[FOSS-GPS] RTKLib MIPS requirements?

Michele Bavaro mic.bavaro at yahoo.co.uk
Tue Aug 14 13:41:05 PDT 2012


Hello Danny,

I agree completely with you: the trend is very clear.

Tshepang,
By saying that there is no commercial product development path beyond PI 
I meant that there is very little to reuse of the PI once you have 
RTKLIB running on it. It is surely a good learning exercise - as it was 
with Beagleboard at the time Mr. Takasu did it ...but it is not that 
useful afterwards. As Danny pointed out -would you want to make a real 
"market piece"- you would immediately need to move away from the 
Broadcom SoC and use another chip, perhaps another toolchain (Cortex-Ax 
are now everywhere), another BSP, and perhaps even operating system (I 
would love to see RTKLIB on Android).

New embedded platforms are coming out every day. Personally I love 
Olinuxino, Gumstix, Beaglebone, Colibri, FriendlyARM, AcmeSystems, etc.. 
and I see three potential reasons to port RTKLIB on an embedded platform:

- having fun whilst learning
Who could ever desire otherwise?? Can't really argue about this :)

- innovating
But.. running RTKLIB on PI is "easy"..I must say thanks to the really 
good quality of the code itself.
IMHO it wouldn't be a major breakthrough compared to Beagleboard (3 
years later).
We run it now on an ARM9 and it took us 3 days of work to go from a 
blank SD card and development environment to having rtkrcv/rnx2rtkp running.
When Danny said he was targeting a STM32F4 (168MHz, 128kBytes RAM) 
...well that would have been showing off!

- making a cheap, inexpensive product for new unexplored markets
Then again, I think there are better choices than PI to shorten the time 
from the lab to the shelves

Beyond the critical thinking above let me make clear that I fully 
respect RaspberryPI: it's a great SBC!

All the best,
Michele


On 14/08/2012 21:28, Danny Miller wrote:
> Raspberry Pi was developed by a nonprofit entity.  They don't make 
> money on them.  This already seemed to be a problem IMHO when the 
> release date got pushed and production went very slowly as they 
> released units basically one-by-one.  You don't usually see this sort 
> of thing in capitalist enterprises.  Apple might not have enough iPads 
> for "everyone" on their release date, but that's because they've 
> convinced 10% of the US population that they must have one on Day 1 of 
> the release and they've bought up all the free mfg in China to make them.
>
> While they made a lot of them, they're not guaranteed to make them 
> forever or update the tech (it's a maintenance issue).  The Raspberry 
> Pi Foundation is not only nonprofit, it's literally only 6 people.  
> When one or two moves on or gets hit by a bus, it's plausible the 
> project will die.
>
> It's also a problem that the Broadcom chip at the core of the Pi is 
> NOT for sale elsewhere.  It was a special, personal agreement between 
> Broadcom execs and RPF members.  There will be no competitors using 
> the Broadcom chip unless things change substantially.
>
> However, this isn't actually a problem.  The RPI is merely the 
> vanguard of a new tech of cheap, powerful, single-board Linux 
> computers.   The "Broadcom" chip is actually an industry-standard 
> ARM6K core, sold as design IP to many OEMs. Broadcom added the video 
> core and memory around the ARM instruction core and fabbed it.  Any 
> mfg with the ARM6K (or another ARM core up to the task) could do it 
> even without the video core, but might be interfacing through a 
> low-resolution LCD protocol hacked together, or through a terminal 
> port (issuing and accepting ASCII command lines through a data port 
> such as UART, USB, ethernet, etc).
>
> But that's beside the point, because the Raspberry Pi will soon be 
> matched by an equal or superior core running Linux.  There already ARE 
> ones specified.  BeagleBoard was an early one- and expensive, for what 
> it did, relatively speaking- but Moore's Law expansion applies to 
> Single Board Computers.  Next year they've got the open-source OUYA 
> gaming console planned for release at $99, which is a helluva LOT more 
> power than the RPI.
>
> IMHO we can expect to see commercially profitable Linux SPCs of 
> comparable core power to the RPI, with supporting Linux distros, at 
> Mouser, Digikey, etc within a couple of years.  Note the console 
> RTKLIB sources are not machine-specific.  If the ports are hooked up, 
> there's an FPU or enough core to implement FP calcs with regular 
> instructions, and enough RAM, it should run. Well you need a compiler 
> for that core- but right now all this stuff is one of the ARM cores 
> and we have GCC compilers for the ARM cores, and can expect a GCC for 
> any core to come out in the future.
>
> Danny
>
> On 8/14/2012 12:56 PM, Tshepang Lekhonkhobe wrote:
>> On 14/08/2012 11:31, Michele Bavaro wrote:
>>> Raspberry-PI is useless for me as there is no commercial product
>>> development path beyond it.
>>
>> Can you explain what this, 'commercial development path', means?
>> _______________________________________________
>> This message is sent to you from FOSS-GPS at lists.osgeo.org mailing list.
>> Visit http://lists.osgeo.org/mailman/listinfo/foss-gps to manage your 
>> subscription
>> For more information, check http://wiki.osgeo.org/wiki/FOSS-GPS
>>
>
>
>
> _______________________________________________
> This message is sent to you from FOSS-GPS at lists.osgeo.org mailing list.
> Visit http://lists.osgeo.org/mailman/listinfo/foss-gps to manage your subscription
> For more information, check http://wiki.osgeo.org/wiki/FOSS-GPS

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/foss-gps/attachments/20120814/f53ce389/attachment.html>


More information about the FOSS-GPS mailing list