Pyra and Raspberry Pi?

[lukey]

Seems to me that the only things the RPi and Pyra have in common are the community support and the openness ... so the only way to merge those two project could be via software, with same OS that uses the same packages etc...

1) Check out this handheld computer with a standard Linux distribution using an OMAP5
2) Check out this computer casing for the Raspberry Pi

3) Turn this cool handheld Computer to an Ultra-Mobile Raspberry Pi with Raspbian

 

[Grench]

This idea is a 100% move the opposite direction. It is stating that this project needs a less powerful SoC with fewer features. That the Pyra should be seen as a comodity accessory.

I would much rather see the Pyra move in the other direction. The next SoC (year+ away) should be on a 14nm process and include support for all ports & hardware on the current main board and preferably have all on-SoC component pieces supported in the main Linux kernel.

-Nvidia has a monster of a mobile ARM SoC coming up - they're unlikely to sell it to a small project or share documentation.
-Qualcom has some awesome stuff - they're unlikely to to sell it to a small project or share documentation.
-TI has exited the game - however, we should send them a big cheesy card with all our signatures as a thank you for the SoC and support that we have gotten.
-Intel has ended development for a directo successor to the Cherry Trail SoC. However, it is readily available, well documented, on-SoC components are supported in the Linux kernel.

Whatever manufacturer or architecture gets considered for the next SoC, it needs to be a step forward in both speed and performance/Watt from the current OMAP 5 AND support the hardware on the existing main board. The Pi SoC won't be doing all of that inside of the next 5+ years.

 

[Askarus]

Exactly.

-Nvidia has a monster of a mobile ARM SoC coming up - they're unlikely to sell it to a small project or share documentation.

Might be possible.
The issue why we didn't get the Nvidia chip was because Nvidia only wants the board to be produced by an Nvidia certified manufacturer and we wanted Global Components to produce the PCBs.
If we'll produce the CPU-Board elsewhere where Nvidia wants us to do we might have better chances.

 

[sulu]

it wouldn't exactly be free press, since it would cost some money to get an rpi compute module in for the Pyra CPU board (and that in and of itself wouldn't necessarily be easy). but it would raise awareness of the project in a similar demographic, even if it's more expensive than an rpi. if even 1% of the rpi fans liked the product, that might still be a net win. and we could explain that the original Pyra CPU is a better CPU anyways .

You don't actually need to integrate a RPi into the Pyra for that. The PR alone will suffice.
Maybe someone should ask the people at PGS for their expertise. They seem to be quite capable of making good PR.

 

[ible]

oh, so ED posts a video of the "Raspberry Pyra" on youtube, but says that it's just a prototype using a slightly different CPU?

 

[rygD]

I mean for future generations..

Yes, think of the children.

 

[rSl]

imo the main physical difference between the rpi and the pyra is that one is a sbc board and the other is a handheld (tada!).
in a handheld type computer one really needs a lot of functionality embedded, as it has to be somewhat selfcontained so one can hold it... in the hands.

so this makes a handheld initially more complex and costly as an sbc board, where one can add things externally without caring for this handheld factor.
if i remember correctly, hardkernel wanted to buy this soc from the rpi foundation/broadcom for a wearable odroid, and they did not sell them any.
what i do not like about the rpi foundation is that for the rpi2 and above they stopped releasing schematics and the bootloader is still closedsource, i think.

having said this, i still like my rpi2 alot and i'm using it as my current main wall-powered homecomputer.
i also have a rpi3 but it's still in it's cardbox as i dislike to use a fan to make it cool enough for long compute-sessions without constantly throtteling down the speed.

that the development of the rpi is not communicated in their boards is not really great too imho.
a new supersecretly developed rpi gets released, but without community involvement in the design and no eben upton writing in his own boards, unlike ed
who makes polls about colors and whishes and is a supernice community host all around here on these boards who really cares.

i whish the rpi foundation all the best, but for the future i like the pyra to be my main wall-powered homecomputer and handheld in one system.

 

[benoitb]

imo the main physical difference between the rpi and the pyra is that one is a sbc board and the other is a handheld (tada!).
in a handheld type computer one really needs a lot of functionality embedded, as it has to be somewhat selfcontained so one can hold it... in the hands.

Did you write this while knowing about the Raspberry Pi compute module ?
https://www.raspberrypi.org/blog/raspberry-pi-compute-module-new-product/

It is the same type of product as Pyra's CPU board. As far as I understand, thay only offer Raspberry Pi 1's SoC and level of performance in this form factor however.

 

[kaprikawn]

This idea is a 100% move the opposite direction. It is stating that this project needs a less powerful SoC with fewer features. That the Pyra should be seen as a comodity accessory.

I would much rather see the Pyra move in the other direction. The next SoC (year+ away) should be on a 14nm process and include support for all ports & hardware on the current main board and preferably have all on-SoC component pieces supported in the main Linux kernel.

-Nvidia has a monster of a mobile ARM SoC coming up - they're unlikely to sell it to a small project or share documentation.
-Qualcom has some awesome stuff - they're unlikely to to sell it to a small project or share documentation.
-TI has exited the game - however, we should send them a big cheesy card with all our signatures as a thank you for the SoC and support that we have gotten.
-Intel has ended development for a directo successor to the Cherry Trail SoC. However, it is readily available, well documented, on-SoC components are supported in the Linux kernel.

Whatever manufacturer or architecture gets considered for the next SoC, it needs to be a step forward in both speed and performance/Watt from the current OMAP 5 AND support the hardware on the existing main board. The Pi SoC won't be doing all of that inside of the next 5+ years.

I can't help but feel like you're only looking at this from specifically your point-of-view. I don't want or need a Pyra with RPi either. I want future upgrades with the latest SoC too. People who frequent this board, who are invested in the Pyra, have reasonably similar ideas about what they want from the Pyra on the whole I'd imagine.

What I think is that aligning the RPi name to the Pyra, even loosely, could bring in other people whose outlook is different to ours, whose requirements are different to ours.

I agree with the sentiment that RPi enthusiasts are largely looking for cheap and cheerful, and that's at odds with the premium cost of the Pyra. But the RPi community is not a hive mind who all think as one. The RPi community is large, and even if a small proportion of them become interested in the Pyra, in absolute numbers that could translate into a fair few people.

 

[Yasu]

The idea of having a DS sized portable Pi would be fun, but it doesn't sound like the Pyra and Pi have that much in common. Pyra is meant to be up to date and powerful, the Pi is meant to be cheap, hackable and have lots of home made addons.

But it would be nice to have Raspian as an alternative OS for the Pyra.

 

[ible]

Did you write this while knowing about the Raspberry Pi compute module ?
https://www.raspberrypi.org/blog/raspberry-pi-compute-module-new-product/

It is the same type of product as Pyra's CPU board. As far as I understand, thay only offer Raspberry Pi 1's SoC and level of performance in this form factor however.

there's chatter of a compute module 3 coming out (same processor as rpi 3).

http://hackaday.com/2016/07/15/the-raspberry-pi-3-compute-module-is-on-its-way/

 

[sulu]

But it would be nice to have Raspian as an alternative OS for the Pyra.

Why?
The original point of Raspbian was to recompile Debian, because the ARMv6 CPU wasn't compatible with Debian/armel (which requires an ARMv7 CPU). That point became obsolete with the introduction of the RPi2 (that even supports armhf). All that's needed here is a custom kernel. The userland can be taken straight from the Debian repository.

So what's the point in having an "alternative" userland, that basically does the same, just worse, because it uses less of the CPU's potential, and is likely maintained worse?

 

[Letalis Sonus]

Many components would turn out completely useless because the RPi simply lacks the required interfaces. You can't even use the screen, the RPi's SPI interface is pretty much unusable with anything but their own display and I wouldn't be surprised if its two data lanes are simply not enough to drive the internal screen, not to mention the rotation issue returns.

There's no USB 3.0 at all, there's only a single SD controller, there's no eMMC interface, there's no SATA. There's still only a single USB 2.0 host port that you'd need to share with the PLS8, using an internal USB hub pretty much kills OTG as well. Wifi only works with the RPi 3's SoC (if it works at all), the others lack SDIO.

Then there's the amount of GPIOs - I don't know how much are needed for the Pyra board and how many either SoCs lose because you need the pins for other stuff, but the OMAP has almost 5 times as many potential GPIOs than the Broadcom SoC (at least the first one, the SoCs of the RPi 2 and 3 still lack proper public documentation!).

 

[Yasu]

Why?
The original point of Raspbian was to recompile Debian, because the ARMv6 CPU wasn't compatible with Debian/armel (which requires an ARMv7 CPU). That point became obsolete with the introduction of the RPi2 (that even supports armhf). All that's needed here is a custom kernel. The userland can be taken straight from the Debian repository.

Fair point.

 

[rSl]

Did you write this while knowing about the Raspberry Pi compute module ?
https://www.raspberrypi.org/blog/raspberry-pi-compute-module-new-product/

It is the same type of product as Pyra's CPU board. As far as I understand, thay only offer Raspberry Pi 1's SoC and level of performance in this form factor however.

problem is, it has a different connector and size to be a pluginable replacement and is a bit on the slow side too.
the rpi3 compute-module ible mentioned would be a lot beefier but also a lot hotter and has stille not the right size and connectors.
other con is, it might have only one gig of ram like the rpi3 and pyra will have 4gigs.
so it boils down to, can we get their soc to build a pyra cpu-board. and i guess after reading the story with hardkernel, the answer is no.

 

[RobertBil]

Hi..as per my knowledge the raspberry foundation looks to bring digital tools to a bigger audience.Aligning quite well with free software and security.It would be a good match, but they are not in the business of selling Pyras.I think quite a lot of features like touchscreen and lid logo and illumination of that and keyboard would need to be stripped before it would make sense, not only with the power-envelope of pi-like products, but also price overall.

 

[TrashyMG]

Hi..as per my knowledge the raspberry foundation looks to bring digital tools to a bigger audience.Aligning quite well with free software and security.It would be a good match, but they are not in the business of selling Pyras.I think quite a lot of features like touchscreen and lid logo and illumination of that and keyboard would need to be stripped before it would make sense, not only with the power-envelope of pi-like products, but also price overall.

If you figure development cost, parts and board population, the price difference between putting an OMAP5 and one of those Broadcom Pi SoCs would only be marginally different. Not to mention I don't really see the benefit, The end result is a very similar Linux armhf environment.

 

[directive0]

I think dragonbox should consider making a Pyra mainboard with the intention of just accepting a Raspberry Pi zero. Ostensibly this would make the Pyra body capable of using a raspberry pi. Sure it wouldn't be fast or anything, but I know tons of people who would just like a nice gaming enclosure for their Pi's.

RPI's are also at this weird price point where a lot of people think "Oh it's just 5 dollars, no big deal!" and then spend over a hundred on weird peripherals like touchscreen and CNC milled enclosures.

Although I'm not sure it would be anywhere near profitable for DB after you factor in new board manufacturing and having to write support software for all the GPIO and drivers.

 

[TrashyMG]

I think dragonbox should consider making a Pyra mainboard with the intention of just accepting a Raspberry Pi zero. Ostensibly this would make the Pyra body capable of using a raspberry pi. Sure it wouldn't be fast or anything, but I know tons of people who would just like a nice gaming enclosure for their Pi's.

RPI's are also at this weird price point where a lot of people think "Oh it's just 5 dollars, no big deal!" and then spend over a hundred on weird peripherals like touchscreen and CNC milled enclosures.

Although I'm not sure it would be anywhere near profitable for DB after you factor in new board manufacturing and having to write support software for all the GPIO and drivers.

If you're going through the effort of making an entirely new Mainboard that won't accept a Pyra designed module, you may as well throw a working SoC on the board directly, especially one that cost < 5 Dollars.

 

[pyrat]

I don't think having RPi internals to a Pyra makes a lot of sense (in a technical context) to those of us who are informed of the situation.

ACK

If you can get a Pyra shell with RPi internals up-and-running then it's free press.

I'm no electronic engineer by any stretch (don't get me started on why). But I don't have any hard feelings about them, so I wouldn't wish them such a terrible thing as designing a board just for PR value.
It must be a work hard enough to hope to entertain, while working on it, thoughts of somebody somewhere getting fun or profit or good uses from the result, not just paying for it and then discovering
it wasn't really useful, it just looked cool on the website.

People who know about any subject have a responsability to return what they have got from society and the hard work of all those before from what they have learnt.
And part of that return is not submitting to uninformed useless requests from the masses. If the masses want a useless board, let them design it themselves.

We know shitty markets and misery make people do useless jobs to avoid starvation, but can we at least not plan for it ?