[Arm-netbook] [fedora-arm] 1ghz ARM Laptop (12in 1280x800 LCD)

[Luke Kenneth Casson Leighton]

[dropping fedora list.  gordan, we have to wait for the membership to
be approved on arm-netbook, by alain - apologies!]

On Tue, Feb 1, 2011 at 10:58 PM, Gordan Bobic <gordan at bobich.net> wrote:

>>  tell me about it.  it's why i will be speaking to ziilabs to obtain
>> details of the Cell's assembly language etc.  i have to come up with a
>> convincing and reassuring strategy as to why they should do exactly
>> that.
>
> You should also consider that getting FSF drivers for this is of very
> limited usefulness. This DSP is very proprietary. It would be far more
> valuable (if more difficult) to get the PowerVR drivers as that would cover
> the majority of ARM based SoC modules.

 and a _ton_ of netbooks.  i was really shocked at how many machines
are on the List_of_PowerVR_Products wikipedia page, due to intel's
licensing of SGX for the GMA500 chipset.  esp. given that it's
supposed to be just an embedded 3D engine.

> I don't think it'd be worth the
> effort, especially for a CPU (A8) that is already on the verge of
> deprecation.

 that'd be a _great_ way to convince ziilabs that they should in fact
do exactly what i'm asking for :)  "your CPU is on the verge of
deprecation, so why fiight the inevitabllllle..." :)

 it's about cost, gordon.  the newer the CPU, the higher the
development costs: less people have worked with it, they go slower,
but still make more mistakes, have to read more 200-page PDFs - and
time is money.  i'm seeing quotes of 1500 hours of work, costing $100k
to $150k from people (USA-based or EU-based) if you want "the latest"
Cortex A9 CPUs such as the spea1310, ericsson u8500, OMAP4440 and so
on.  and those are all 0.5mm pitch BGA, 11mm sq because they go into
"smartphones" where space is an absolute premium, and cost is not a
real significant issue (mobile phone vendors actually pay money - a
significant percentage of its value - per device to the manufacturers
if they give them exclusive access to the product!)

 whereas a 0.8mm pitch BGA part that's a little bit bigger, you can
use standard solder reflow processes, so pretty much any chinese OEM
house can knock something up, and many of the smaller factories just
experiment repeatedly with layouts until they get it right!  crazy
sods... :)

 i don't want to discourage you, especially as the pandaboard is
definitely the way to go, but to make this a viable (and low-cost yet
still reasonably high-end) product, an older, more mature and more
well-understood 1ghz+ Cortex A8 is the sensible safe choice right now
(and next year, probably it'll be Cortex A9s that are in the same
position)

 so once a useful and useable laptop exists and actually sells,
because it's 1ghz (or maybe 1.2 or 1.5ghz if not insanely costly like
the TI AM3894 is), assume that takes 6-8 months or so: by then, some
of the A9s will have been out (and maybe even some 750mhz or 1ghz
Dual-Core MIPS in 45nm) for long enough so that they can be
considered.

 and, maybe even some profits made so that they can be afforded!

 i guess that summarises as walk first, run later.  heck, reading
between the lines i believe that's what genesi are doing.  the ekiga
mx, based on the 800mhz part, sounds like it was the "trial run".  i'm
reaally interested to hear what the specs are on the new system being
developed, bill :)

 l.