EOMA-68 Based KDE Vivaldi Tablet Engineering Boards Ship 33
sfcrazy writes "Aaron Seigo, a lead KDE developer, says that the ambitious KDE tablet Vivaldi is shipping to the team for quality testing. Seigo writes on his Google+ page, 'A great start to the week with a warm, sunny, quiet Monday. Well, almost quiet. The first Vivaldi tablets, new dual-core engineering boards and the custom EOMA68 developer workbenches we commissioned have all been shipped out. Don't get too excited: the tablets are pre-certification (EC/FCC) and are on their way to us so we can verify the Q/A targets we set out. Still ...'"
It looks like long-time reader lkcl's EOMA-68 initiative is working out; in related news the first batch of Allwinner A10 EOMA-68 cards is shipping to the "...20 Free Software developers brave enough to take one of these at this very early phase." Update: 07/23 17:16 GMT by U L : Correction from lkcl: the first batch of EOMA-68 cards are actually using the Allwinner A20, a bit of an upgrade from the original design.
Whatever you want it to be (Score:3, Informative)
Re: Wildly confusing subject line (Score:3, Informative)
Yes. It does make sense.
Tablets are great, right? You figure out what size you want, what extra features (dockable keyboard ala Asus Transformer, Wacom stylus, etc.), and you're set for years -- we've finally got good enough resolutions (1920x1200, 2048x1536, and higher) that future screen kicks aren't real compelling upgrade motivators. The only reason you'll upgrade is more CPU/GPU/RAM muscle, so it makes sense to let you buy a new CPU card when the next gen SoC comes out, and upgrade your existing tablet easily. So there's one use case that makes this a good thing for a consumer.
Now that was probably obvious -- the slightly less obvious thing is, it ALSO makes sense for the producers -- because while you're just buying a new CPU card instead of a new tablet, this leaves you with a spare old-but-usable CPU card. So you buy a lame tablet for your kid, and slot the old CPU card in there. Or you buy a dedicated ebook-reader, and slot it there. Or whatever; the point is, it doesn't hurt the manufacturers as much as you might think.
Moreover, it massively helps them, because instead of completely redesigning their tablet product line every 6-12 months with the arrival of a new SoC, they keep shipping the exact same hardware, except for the CPU card (ok, and maybe change the model number on the box) -- so they just order a batch of new CPU cards with the new SoC, and have an instant spec upgrade.
Or say you want to sell a range of tablets (say two price/performance points in each of ARM and x86), no problem -- instead of 4 SKUs meaning at least 2, probably 4 different boards, with 4 different BOMs (2 boards assumes you have two pin-compatible SOCs of each arch, as well as pin-compatible EMMCs and such), and redesigning the whole mess once a year, they get 4 SKUs from ONE board, with EVERY soldered-on component identical, and one pluggable element for all the differences -- and you still only need to redesign for changes of screen and integrated peripherals (e.g. WLAN or WWAN).
It really makes an amazing amount of sense, once you think it through -- you just have to get a transition through the chicken-or-egg problem from here to there somehow.
Re:Looks interesting (Score:4, Informative)
I would certainly like a tablet with full Linux support, especially with KDE Plasma. Never heard of the "EOMA-68" standard before, but it looks intriguing . Not sure why they specced 10Mbit ethernet support as mandatory minimum for the CPU card. 10Mbit networks must be very rare these days, and the cheap misers who still operates them are unlikely to purchase a tablet. Am I missing something?
Well, the big philosophical idea is that ANY EOMA-68 CPU card slots in ANY EOMA-68 machine (note that EOMA is not entirely, or even primarily about tablets -- that's just the first hardware product using it), and works. That's why Luke (aka lkcl) is quite adamant there are no "optional" features in the spec -- the only exception is for interfaces (e.g. USB, 10/100/1000-BASE-T) that can fully autonegotiate in both directions, so that there's neither a slow-machine/fast-cpu-card, nor slow-cpu-card/fast-machine case where it becomes incompatible. So you can have 10-only (not likely, but hey, we'll allow it), 10/100 (likely on low-spec SoCs, or when an SoC has no ethernet, so we fulfill the ethernet requirement with a USB-NIC), or 10/100/1000 (high-spec SoCs).
As for specific machines that would be apt to need 10Mbit, and fail with a 100/1000 only CPU card -- think of a router for a domestic DSL connection (or a "plug-computer" like pogoplug serving this role. Most DSL modems over a certain age are strictly 10Mbit.... if the standard doesn't require every CPU cards to be able to negotiate down, you'll either need to:
[A] supply your own 10Mbit NIC (hanging off the USB interface like the Raspberry Pi)
[B] supply your own 10/100 autonegotiating switch
(either of those drives up BOM substantially) or
[C] break compatibility by making it work with only those EOMA-68 CPU cards that happen to support 10Mbit
(which inevitably frustrates users who try to swap a seemingly-compatible CPU card in, and find it mysteriously stops working...)
None of those is really winning, agreed? So we either split the standard for plug-computers and everything else (meaning when you upgrade your netbook or tablet, you can no longer slot the old one in a cheap plug-computer to run as e.g. "home server and router"), or we require 10 Mbit as the minimum -- which isn't really hard to meet, anyway...
Re:Umm... WHAT is this? (Score:3, Informative)
Yes.
It's the main guts of a computer, stuck in something the size of a PCMCIA card, and you can stick that in whatever hardware project you want.
Prototyping board is one (probably the most obvious) potential use.
But that's partly because it's useable for so many uses.
correction: it's an A20 CPU Card (Score:4, Informative)
the A10 is out-of-date so we're using the pin-compatible A20 instead. dual core ARM Cortex A7.