New Beagleboard similar specs as Pyra??

[ible]

http://www.elinux.org/Beagleboard:BeagleBoard-X15
https://boards.openpandora.org/pyramain/pyraspecs.html/Are these pretty much the same board?  maybe the communities can be mutually beneficial for getting things like the 3D stuff going, and we can figure out how to use some of the other random stuff (like the M4's).

 

[EvilDragon]

It uses the current AM-processor from TI, which is very similar. It has the same 3D GPU, so yes, that could help.

 

[fantomid]

The first article about BeagleBoard-X15 came in November 2014 on linuxgizmos.
Release date may be this year. Didn't find any information relative to this date.

 

[WizardStan]

I believe they just started shipping release builds, the FINAL final prototype before going mass manufacturing.  They aren't that far ahead of the Pyra, really.

 

[levi]

Might be interesting as a dev board though, surely?

 

[Wally]

This has to be good news for both the Pyra and Beagleboard meaning we can have packages shared across both devices somewhat?

 

[ible]

certainly more people buying the chip may help TI get someone to help with the 3D driver...

 

[rygD]

Didn't ED say he had someone that could work on the 3D (I may be getting mixed up)?  Maybe DragonBox and the Beagleboard people can split the cost of getting someone to do it.

 

[Grench]

I don't understand why getting a Linux 3D driver from TI would be an issue for the either or both projects.

http://www.ti.com/product/am5728
TI lists the AM5728 as having 3d graphics.
TI lists the AM5728 as having Linux as a supported OS.
If the 3D capabilities are not 'open' and can only be obtained through a proprietary binary - then they need to provide that binary or they don't actually have Linux support for the board features.

As far as I can tell, TI claims to support Linux on the OMAP5432 as well.
http://www.ti.com/tool/OMAP5432-EVM-EVAL-SW

I'm not sure how TI can advertise that the SoCs both have 3D graphics and work with Linux and not provide driver support to make it functional.  Frankly, TI should either provide a proprietary driver and ongoing support throughout the life cycle of products based on their chip OR open the documentation on the 3D portion of the SoC so that others can do it.

 

[EvilDragon]

They provide the driver. It works on their outdated Ubuntu 12.04 setup with an ancient kernel and X.org.

But we want to make it work with current and future kernel and X.org versions, that's why we need some magical hackings.

It's the same on the Smartphones: My Droid4 runs with the latest Lollipop but still uses an outdated kernel - as no compiled 3D drivers for more recent versions exist.

 

[WizardStan]

As with all other binary drivers, there's an open source interface into that binary which needs to be compiled for the kernel.  The one that TI provides is not compatible with the current kernel. It "only" needs to be updated, which is not an easy task.
It can be done but it'll take time and a lot of effort.  I haven't looked at what is involved so I can't comment on exactly how much effort but the fact that it hasn't been done I'm going to go with "a lot".

 

[fantomid]

Renesas announced the availability of a SoC with specs closed to the Pyra's one.
And the SGX seems to be the same.
Another chance to get up to date binary drivers. I hope so.

 

[fantomid]

Phytec too !!
2016 will be the year of the TI Cortex-A15 SoC and the year of the Pyra

 

[levi]

Also, the Elesar Titanium, the specs of which look suspiciously Pyra-like to my eye, though I've not compared part numbers (also, despite being announced in October, and demonstrated since, it's still not available to the public).

 

[TybiBeaton]

Can you please share the all specifications of this board here?


I want to work with the Beaglebone, but don't have any idea about the Beaglebone and the Pandora both?


How i can start with them and what are the other components required for them in their hardware setup?


surface mount assembly

 

[Exophase]

Also, the Elesar Titanium, the specs of which look suspiciously Pyra-like to my eye, though I've not compared part numbers (also, despite being announced in October, and demonstrated since, it's still not available to the public).

The AM5728 looks a little different compared to the OMAP5432 in Pyra. It has half the CPU L2 cache but 2.5MB of added scratchpad (not cache, and the ECC probably counts as part of that, hence the weird amount).


The DSP capabilities are also significantly expanded. There are now two faster DSPs that support floating point.

 

[notaz]

From the product brief, adding all the power supply requirements it's 31W for the base board and 127W for the peripherals (whatever those might be).


It's quite strange you can order that board with RISC OS preloaded.

 

[levi]

It's not clear whether they include the power requirements of connected USB devices and other board powered devices in the board power or the peripheral power requirements.  If the former, and all 8 USB ports deliver 500ma at 5V, that's 20W there alone.  I'd guess peripherals in that case are the four SATA discs you can hook up should you want.


It's not that odd that it is offered with RISC OS preinstalled.  The guy that designed the board designed previous RISC OS dedicated computers, and I found out about it through various RISC OS boards.  I'd guess what's holding it back at the moment is RISC OS drivers from what I've read, and they're not that interested in offering the Linux configuration ahead of the RISC OS only one.


But for me it's far more interesting as a Linux computer.  RISC OS can't use one of the cores, and can only drive one display at 1080p.  I'm honestly not sure how big the market for a RISC OS board is these days, but people keep releasing new machines every couple of years (thus far mostly using boards like the Beaglebone and Pandaboard).

 

[hlidegp2x]

I don't understand why getting a Linux 3D driver from TI would be an issue for the either or both projects.

The issue is that those drivers are tied to a specific old kernel version making their use for more recent kernel versions pretty hard to run unlike Nvidia tegra drivers which works for ALL kernels. The issue appears the same with other GPU providers.

 

[slaeshjag]

In our case, the kernel part isn't a big issue. It's open source and can be fixed by anyone. The main problem is the userspace libraries that are linked to ancient versions of mesa, and any attempt to use them on a modern distro will end with crashes and sad faces all around. The entire userspace part is closed source blobs, and will require either TI's cooperation or reverse-engineering to fix.