<rant>shirro wrote:fully open drivers.
Mostly its the case of management not understanding what should and shouldn't be Intellectual Property. I run into this a lot at work and when management feels that all code I write is owned by the company and the lawyers would rather not deal with Open Source Licensing its going to be an uphill battle.chris_c wrote: I've never understood this rather self defeating meme about proprietary drivers - the excuse seems to be that the techniques are so obscure and so unusual (yeah really) that the need to be guarded, aside from the fact that a driver even open sourced isn't really going to tell massive amounts about the internal implementation
I hear you, but as it stands it looks like a positive encouragement to reverse engineer, if there were x egl and driver I would just shrug and go ah what the helljecxjo wrote: Mostly its the case of management not understanding what should and shouldn't be Intellectual Property.
Sometimes just opening the door for reverse engineering is enough to "give away" trade secretes
My RPI 2 specs would include an FPGA!
Everyone needs to move on from this, it will be a long time coming. The coders would love to release the source, but for business people, it would be considered madness. I've worked with many SoCs and seen the driver source. Apart from fixing the odd bug it has never led me to change my code. It will not help to make your game faster, only the driver writers can help to do that.shirro wrote:fully open drivers.
We are talking about the next RaspPi, not this one. I assumed the next RaspPi would not be released for a year or three, hence the x4 CPU and RAM. USB3 on the board does not mandate USB3 peripherals, similarly gigabit ethernet. However the pair of them allow a decent NAS or cluster, which the release 1 RaspPi is rather poor at.Lob0426 wrote:What chris_c is asking for is beyond the Panda Board ES and will cost more. Keep the features reasonable and you can keep the price reasonable. An open source GPU would be an advantage for many. USB 3.0 is powerfull, but would triple the price of the board and the higher performance peripherals.
I have to wonder if its possible to implement a high speed parallel bus with the gpio pins, with a kernal driver it could possibly be fast enoughrurwin wrote:... a decent NAS or cluster, which the release 1 RaspPi is rather poor at.
Valid point, but it is never going to happen, move along, nothing to see here.chris_c wrote: open source drivers have nothing to do with wanting to make faster games! you totally miss the point... its about the freedom to add features / abilities to the drivers something that is completely impossible at the moment.
Ah I didn't realise you were the Broadcom CEO, naturally as you say so it must be right!RichardUK wrote:valid point, but it is never going to happen, move along, nothing to see here.
That Xilinx Zynq looks pretty good! But no GPU and a board around it will never be as cheap as the raspi. One good thing Xilinx did is to add support for Zynq to QEMU. Makes it much easier to develop code for it. Xilinx should also sponsor a beaglebone like project around it. The zedboard is a bit too expensive.jecxjo wrote:I honestly can't think of anything I would add to the RPI. I'm running it as a desktop replacement, but I spend most of my time in cli, hardly ever even in an X Windows session. Oh wait I take that back. My RPI 2 specs would include an FPGA!
I think you could get 8 gpio, 2x serial, 2x i2c, 2x spi, 3.3v & GND onto the same 26 pin connector as on the current Pi, so it's not that hard.chris_c wrote:I can see good uses for all those interfaces, but how you'd get it all physically onto a small board I'm not sure!
I very specifically didn't ask for USB, take that connector off and we have more room for other stuffchris_c wrote:I'd even forgo ethernet if there were 2 seperate (working ) USB 2.0 channels, you could hang wifi off one and have your low speed peripherals on another.
ahci = sata and there are SoC's that have itchris_c wrote:a small footprint connector for storage would be sata but I'm not sure there are any SoC's with sata (much less separate usb channels mind you!)
No thanks, don't want anything that multiplexes several capabilities onto the same pins. One common problem I've had with microcontrollers is that you have to pick either hardware i2c or hardware spi and bit-bang the other one.chris_c wrote:GPIO's that are interrupt capable would be very useful - there's no reason they couldn't double as i2c etc,
Nah, it's fine. I have no interest in putting XBMC on it. Besides, I remember the times when we didn't know how to fill 640kchris_c wrote:512mb is ok but still a little tight....
re: USB and audio driver source is available. Fill 'yer boots.chris_c wrote:oh my, those problems sound terrible. I never realised things were is such a state!
audio drivers and usb drivers would be the type of thing that could have been solved fairly quickly by the community - I wonder how long we'll have to wait.
well if they can't fix it - is it broken at a hardware level?jamesh wrote:....and neither have the manufacturer, who have all the chip docs etc.
exactly the total number with Pi's let alone kernel hackers with Pi's is tinyMaybe, just maybe, it's a bit more difficult than you seem to think...
Problem is that are very few people with the required skill set, which makes the community rather small.