Hacker News Reader

Comment by seba_dos1 | Hacker News Reader

When it comes to used SoC, booting GNU/Linux on it isn't really groundbreaking. I'm much more interested in i.mx6/i.mx8m's power consumption and heat generation. Those chips are often targeted for and used in automobile industry, with bulkier power supply and good cooling, so putting them in a phone might pose some big challenges.

Nevertheless, as a guy who used to be active part of Openmoko, OpenPhoenux and Neo900 communities, I wish them luck!

View on HN

For those of us who just want Linux on the phone, but don't know much about hardware and low level stuff: What can we expect?

How powerful are these going to be? I guess asking how they compare to current phones performance-wise is not the point of this phone, but it would be nice to know where this stands.

Also, if it's not too early to make predictions, how long will the thing last battery-wise? Iirc the laptop situation in Linux had some stuff like powertop which helps, but it's still tricky.

What smartphone SoC has open drivers? If they can make a workable smartphone which runs on the mainline kernel, that seems like dramatic progress.

It's not hard to boot plenty of SoCs on mainline kernel, plenty of open devices use TI OMAPs, for instance, that's pretty well supported. What's hard is to provide proper GPU acceleration, support for all included hardware and power management tailored for particular device. Seems like i.mx6/8 are good with the first one (sadly, OMAP sucks there), that's nice. However, regarding the rest - you don't expect that Librem 5 will be really usable as a phone on mainline Linux on launch and some time (possibly years) after launch, do you?

GTA04 project has had (and still has) a massive aspiration to upstream all the kernel changes and drivers, but they're still not done after many years (although close), because a lot of things that actually make the mobile device usable are device specific hacks that rarely are accepted upstream. Now it boots on upstream kernel (or not, it changes from release to release), but if you really want to use it as a phone, you still have to use Goldelico kernel. Which is of course completely open, just not mainline. Earlier, with Neo Freerunner, the situation was similar, aside of the fact that amount of non-upstreamable hacks were much higher and there was no stuff like device tree back then, so nobody made any serious progress on upstreaming.

Also, booting is one thing. Getting everything in shape to be usable as a every-day device with small battery is another. The road from "booting on mainline" to "workable smartphone on mainline" isn't doable in a year or two, when you also have to design and produce the device during that time.

However, that's not a big issue at all. Mainline eventually will get better and better, and if you have active community and try to minimize the deviation from mainline, rebasing isn't that hard. What can be an issue is power and thermal requirements of used chipset. I don't know about any mobile device that has used it before (but that's doesn't mean it doesn't exist), and I've already seen projects rejecting them as "not for mobile", that's why I'm eagerly waiting for more info about this SoC's performance.

Thanks for the answer, I'm intrigued by what you're saying about the hardware being unsuitable for mobile, surely these are fairly standard components? The range of CPU's they're discussing use off the peg A72 Cortex cores designed by ARM, mixed with some lower power cores in various configurations, using big.Little. All mobile chips massively over-run their thermal capacity, and have to throttle after some minutes running at full power. Work on keeping within the thermal envelope, managing the array of cores and parceling out tasks, seems more to do with software than hardware, and work which has to be done anyway for Linux to be able to function in that environment, and also to be a relatively generalizable problem. That seems to be why you'd support a project like this.

Why do you need device specific hacks, and what about the hardware is inappropriate? Genuinely interested to know.

> What's hard is to provide proper GPU acceleration

I'm still looking for a non-commercial BeagleBone Black image that includes the TI HW accelerated GPU drivers.

The imx6 and imx8 are for the automobile industry but not the imx8m . Librem really wanted to use the M for the reason it used less power ,that's what they said anyway. I believe they will use the quad core A53 variant

http://media.nxp.com/phoenix.zhtml?c=254228&p=irol-newsArtic...

Reply on news.ycombinator.com