> Our resources are limited. If we remain focused on "just" a GPU
> (as if that weren't a big enough project) we will see results
> sooner. A SoC/APU requires a CPU, MMU, FPU, ICU, SATA, Ethernet, etc.
> in addition to the GPU and it all needs to work correctly, and given
> mask costs it better work correctly the first time.

I'll agree that we should focus on *just* a GPU when someone can
provide some business logic on how we can ship a million GPUs
from that single first mask run.

My opinion is it's more likely that a full Libre-hardware SOC with
GPU, cpu, MMU, FPU, and InfiniBand will ship a million units. If the
SoC has InfiniBand, I can dream up an architecture where if you want
higher GPU performance, you just plug in more SoCs to your infiniband
switch backplane.

> Speaking of offtopic discussions, it feels like time for a reminder about
> the Open Hardware list:
> 
> The Open Hardware list is for general discussion of
> Free/Libre/Open Hardware, including many topics that
> are off-topic for the Open Graphics list.

The open-hardware list archives are members-only. I'll be
happy to take this discussion there if one of the following
conditions are met:

a) someone is paying me to keep discussions closed
or
b) the archives are open for public inspection.

I like referring to thing's I've posted in a public mailing
list archive 2-10 years later. (Speaking of which, I would 
like to mirror the list archive(s) on my server)
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)

Reply via email to