The future of PRU in TI's SoC evolution

John Beetem wrote:

 

Will it be possible to program the GPU, or is this the usual "look but don't touch" GPU nonsense?

It's ARM Holdings' single biggest claim to dumbness that they made their MALI range of GPUs proprietary instead of openly programmable like their ARM CPU cores.  They had a fantastic business model going that created great synergy in the ARM community, and then they turned their brains off and followed the proprietary path that splits their community instead.  (Sure, there's a reason it's closed as they bought in a third party design, but they could have designed their own GPU instead or else negotiated a deal allowing MALI to be open.)

The only good light on the horizon is that the open source Limadriver is reputed (by Olimex) to work well.  The website news hasn't been updated since last March though and their Gitorious repo not since October, so there is room for worry as well.  (Plus the fact that only MALI-2* and MALI-4* seem to be covered for now.)

Morgaine

Morgaine Dinova wrote:

 

It's ARM Holdings' single biggest claim to dumbness that they made their MALI range of GPUs proprietary instead of openly programmable like their ARM CPU cores.  They had a fantastic business model going that created great synergy in the ARM community, and then they turned their brains off and followed the proprietary path that splits their community instead.  (Sure, there's a reason it's closed as they bought in a third party design, but they could have designed their own GPU instead or else negotiated a deal allowing MALI to be open.)

I suspect that the primary reason Mali is closed (and perhaps other GPUs as well) is that ARM (etc.) fear that if the details were made public they'd have to fight off myriad [*] frivolous patent lawsuits.  By keeping the details secret is makes it harder for aggressors to troll [*] for alleged violations.  OTOH, ARM doesn't seem to be having a cow about Lima.  Maybe they figure that if some grubby FLOSSers [**] open up Mali then ARM can simply say "oh, that's not really the way we do it so don't bother trying to sue".

All this, of course, slows down Progress of Science and useful Arts, but then what are patents for?  [***]

JMO/IANAL

[*] all puns are intentional.

[**] humorous reference to how FLOSS advocates are often regarded.

[***] sarcasm is intentional too.

John Beetem wrote:

 

Oh well, nothing to see here.  I sure hope Parallella is successful!

I'd have used a wink-smilie in self-defence there!

• Firstly, there is no GPU at all on Parallella (so it's easy for it not to be closed),
• and secondly, the Zynq contains all that oh-so-sekrit FPGA programming stuff.

So while I very much look forward to programming the fully open Epiphany device, Parallella isn't the ideal ripost to complaints about ARM's closed MALI GPU.

(PS. In the absence of a GPU, people have been seeking alternatives like sending graphics off-board to the Pi's wholly closed OpenGL ES 2.0 VideoCore renderer.  I've tested shodruk's demo code and it works nicely.).

Morgaine.

Luc Verhaegen is quite active in the sunxi mailing lists and seemingly uses Olimex boards for development work on Limadriver. Yes, slightly worrying that there appears to be no public progress recently.

There are however various bits and pieces that help but don't quite seem to have come together into a set of stable drivers just yet. So you do need to be prepared to build a lot of the parts yourself at the moment.

Morgaine Dinova wrote:

 

John Beetem wrote:

 

Oh well, nothing to see here.  I sure hope Parallella is successful!

I'd have used a wink-smilie in self-defence there!

 

• Firstly, there is no GPU at all on Parallella (so it's easy for it not to be closed),
• and secondly, the Zynq contains all that oh-so-sekrit FPGA programming stuff.

 

So while I very much look forward to programming the fully open Epiphany device, Parallella isn't the ideal ripost to complaints about ARM's closed MALI GPU.

I was thinking of Parallella's Epiphany when I made the comment.  Also, I'm not concerned if Epiphany turns out to be a poor GPU.  What's important here is that Epiphany provides a way to do closely-coupled general-purpose parallel processing on an open architecture.  People have used GPUs as general-purpose parallel machines, but I don't know how far they've gotten with closed architectures.  If Parallella snatches this work away from GPUs by being more open, it encourages GPU-makers to reconsider their situation.

I came across something interesting recently that's rather relevant to us discussing the merits of using some Cortex-M* cores in place of the PRUs.

It seems that the am33xx series already has a Cortex-M3 in it. I'm not sure it's terribly useful, it's seemingly used as some part of the power management capabilities and has some sort of hardware mailbox interface to the main cpu.

Documentation and firmware for it is here http://arago-project.org/git/projects/?p=am33x-cm3.git;a=tree

selsinork wrote:

 

It seems that the am33xx series already has a Cortex-M3 in it. I'm not sure it's terribly useful, it's seemingly used as some part of the power management capabilities and has some sort of hardware mailbox interface to the main cpu.

I guess the appropriate hipster comment would be "Hack all the cores!"  Yep, even that one.

As an aside, back when we discovered that OMAP5 has replaced the two PRUs of AM335x with two Cortex-M4, it did occur to me that if they'd dedicated the same number of gates to adding more PRUs instead, OMAP5 would have its own integrated array-of-cores, a bit like Parallella.

Morgaine.

I think it's fairly clear that TI have little interest in the PRU, so even if they had done that without some commitment to active support it I think it would have been pointless.

What's becoming painfully clear to me is that the boards/SoCs with an active developer community, not 'user' community, are making much more progress. With little in the way of developer community behind the beaglebone and no interest from TI, what future for the PRU ?

selsinork wrote:

 

I think it's fairly clear that TI have little interest in the PRU, so even if they had done that without some commitment to active support it I think it would have been pointless.

What's becoming painfully clear to me is that the boards/SoCs with an active developer community, not 'user' community, are making much more progress. With little in the way of developer community behind the beaglebone and no interest from TI, what future for the PRU ?

I haven't looked into PRU in great detail, but the first impression it made on me is that it's a user/developer-programmable version of the internal RISC processor in Freescale's QUICC communications SoCs.  I've had a lot of experience with various generations of QUICC, and they do support many communications protocols like HDLC, ATM, and async HDLC.  However, the QUICC internal RISC processor is a closed architecture: you can download binary blobs, but you can't write your own.

So I see PRU having the advantage of programmability.  That said, I don't personally have an application for it at this time, and I don't know whether TI is taking any embedded communications sockets away from Freescale.  It may depend on whether EtherCAT takes off and who supports it -- I haven't looked into that either.

If the PRU subsystem is available in the 10-year long-term support AM335x parts, then it doesn't really matter at all whether TI "has interest" in supporting the PRUs --- it's still completely in our interest  to give it FOSS support.  10 years is long enough to underpin a whole generation of open projects, and the latest shiny is of little relevance.

And all these BBBs aren't going to vanish once an OMAP5 board comes out.  Their PRUs certainly merit our support, for decades ahead.

Morgaine.