ARM as a primary architecture

Przemek Klosowski przemek.klosowski at nist.gov
Fri Mar 23 14:33:44 UTC 2012


On 03/22/2012 10:42 PM, Kevin Kofler wrote:
> Przemek Klosowski wrote:
>> Fair enough, but the trends are well established, and the data are
>>  for shipments so the actual deployed numbers are compounded (flat
>>  shipments translate to steady growth, linear or faster shipment
>> growth means quadratic or maybe even exponential growth).
>
> Quite the opposite, shipments are related to the first (discrete)
> derivative of active deployments, which will necessarily amplify the
>  effects: if the number of shipments decreases, that just means that
>  growth in deployments is slowing down, not that the number of
> deployments decreases.

I think we're saying the same thing here; the point is that compounding
amplifies even small trends.

> A computer is often a per-household device. A cellphone is per
> person, or even multiple devices per person (e.g. multiple phones
> with different contracts, or a phone and a tablet, etc.). So the
> total number is necessarily going to be larger without there being
> more users.

Of course, but numbers are numbers; we just have to pay attention to 
lighter, mobile formfactors. It is the same reasoning as in the desktop 
vs. server dilemma 15 years ago: you could say that 'servers are 
per-company devices', so Linux has been concentrating on the desktops, 
where the numbers were---while still paying attention to servers. As we 
know it's a good idea to optimize for both desktop and server; they 
aren't contradictory.

As I wrote responding to Nicolas, the GUI interaction concepts from
mobiles are relevant to all platforms, and it is appropriate to think 
ahead and seek a model that serves both. Microsoft also seems to follow 
this direction, and implements it, however awkwardly, in Windows 8.


More information about the devel mailing list