[Arm-netbook] EOMA server standard

[Baybal Ni]

What do you think about sticking with "twisted pair on PCB" then? It's
the same conventional twisted pair ethernet, except for no transformer
couplers and it uses tracks on PCB instead of wires. This way we are
getting free from "different phy" problem. Integrators will then only
have to choose between 1000T(8 tracks) and 100X(4 tracks)

On 25 October 2012 15:59, luke.leighton <luke.leighton at gmail.com> wrote:
> On Thu, Oct 25, 2012 at 10:59 PM, Baybal Ni <nikulinpi at gmail.com> wrote:
>> Answering my own question, I just thought that PHY's are integrated
>> too deep within the networking architecture.
>
>  effectively, yes.  whatever is chosen as the "standard", that then
> *has* to be what is either supported by the SoC, or it has to be
> converted to irrespective of what the SoC might happen to have.
>
>  it's the same thing with LVDS, if you remember (or MIPI - let's use
> MIPI because it's more modern).  let's say that EOMA-68 had MIPI
> multi-lane taking up only say... 15 pins (whatever it is) instead of
> 24-pin RGB/TTL taking up 28.  MIPI would be great, right!  more pins!
> free up space, make some more functions yaaay!
>
>  ... except, now if you want a low-cost LCD with a low-cost SoC,
> here's what you have to do as the BOM:
>
>  * chassis: low-cost LCD ($10 say)
>  * chassis: not-really-very-cheap RGB/TTL-to-MIPI converter IC ($5 but
> however much it is you don't care how much: it's an overhead, plain
> and simple)
>  * botched-EOMA-68-standard-CPU-CARD: not-really-very-cheap
> MIPI-to-RGB/TTL converter IC (another $5 say)
>
> you get the picture?  whatever it is, because you picked a
> non-lowest-common-denominator standard (MIPI is by percentage-wise of
> availability across SoCs *so* not common) you then forced *everyone*
> to add a massive and critically *uncompetitive* overhead to *all* such
> "low-cost" products.
>
>  thus destroying any chance that the standard might have.
>
> i'm keenly aware that picking the lowest-common-denominator also
> eliminates some of the possibilities for using built-in LVDS, or
> built-in audio, or built-in.... whatever-it-is, but... tough.
>
> at some point down the line, it will be possible to get an Integrated
> EOMA-68 I/O chip done, which has all the features built-in that are
> required of e.g. a laptop, or a tablet, or a TV, or a camera and so
> on.  but, not right now.  no, actually cancel that: i'm not going to
> say "no" :)
>
> anyway, that's quite a bit of explanation of why it's hard to pick one
> particular PHY interface :)  i still don't exactly know what the
> answer is, though.  i'm 50% ready to give up on "standardising" for
> servers, and 50% still thinking "this really really should be
> possible, y'ken".
>
> l.
>
> _______________________________________________
> arm-netbook mailing list arm-netbook at lists.phcomp.co.uk
> http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
> Send large attachments to arm-netbook at files.phcomp.co.uk