It Looks Unlikely The Improv ARM Mer-Running Board Will Ship

[phoronix]

Phoronix: It Looks Unlikely The Improv ARM Mer-Running Board Will Ship

Back in the middle of April I wrote about the Improv ARM development board not yet shipping and many of the early pre-order customers being frustrated that the open-source-friendly hardware still isn't shipping months past its original ship date. Since then, there's still been no official update and it looks like one of their suppliers isn't even working with them anymore...

http://www.phoronix.com/vr.php?view=MTY4NDA

[shmerl]

Unfortunately all this is starting to look like vaporware. Vivaldi project especially. Falling behind schedule a few times can happen, but when it takes years and years - most users just lose hope.

[0xBADCODE]

Whatever! Chinese manufacturers already filled gap - there is ton of cheap boards at about $50, coming with at least Allwinner A20 (1.2GHz, 2 core, SATA + giga ethernet) and 1Gb RAM and ton of I/O expansion. This can be CPU module, dev-bord like thing or ready computer-like thing. I do think it rather pointless to compete with China in manufacturing cheap stuff. They overtook this market. And yeah, its pretty possible to boot usual Linux on such boards (Debian/xubuntu/etc are common).

[BSDude]

I do think it rather pointless to compete with China in manufacturing cheap stuff. They overtook this market.

It is a matter of time. Chinese workers are starting to complain and boycott the manufacturers to increase their salaries. The richer China gets the costlier it is to keep prices down. Short term they're the leaders, long term it's going to shift to India or the African continent.

[wizard69]

Is anybody really surprised?

I took one look at their web site a year or so ago and left in a hurry because it was obvious there was no business sense to be found. Beyond that the whole idea of putting your CPU on a plug in board is completely bogus.

[wizard69]

People at times get all worked up over trivial difference in costs.

Whatever! Chinese manufacturers already filled gap - there is ton of cheap boards at about $50, coming with at least Allwinner A20 (1.2GHz, 2 core, SATA + giga ethernet) and 1Gb RAM and ton of I/O expansion.

True there are interesting offerings out there. The problem is how long will the companies be around and how long will the actual designs be around. A cheap board that only exists for six months to a year is useless for the embedded market. Being cheap really isn't an advantage if you can't set life spans for the boards. The boards may be good as a toy, an educational platform or even a limited production application but would be useless for a product with an extended life. Sadly extended life these days seems to be anything beyond a year. I'd like to see at least one of these low cost manufactures demonstrate a five year plan to keep a board supported. Yes five years is a long time but the alternative is pretty ugly for many potential users.

This can be CPU module, dev-bord like thing or ready computer-like thing. I do think it rather pointless to compete with China in manufacturing cheap stuff. They overtook this market.

Not in all stuff. Many manufactures have returned to US production especially for non electronics. The problem with electronics manufacture in the US is one of legacy systems not competing with advanced Chinese factories. Well that an the extremely high overhead in the US from taxes, insurance and other evils instituted recently. There are a number of things, electronically speaking, made in the US that compete well with Chinese products.

And yeah, its pretty possible to boot usual Linux on such boards (Debian/xubuntu/etc are common).

This is possibly the most important aspect of these boards, you get a usable Linux platform dirt cheap. This provides for many possibilities use case wise. Further it makes the attempts to recycle old power hungry hardware look a bit foolish. This especially the case if you want to embed something and have it running 24/7, you can run some of these boards all year on the amount of power a traditional refactored PC will use in a week.

[0xBADCODE]

It is a matter of time. Chinese workers are starting to complain and boycott the manufacturers to increase their salaries.

Get the idea: deals are conducted here and now. Either I'm getting acceptable terms and complete deal, or I'm getting off. Should market conditions change, market would adapt. And there was unexpectedly large demand on these cheap ARM boards as they're great for low power home automations, all kinds of kiosks, various embedded computers/controllers and somesuch. So they even frequently happen to be "sold out", etc. To great surprise of everyone. Since there is demand, more and more fabs are getting idea and starting to manufacture these things. And have you ever heard about fully automated factories? Have you ever seen how pick and place machine puts SMD things on board much faster than any human can ever dream? Have you seem how the whole PCB being soldered in one shot in fully automated way? There is no real need for monotonous repetetive human labour. Over time, fully automated manufacturing will be cheaper. And then machines are okay with working 24/7, demand no salary and do not ask for holidays. For that reason they're likely to replace manual labour. Just as it already happened zillion times in industry. Machines are getting smarter. So my bet is: at the end of day there will be no monotonous manual labour at all in the world. Machines are better at that and they do not ask for salary at all. For this reason I think these with little brains will be seriously out of luck at this point. If you can be replaced with robot, I have really bad news for you: robots are coming and about to replace you.

True there are interesting offerings out there. The problem is how long will the companies be around and how long will the actual designs be around. A cheap board that only exists for six months to a year is useless for the embedded market.

Really depends. For relatively small-scale deployments its not a big deal. If you care, you can buy some extra stock to ensure availability/replaceability. And some "less critical" applications like home automation/kiosks/etc are both popular and not really hard to refit to "similar" board, should you really need that and board gone totally unobtainable. So it turned out not everyone needs super-available boards at super-high prices. Low-cost boards got unexpectedly broad markets. Orders of magnitude bigger than initial batches of boards expected after looking on embedded markets you refer to.

I'd like to see at least one of these low cost manufactures demonstrate a five year plan to keep a board supported.

I do not really care about that. If you really need it, go for Texas Instruments and somesuch. But they have crappy hardware at very bad prices. Also, TI OMAP is overengineered to the hell and very complicated IC while it price to performance ratio suxx really bad. So availability in future is virtually the only advantage they have. But only few application really neeed such availability. Far more applications are okay without it.

Not in all stuff. Many manufactures have returned to US production especially for non electronics.

As for me, USA goods are usually horribly overpriced. And for me price is a deal breaker, no matter what. I can find two zillion uses for $15 low power and smart modules. I can find some reasonable uses for $50 modules. At $75 it would get hard. At $100... just forget about it. Its just dev toy and much uses are just getting not feasible at this price level. Price matters, $15 module haves much more uses than $150 module.

you can run some of these boards all year on the amount of power a traditional refactored PC will use in a week.

Exactly. Not to mention that if some system is valuable (e.g. home automation controller) - small board is easy to provide with uninterruptable supply for a looooooong time, many hours to days of autonomous run time. Power hugry PC would exhaust even large and powerful UPS in mater of hours. And even notebook would exhaust it battery in active modes in few hours. Boards like this can last for literally several days using batteries equivalent to those used in notebooks.

[lkcl]

!cheap!

Whatever! Chinese manufacturers already filled gap - there is ton of cheap boards at about $50,

sorry i missed this when it first came out, i was very busy. the comment you make entirely misses the value of the EOMA68 concept. many engineers looking to leverage these cheap boards as a way to remain cost-competitive (i.e. instead of spending $20,000 in NREs then buying components that are only available in quantity 2500 to make a run of 500) have bought a stack of cheap boards at $50 and found them all to be GPL-violating or that they out-of-date within a year. or not contain sufficient documentation, either on the processor or the board.

by contrast the EOMA68 boards that i am designing - apart from being entirely designed to be manufactured in china - conform to a decade-long standard (so the base can be designed once and only once), and all CPU Cards will have full source code available.

I do think it rather pointless to compete with China in manufacturing cheap stuff. They overtook this market.

we're not (we're *using* them), and they haven't. a) they're cloning boards in the same factories that make shoes, handbags and spoons: the boss just decided one day to "diversify" or "get into tech". one guy i know, in between sending me offers of hardware i'm never going to buy occasionally offers me 50,000 condoms as well.

but more importantly, b) the biggest competitor for "cheap stuff" is... CHINA! many of the people who were sending me offers of hardware a year ago are no longer in business, because the profit margins are now so small that they simply cannot afford to continue to produce the "cheap stuff". and, as there isn't any HIGHER VALUE stuff, they're getting out.

And yeah, its pretty possible to boot usual Linux on such boards (Debian/xubuntu/etc are common).

actually, it isn't. a specialist reverse-engineering expert may be able to succeed in anywhere between three to twelve weeks but anyone else usually has to put up with the fact that the kernel is usually GPL-violating, based on android, DRM-locked to the processor, out-of-date (or going there fast), and any combination of those factors and people usually DO NOT do enough research. for example: mjg59 gave up maintaining the GPL-violating tablets list that he started, once it was clear just how endemic the situation really is (around 98%).

just today we received on the debian-arm mailing list yet another request for assistance in getting debian up and running on yet another ARM system: an ACER tablet this time i believe. there was a sting in the tail: it was pointed out that because the system runs a GPL-violating DRM-locked 2.6 kernel that they are running out of time for the system to be useful. ordinarily you would run a chroot environment under the GPL-violating kernel but because it has a 2.6 kernel (that the manufacturer is never going to update, or provide the source for) that AUTOMATICALLY prohibits you from using anything other than debian 6 (unless you are happy to run heavily-modified versions of debian, which is a huge amount of hassle well beyond the average person's resources).

and no, it's not acceptable to say "ok just get another board" because it took long enough to evaluate the existing ones, track down the source code and find creative ways to run what *YOU* want to run on them rather than what the GPL-violating factory *SUPPLIED* on them - so long in fact that by the time you chose one it became out-of-date!

does that help make it clear what the value proposition is of working with someone who is primarily a Software Libre Developer? i've taught myself hardware design precisely because i have seen so many people fail time after time to do exactly what you believe to be so very very easy. i've heard stories of at least 2 entrepreneurs who sadly commited hundreds of thousands of dollars to buy thousands of units on the PROMISE of receiving the source code under a GPL request, only to be told after the money is sent "i'm sorry: we lied - we don't actually have the source code because we make spoons and other widgets y'know, this is just like yet another line for us, we just bought the design and the binaries from some random chinese ODM house. now, can we get any more cash out of you, sucker?"

this is the irony and reality of working with what you describe as "cheap stuff".

so this is one of the many many reasons why i came up with the EOMA68 concept. if it was "just another cheap stuff knock-off" i would have quit a long time ago. there are plenty of people doing arduino-compatible or beagle-compatible or X-Y-Z-board.org clones but that is NOT what the EOMA68 concept is there to compete with. one idea is to take the schematics and proven successful X-Y-Z-board.org PCB files and turn them *into* EOMA68 CPU Cards, then turn them over to mass-volume china manufacturing in order to get the price down.

then when that SoC becomes out-of-date, that's fine, you can move to the next one *without* having to do a total system redesign.

there are many more other benefits, right the way down the chain. so this is why i am continuing with the project, and am committed to its success. it's a long-term project that has to be got right. once the first products are out the door, the standard CANNOT be changed. i would rather the standard be correct than make the mistake of it being wrong and the decade-long project failing because the standard could not keep up.

so it's not so much about selling a "single board" as it is about selling a long-term concept. which is why there is a sign-up page with crowdsupply which will be going live in a few weeks: https://www.crowdsupply.com/eoma68/micro-desktop