Den onsdagen den 1:e januari 2014 kl. 13:13:55 UTC+1 skrev Anguel: > > Happy New Year everybody! > > Well, reading all those comments here about non-profit and break-even > calculations my theory is the following: > > On the other hand it is very interesting for me to know how much this > "non-profit" scheme spends for advertising and marketing, I bet it is a > very different number in contrast to the money spent for actual Beaglebone > HW + SW development. Talking about all this "non-profit" stuff people > overlook the fact that the Beaglebone is actually a very effective large > marketing machine working for TI and its Linux processors. Just look at all > the articles about the Beaglebones, all the Google Adwords and other > advertising in magazines you see around, people wrinting in forums, > spending time to make YouTube videos just push TI popularity + sales for > free and they do it better than any internal marketing department. Also, I > still don't get it why TI should make less money when 100.000 BBBs are sold > than they do when selling large quantities of a Linux MCU to a big > company... >
TI probably have a significant number of projects which has 100ku volume or more, which does not require special support. If on average each customer buys 10 boards and files 1 request to TI support, then definitely it is more expensive than a single customer which buys 100,000 chips. If the chip is $5 and the profit margin is 50%, then they earn $25 per customer. If an engineer costing $100/hour is spending 15 minutes on answering that question, then 100% of the profit is used up. Simple math. > > Unfortunately, as we know, the marketing concept did not work as well as > expected: The only SW developer at Beagleboard.org decided to choose the > niche Angstrom Linux for some reason. For Angstrom there was no community, > no existing docs or books, no forums you could find any questions answered. > Angstrom is the OpenEmbedded/Yocto project repackaged. There are two mailing lists for Angstrom, but the main activity is the OpenEmbedded mailing lists. The core OpenEmbedded project is not concerned with the BSP for specific parts, but the semiconductor companies involved with embedded linux. (I.E: Intel, TI, Freescale and Atmel) all are making Yocto the recommended way to build a BSP. The toolchain used by Angstrom/OpenEmbedded is the Linaro toochain which is funded by most major mobile phone developers. I think you should check facts before rarting. > So developers not only had to know Linux at bare-metal level but they also > had to learn complex OpenEmbedded and spend too much time to get simple > things done in the Angstrom way, if they did not give up immediately. So > nobody really contributed anything to Angstrom Beaglebone development. > The way it works is that peiople contribute to the OpenEmbedded/Yocto project and then the people behine Angstrom (which incidently are the core people behind OpenEmbedded) make a snapshot which is available in a branch in the Angstrom Distribution. The default branch is Yocto 1.3 compliant, but there are Yocto 1,4 and 1.5 versions available as well. > Another problem was the fact, that TI did not even succeed to deliver a > proper kernel with working drivers to the community at that time. They also > sticked to their parallel Arago (EZSDK) project instead of working together > with the community and Angstrom (Angstrom is a real celebrity in contrast > to Arago). Additionally, at this time Linus Torvalds decided to force > developers to move to device tree. This was another very huge problem that > still has not been solved completely. > > Which is totally unrelated to the choice Angstrom and/or Debian. > Reagardless of the problems above, people continued to buy the BBB because > it is cheap and advertisements suggested it was "easy" to use. Development > boards are a nice business for any manufacturer because everything is > expected to be done by the developer (who is always the one who causes the > problem) and there are no warranties at all. > > > So it looks now that TI / Beagleboard / CircuitCo try to correct the > Angstrom concept by moving to Debian, probably again without investing in > development. If the community will suddenly become as engaded as with the > Raspberry Pi is a different question. But at least they would not have to > rely on one single Angstrom developer to solve all their problems. > > Moving to Debian will not solve any kernel problems. It certainly will not solve problems with capes or any other beaglebone specific stuff, without extra development. No sane company expecting to deliver significant volyme would choose a development board based on a part which cannot be purchased, so I expect that they will continue to sell a lot. Ulf > Anguel > > -- For more options, visit http://beagleboard.org/discuss --- You received this message because you are subscribed to the Google Groups "BeagleBoard" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/groups/opt_out.
