU and memory consume power as they're used. This means, without proper power management systems on the OpenMoko platform, the Neo's battery life is rather short - in my case, about 5 hours.
There's another problem - the power controller firmware doesn't flag "critical" use - what this means is rather than shutting down when the phone's power gets to 5% it will continue to draw power, right until the battery is 100% drained and the system dies - or as SpeedEvil says, "rests" - mercilessly. Why is this so important? Because the Neo has an advantage called "Quick Charge" - when powered on (i.e. running the Linux kernel) it is able to tap the full power it recieves in the USB charging port resulting in a faster charge. However, when the phone is NOT running Linux (dead or running the bootloader, u-boot) it draws about 1/5th of that power resulting in painfully slow charges. So a phone that has "rested" totally need to be slow charged for about an hour before it has enough power to boot into Linux and enable quick charge.
The promising part is the reasons for the total battery death are know and can and will be fixed over time. That's not so much of an important thing.
What IS important to me is that pondering on this caused a paradigm shift. Having been a GNU/Linux user for the past several years (almost a half decade now?) I've seen several Linux kernel release announcements from the exciting release of KVM to things I considered mundane - drive I/O improvements or some such.
One specific area I've always deemed unimportant - power management. Up until recently, it never mattered to me as I'm a full-time desktop user. I hate laptops and have always found them to be novelties. Now, having really looked at how dramatic this is on my Neo, I have a new appreciation for the time and energy that kernel hackers put into these things - and I realize that every patch submittited is someone else's "power management"; their issues that everyone else is oblivious to.
Taking this track along a bit more, I realized just how powerful the Free Software community is. Until today, I've never thought about power on my computer before. The more CPU speed needed, the more power it consumes. The more memory swaps, the more power. More HD? Bright screen? MORE POWER!
My friend Danijel finds a connect between being a geek and "being green" - I do not. Personally, I feel that the concept of Global Warming (implying it's our 'fault' and that we actually have the capacity to screw a planet over in under a hundred years - or more crazily, that we have the capacity to "fix" it if such things were true) to be laughable. I don't, however, think we can ignore issues of pollution, energy consumption and waste - just the opposite in fact. I think these issues are SO important that, like Freedom, they should be addressed on their merits alone, without propaganda scare tactics of an impending global catastrophy to motivate people to do something about it.
That said, until today, my eyes were closed to the sheer about of waste generated by shitty code. The FSF has been pushing that angle for a little while with the Bad Vista campaign, but this isn't an attack on Vista because it applies to us - users of software libre - just as much. How much electricity are we wasting because of crappily optimized code?
Think about this for a moment... There are datacenters FULL of servers on redundant power systems. Each of these servers is running some sort of AMP stack, spinning hard drives and swapping bits around in memory. If Apache itself has a bug that causes just a bit of unneeded movement it has a cascading effect running 24 hours a day on hundreds of machines. How much energy is being consumed because a developer didn't think about being efficient while writing his code?
Free Software users and developers have an added kind of power, the power to consume less. The amount of waste generated by that hypothetical Apache bug is decently large, but imagine now if the bug was in Linux itself - now it's affecting desktop users and laptop users as well as a lot of servers. This isn't merely a case of "bloat" because even on very slim, "minimal" systems it's possible to be doing more than is strictly needed.
We all need to expand our horizons sometimes and I'm glad for having the chance to have done it here with the Neo - it makes me feel powerful. :) Pun intended, unless it's tacky. Then I won't own up to it.
Leave a Comment
Comment XML feeds: RSS | Atom
# Re: Power from a different angle
Excellent post!
I believe a geek-green connect you mentioned me making was actually more about Free Software geeks than in general, because Free Software is *usually* known to be more efficient because of peer review and because it is considered more of an art and comes much more out of love than out of pressure by some deadlines that have to be met.
I think a big part of the reason why Vista is so power hungry is that the code is too monolithic, developed too much on the rush and hence probably quite hard to manage. So stuff just ends up piling up on top of each other.. I'm not a programmer, but I imagine if some Free Software programmer were to look at the Vista code (s)he'd probably be appalled at the size of the software spaghetti (s)he'd see in there. :P
Anyway, about power management, I never had a laptop and am also a full desktop user and hence never really felt the need for efficient power management on my skin. However, simply due to environmental concerns I've been a bit sensitive to the topic, which is why I at least usually shut my computer off when I sleep rather than leaving it always on as I used to (cause it's cool to be always on IRC and have a comp always accessible at a touch :P ).
I also appreciate the fact that Intel Core Duo processor I have in there now is also quite power efficient, and I think modern components I have in there are overall a lot better than my last computer in regards to this. I'll be paying attention to this issue in the future buying decisions as well.
About software bloat, I totally agree. I use GNOME, which isn't exactly on the slim side, but if kept in check isn't all that terrible I suppose. But things like Mono and unfortunately FIREFOX - they really burn.. Firefox may be the single worst software hog I've run in my PC for a long time. Leave it on for long enough and all 1.5 GB of my memory is gone and then starts the swapping. I guess I just have to learn to close it more often, but then again it still points to something being quite wrong with FF's memory management..
Anyway.. long comment.. so I stop now. Thanks for a great entry!
Cheers