[Arm-netbook] xilinx zynq-7000 800mhz ARM Dual-Core Cortex A9

[Gordan Bobic]

On 08/20/2011 01:29 PM, Alain Williams wrote:

>> phone/slate gaming, but that's a very limited use case. For server use
>> we don't need a GPU at all (serial consile a-la SheevaPlug is
>> sufficient), and for 99% of desktop use, you only need simple frame
>> buffer, possibly with very basic acceleration just to get the Window
>> manager performing well enough for LibreOffice. That covers all normal
>> business use.
>
> The trouble is who is the market. For servers - not a problem.
> For PC use: myself I have no interest in games, so what you propose
> (+ maybe videos) I am happy.

I do have an interest in gaming, but I am fully aware that I'm not going 
to be playing Crysis with all the settings turned up to 11 on a 
3840x2400 monitor with a computer that has a power budget of 3W. The 
power envelope gap is a factor of about 300x.

> However: that is not how things are sold,
> the marketing is all about features list - even if people don't use
> most of them, even if you try to position it as a lower cost/better battery live/
> whatever option.

I largely disagree - most people will buy whatever is cheap.


Either way, the key point here is that we are not looking for a chip for 
a fancy slate or phone, where the odd person might actually play a 3D 
capable game. We are looking for a chip for a business 
server/desktop/laptop.

Our primary (only?) target is Linux (normal Linux, not Android, not 
MeeGo). Last I checked the number of games that will actually run on 
Linux is at most in low double figures (ID games, earlier UTs, and the 
handful of games that the long gone Loki Software ported). Of those the 
number of the ones that source is available for is in low double figures 
(Quake 1/2/3 is the only thing that comes to mind), so we have a grand 
total of 1 game that would actually benefit from having 3D graphics on 
an ARM in our only use-case. I really don't think it's a big deal.

> People expect to pay the price of a mini but get a Rolls Royce.

We are talking about a completely different market here. No low power, 
ultra-portable, 8+ hour battery life laptop is a gaming rig, and is 
unlikely to be for years, maybe even decades.

If we were to look at opengraphics, cut out all 3D capability and limit 
it to 2D video acceleration (Xvid), then that might be the most ideal 
solution if it can be squeezed into a power envelope of about 1W or 2W 
at the most.

But again, even if this was plausible (which remains to be assessed by 
someone with silicon design skill), there's still the question of 
getting the silicon - which is likely to be so expensive to tool up for 
that it won't be plausibly doable on the basis of non-profit 
organization sponsorship.

Gordan