Pandora Development Laptop

[sindbad]

For a while now I've been thinking that I don't really need x86 on my laptop. And I'd love it if my laptop was entirely silent, didn't heat up like crazy and had a long lasting battery life.

Perhaps OpenPandora might want to build such a laptop, to make developing for the Pandora easier?
Here's the draft specs of my (almost) ideal laptop:

Chipset: Cortex A8-9 or similar, 0.5-1GB ram
Storage: 10GB+ SSD, at least one SD slot
Display: 13", ideally Pixel Qi/similar sunlight readable
Keyboard: full size, utility buttons (brightness, volume, etc.) as Fn+F keys
Pointing: large (capacitive) trackpad, maybe one pandora nub too
Connectors: 2-4 USBs, audio out(&in*), VGA/DVI/other digital output + adapter
Audio: hardware volume control (or at least mute switch), integrated mic, speakers (however small)
Video: DSP support in video player (mplayer/gstreamer), webcam
Battery: a minimum of 5h, as much as possible within 1-1.5kg

There could be an optional USB/BT controller that mimics the Pandora's layout as much as possible.

What do you all think?

 

[Aurekana]

You could gut an old laptop and replace the motherboard with a beagleboard. Might be interesting interfacing the LCD and keyboard though...

 

[Exophase]

lkcl was and quite possibly still interested in doing something like this:

http://www.gp32x.de/board/index.php?/topic/52705-help-needed-replicating-success-of-openpandora-for-a-netbooklaptop

That said, you've got to understand that while gaming handhelds like Pandora are a niche, netbooks/laptops are not. Someone is bound to hit most or all of the things you want. There are already ARM based netbooks out there, like Toshiba AC100. And with the release of Nufront devices next year we'll see offerings much more competitive with current Atoms and possibly even low end i3's and the upcoming Bobcats. You also have to understand that with a 13" screen you're not going to save a huge amount of battery life using a 1W SoC instead of a 5W one. You can probably run an Atom netbook without a fan too. Any higher end x86 is going to blow away any currently available ARM solution so there's no real point comparing them.

OP wouldn't be doing themselves a lot of favors in entering this sort of market where they'd easily be put to bed. They're spreading themselves thin enough as it is.

 

[sindbad]

Didn't know about lkcl's idea.

Right now, ARM's advantage in power usage is almost an order of magnitude. Especially when it comes to idle and suspending. 10" ARM netbooks, like the Efika MX smartbook, still have impressive battery life, at weight and thickness lower than current atom netbooks. 13" isn't much bigger, especially with low-power displays like the Pixel Qi.

I do understand why it might now be in OP's interest to get involved in such a project. I was just checking

 

[Exophase]

A typical ARM SoC doesn't have an order of magnitude power usage advantage over a typical current Atom chipset, especially not when scaled to similar performance levels. Having a 13" screen doesn't just mean you're taking up probably more power with it than the SoC, but it also means you have that much more room for a battery. The Efika MX you mentioned only claims "less than 12W." Its impressive battery life is only 6 hours, which is less than you'd get from a typical Atom netbook, especially after limited to one core at 800MHz. The only real advantage is saving a small amount of weight on battery; the actual weight of about 2lbs isn't that much different from the 2-3 you tend to see for Atom netbooks. All for a price that's higher than Atom netbooks cost. Is it really worth it?

 

[sindbad]

It all depends on how you define your battery life. The Efika MX has been observed to compile kernels back-to-back for 8h at the recent Ubuntu Developer Summit. I think the efika folk are being overly conservative.

Most Atom netbooks have their battery life measured at idle, with everything turned off. If you buy a high-quality battery for it, you might reach 8h, but it's unlikely, noisy and hot. x86 is a heavy burden in terms of heat and power, and it really shows in small devices.

For me, something like the Efika MX with a 13" Pixel Qi display and maybe with Cortex A9 would be worth at least as much as any other 13" laptop.

 

[Exophase]

It all depends on how you define your battery life. The Efika MX has been observed to compile kernels back-to-back for 8h at the recent Ubuntu Developer Summit. I think the efika folk are being overly conservative.

Was the screen on while it was compiling kernels? It certainly wouldn't need to be. I just don't see how something that's rated at "less than 12W" gets 8 hours on a "3 cell battery" assuming that that means around 24 Wh.

Most Atom netbooks have their battery life measured at idle, with everything turned off.

That isn't true at all. Any netbook battery benchmark worth anything will include numbers at full load and those numbers can reach > 7h. You might need 2x the battery of the Efika MX to do it, but you get 2x the performance...

Bear in mind that the N4xx Atom platforms have lower power consumption than the N2xx ones did because they integrate the memory controller and graphics in the SoC. TDP of CPU + support chipset is around 6.5W. That's up to 1.83GHz. You can get dual core 1.5GHz at a TDP of 8.5W.

If you buy a high-quality battery for it, you might reach 8h, but it's unlikely, noisy and hot. x86 is a heavy burden in terms of heat and power, and it really shows in small devices.

Plenty of Atom netbooks are fanless and don't have harddrives, so where is this noise coming from? The heat is also not exactly a highly claimed problem. I think you are under a lot of misconceptions, the power burden Atom has merely for being x86 is greatly exaggerated. Don't get me wrong, ARM, particularly Cortex-A9, has better perf/W and makes a lot more sense in deeply mobile like phones. But netbooks aren't phones. For ARM to work in netbooks they have to offer performance that is at least as good and prices that are at least as low, oh, and they need to not run Android (no thanks Toshiba AC100). The upcoming Nufront dual-core 2GHz Cortex-A9 SoC could make a big difference here. But an i.MX51 is incredibly weak and the power saved just isn't worth much of anything when compared to the power spent on the rest of the thing.

For me, something like the Efika MX with a 13" Pixel Qi display and maybe with Cortex A9 would be worth at least as much as any other 13" laptop.

The only thing a Cortex-A9 laptop with a 13" Pixel Qi display has in common with Efika MX is that they're both ARM, which by itself isn't an asset for most people. It doesn't really help that Efika MX is priced higher than netbooks with superior performance..

 

[sindbad]

Was the screen on while it was compiling kernels? It certainly wouldn't need to be. I just don't see how something that's rated at "less than 12W" gets 8 hours on a "3 cell battery" assuming that that means around 24 Wh.

The person who did this was also reading something on it, so the screen was on. I don't know about brightness, though.

Most Atom netbooks have their battery life measured at idle, with everything turned off.

That isn't true at all. Any netbook battery benchmark worth anything will include numbers at full load and those numbers can reach > 7h. You might need 2x the battery of the Efika MX to do it, but you get 2x the performance...

Manufacturers do measure and advertise battery life at (mostly) idle. And I don't want 2x the battery for 2x the performance, I want 2x the battery for 1x the performance for 2x te battery life on a laptop. And 1x the battery for 1x the performance for lower weight on a netbook.

Plenty of Atom netbooks are fanless and don't have harddrives, so where is this noise coming from? The heat is also not exactly a highly claimed problem.

All the fanless Atom netbooks have large heatsinks and they still tend to overheat and fail if under high load a lot. Perhaps future Atom designs won't have this issue, but the current generation can't reliably run without a fan under high load. VIA's CPUs are better, my OLPC XO 1.5 does get warm when pushed hard, but not very.

I think you are under a lot of misconceptions, the power burden Atom has merely for being x86 is greatly exaggerated. Don't get me wrong, ARM, particularly Cortex-A9, has better perf/W and makes a lot more sense in deeply mobile like phones. But netbooks aren't phones. For ARM to work in netbooks they have to offer performance that is at least as good and prices that are at least as low, oh, and they need to not run Android (no thanks Toshiba AC100). The upcoming Nufront dual-core 2GHz Cortex-A9 SoC could make a big difference here. But an i.MX51 is incredibly weak and the power saved just isn't worth much of anything when compared to the power spent on the rest of the thing.

I believe you are under a misconception as well. CPUs haven't been the bottleneck for a long time, even weak ones like i.MX51. This is especially true for netbook-like usage. My OLPC XO 1.0 has a 400Mhz CPU and it's fine. The only problem is RAM (256MB) .It's hard drives or slow SSDs that slow them down (or the network, but that's obvious).

And yeah, while Android is great on phones, it's crap on netbooks. Toshiba are idiots for not using Ubuntu or something similar.

The only thing a Cortex-A9 laptop with a 13" Pixel Qi display has in common with Efika MX is that they're both ARM, which by itself isn't an asset for most people. It doesn't really help that Efika MX is priced higher than netbooks with superior performance..

Ok, they don't have much more in common than x86 Atom netbooks and x83 i3 13" laptops do.

ARM netbooks have been marketed badly so far. ARM manufacturers should advertise 12h battery life, since by the common standards in the x86 world that's at least how much they'd get. They also need to stop putting silly smartphone OSes on them.

 

[Exophase]

Manufacturers do measure and advertise battery life at (mostly) idle. And I don't want 2x the battery for 2x the performance, I want 2x the battery for 1x the performance for 2x te battery life on a laptop. And 1x the battery for 1x the performance for lower weight on a netbook.

I only go off of tests that independent reviewers do. There are N450 Atom laptops that can achieve over 10 hours on a "6-cell" battery, and not while idling. You should look at some real reviews.

All the fanless Atom netbooks have large heatsinks and they still tend to overheat and fail if under high load a lot. Perhaps future Atom designs won't have this issue, but the current generation can't reliably run without a fan under high load. VIA's CPUs are better, my OLPC XO 1.5 does get warm when pushed hard, but not very.

Do you have any hard data on failure rate due to heat? OLPC XO isn't a VIA CPU, it's a Geode LX, manufactured by AMD but with Cyrix heritage. And it doesn't generate much heat because it's incredibly slow.

I believe you are under a misconception as well. CPUs haven't been the bottleneck for a long time, even weak ones like i.MX51. This is especially true for netbook-like usage. My OLPC XO 1.0 has a 400Mhz CPU and it's fine. The only problem is RAM (256MB) .It's hard drives or slow SSDs that slow them down (or the network, but that's obvious).

By that line of reasoning Apple should have never released a phone past 3G, but a lot of people have been very happy with the performance improvements in 3GS and iPhone 4.. Your needs may not be very demanding but you can't claim they represent the majority. If you're happy with a Geode LX at 400MHz then you should be happy with an Atom at 400MHz, which would perform better and would use far less power than one at 1.6GHz.

And yeah, while Android is great on phones, it's crap on netbooks. Toshiba are idiots for not using Ubuntu or something similar.

Unfortunately, that's one of the few things they can try to do to differentiate itself from the run of the mill. The result sucks though.

Ok, they don't have much more in common than x86 Atom netbooks and x83 i3 13" laptops do.

Yes, and for that matter not much more in common than x86 netbooks and ARM netbooks do, save for the ability to run Windows.

ARM netbooks have been marketed badly so far. ARM manufacturers should advertise 12h battery life, since by the common standards in the x86 world that's at least how much they'd get. They also need to stop putting silly smartphone OSes on them.

They can advertise 100 hour battery life and it won't change anything, they'll just get outed by reviews that show that they're actually not really that much lower power than Atom ones.

In order for a netbook to even begin to challenge the current status quo of Atom + Windows it needs to:

- Have a lot of choices that don't get delayed a year
- Be cheaper than the rest, so far this isn't happening
- Offer at least as much performance as the rest, this definitely hasn't happened yet
- Have a well packaged OS.. it actually took Ubuntu for a while to be even remotely decent for this form factor

But it'll still be at a massive disadvantage by not running Windows.

Proportionately speaking, the power savings advantage you get by going from Atom to Cortex-A8 or Cortex-A9 while at similar performance levels is a tiny fraction of what it is when you go from what middle of the road laptops were to what netbooks are. So there was enough of a compelling push, plus it opened up a new market for a more convenient smaller form factor and most of all something much cheaper. ARM netbooks are fighting in an established market and no matter how much a few fringe people prefer them there has to be enough mass market appeal for them to really take off.

 

[sebt3]

This thread is entertaining

 

[sindbad]

Do you have any hard data on failure rate due to heat?

True, the plural of anecdote is not quite data.

OLPC XO isn't a VIA CPU, it's a Geode LX, manufactured by AMD but with Cyrix heritage. And it doesn't generate much heat because it's incredibly slow.

I was talking about the XO 1.5, which has a VIA C7-m.

By that line of reasoning Apple should have never released a phone past 3G, but a lot of people have been very happy with the performance improvements in 3GS and iPhone 4.. Your needs may not be very demanding but you can't claim they represent the majority. If you're happy with a Geode LX at 400MHz then you should be happy with an Atom at 400MHz, which would perform better and would use far less power than one at 1.6GHz.

In that particular case, there was a need for a faster CPU (which btw isn't all that fast). A geode at 400MHz isn't enough of course, but an Atom/C7-m/Cortex A9 at 1GHz is.

Unfortunately, that's one of the few things they can try to do to differentiate itself from the run of the mill. The result sucks though..

They could claim Android, Chrome OS and Ubuntu, at the user's choice.

Yes, and for that matter not much more in common than x86 netbooks and ARM netbooks do, save for the ability to run Windows.

They can advertise 100 hour battery life and it won't change anything, they'll just get outed by reviews that show that they're actually not really that much lower power than Atom ones.

In order for a netbook to even begin to challenge the current status quo of Atom + Windows it needs to:

- Have a lot of choices that don't get delayed a year
- Be cheaper than the rest, so far this isn't happening
- Offer at least as much performance as the rest, this definitely hasn't happened yet
- Have a well packaged OS.. it actually took Ubuntu for a while to be even remotely decent for this form factor

But it'll still be at a massive disadvantage by not running Windows.

Proportionately speaking, the power savings advantage you get by going from Atom to Cortex-A8 or Cortex-A9 while at similar performance levels is a tiny fraction of what it is when you go from what middle of the road laptops were to what netbooks are. So there was enough of a compelling push, plus it opened up a new market for a more convenient smaller form factor and most of all something much cheaper. ARM netbooks are fighting in an established market and no matter how much a few fringe people prefer them there has to be enough mass market appeal for them to really take off.

Yes, it's sad that Flash managed to delay ARM netbooks. I hope when they do eventually come out they won't be too late.

I don't think the ratio is that sad, especially because of how much better suspend usually works on ARM. We'll have to wait for Efika and friends to come out.

This thread is entertaining

I wouldn't have it any other way

 

[Exophase]

I was talking about the XO 1.5, which has a VIA C7-m.

I had no idea they released more.. how do you have one? Was there another give one get one sort of thing with it? The wiki says it hasn't even reached production, is it out of date?

In that particular case, there was a need for a faster CPU (which btw isn't all that fast). A geode at 400MHz isn't enough of course, but an Atom/C7-m/Cortex A9 at 1GHz is.

Of course everything is a manner of opinion on what is enough. I would argue that if C7-m was ever sufficient it would have taken the netbook market, not Atom.. it was out for years prior.

The thing is, an 800MHz Silverthorne Atom (Z500) will outperform a C7 at 1GHz in most cases, but uses only 0.65W instead of 5W. Of course this isn't a netbook chip, but it shows that x86 is not inappropriate for netbooks based on power consumption alone. I have little doubt that current N class will outperform C7 in perf/Watt too. It's really a very dated chip. Nano, on the other hand, does beat Atom in performance but consumes far too much power to do it.

Unfortunately, that's one of the few things they can try to do to differentiate itself from the run of the mill. The result sucks though..

They could claim Android, Chrome OS and Ubuntu, at the user's choice.

The fact is, user's don't really want flexibility and freedom, they want a device that feels made to do some particular thing in some particular environment. What devices like the AC100 are going for by putting Android on a netbook is similar to what Apple went for by putting iOS on a tablet. Something to differentiate it from the "generic" platforms everyone else perceives.

But Android still sucks on netbooks, and for the niche crowd that actually did end up wanting ARM in a netbook it's a hindrance.

Yes, it's sad that Flash managed to delay ARM netbooks. I hope when they do eventually come out they won't be too late.

I'm afraid it's already too late for something i.MX51 level and even Tegra 2 level will have a hard time picking up any sales. People want something substantially more powerful than their phones, and so far every ARM SoC out is optimized for low power and not high performance, although I can bet you the high performance flow ARM does was made precisely with things like netbooks in mind. There's also the matter of having SoCs that support SATA and PCI natively. I have decent hopes for Nufront netbooks utilizing this, assuming those are made and not just desktops.

I don't think the ratio is that sad, especially because of how much better suspend usually works on ARM. We'll have to wait for Efika and friends to come out.

When things are out I'm sure people will be doing battery life tests to compare the two so I guess we'll see then. Why would suspend work better on ARM? Low power modes might draw less and last longer, but suspend can go all the way down to non-volatile and shut off completely, consuming only whatever residual is taken off the battery while the thing is on. The former is a software problem, the latter is an electrical design problem that's hopefully not connected to any of the main electronics.

 

[sindbad]

I was talking about the XO 1.5, which has a VIA C7-m.

I had no idea they released more.. how do you have one? Was there another give one get one sort of thing with it? The wiki says it hasn't even reached production, is it out of date?

The wiki is up to date. I'm a developer for Sugar and that's how I got one. From the packaging I got it in I'm guessing there might be another G1G1 program, but don't quote me on that. Btw, it's a pretty awesome laptop. Indestructible afaict and the screen is amazing.

The fact is, user's don't really want flexibility and freedom, they want a device that feels made to do some particular thing in some particular environment. What devices like the AC100 are going for by putting Android on a netbook is similar to what Apple went for by putting iOS on a tablet. Something to differentiate it from the "generic" platforms everyone else perceives.

That's sad. Are people so clueless? I guess they might be, looking at iPad sales.

I'm afraid it's already too late for something i.MX51 level and even Tegra 2 level will have a hard time picking up any sales. People want something substantially more powerful than their phones, and so far every ARM SoC out is optimized for low power and not high performance, although I can bet you the high performance flow ARM does was made precisely with things like netbooks in mind. There's also the matter of having SoCs that support SATA and PCI natively. I have decent hopes for Nufront netbooks utilizing this, assuming those are made and not just desktops.

From what I've heard, some of those delayed netbooks are switching to newer designs. OMAP4 looks like it might fit the bill.

When things are out I'm sure people will be doing battery life tests to compare the two so I guess we'll see then. Why would suspend work better on ARM? Low power modes might draw less and last longer, but suspend can go all the way down to non-volatile and shut off completely, consuming only whatever residual is taken off the battery while the thing is on. The former is a software problem, the latter is an electrical design problem that's hopefully not connected to any of the main electronics.

Yes, for one their low power modes are simply better. The XO uses a DCON chip to keep the display on with the CPU off in reading mode, that might not be necessary with ARM.
Also, ARM SoCs tend to have much better RAM, that draws almost no power in suspend and often also wake up from suspend faster. Of course a good x86 SoC could have the same advantage, but so far they don't seem to care.