On Fri, Dec 07, 2012 at 11:40:39PM -0600, Gregory Carter wrote: > Quite honestly if I started writing Verilog code I would kill the > project. The GPU would be a laughing stock, and probably work just > as well as AMD's or Nvidia's. > > You don't want that do you? :-) > > Now, once the card is done, and you want drivers for X or DRI kernel > work done that is a different story. I have 15 years dealing with > LINUX kernel issues, and making it work for business and industry. > > Just to be clear, I sell and build engineering workstations for AT&T > contractors. I almost lost the business trying to do it with binary > blobs from AMD or Nvidia.
What requirements will you have for CPU and memory footprint on these engineering workstations in 1-2 years? Is there a benchmark so I can do some modeling and figure out what cpu you'll need and try to guess what process technology will be required to fabricate it for the requirements you need? Can you write drivers and have applications work in a way that would run distributing the computations over several compute nodes the way the VisWall http://www.visbox.com/viswall.html might potentially do? _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
