On 10/7/05, Thomas Burg <[EMAIL PROTECTED]> wrote: > Ok i read archive of mailing list, and i have still some questions or remarks > : > > - What is the first aim to reach ? > * Get a working prototype to begin and validate some pieces of > software ? > * Get a prototype with some performance abilities ?
First aim: Get a working prototype to begin and validate some HARDWARE. Also, get a sellable product so we can generate revenue. > > - To use 2 FPGA to split fontionnalities is a good thing but I think > it 's not the easiest way to get something working quickly because : > * To make a PCB is timea nd money consuming so make it simple > at first time > * More you have components more the PCB will be large and more > the prize is expensive. > * There is some other tips, costless, to reload the bitstream > of FPGA when computer is turn on : for example a microcontroler unit > who read the bitstream from a mmc when a external action is done (or > by writing at a special adress in the video card) (1) Xilinx PROMs are expensive. (2) Programming a Xilinx on startup takes too long. We need the device to appear instantly on the PCI bus, at power-up. (3) There are no flash-based FPGAs that are large enough to hold the whole design. (4) When we're developing the hardware, it would be very helpful to not necessarily require a reboot when reprogramming the larger part of the design, which will be in the Xilinx chip. > > - Have you look at PCI and vga core available on opencore ? It 's not > necessary to rebuilt the wheel when somebody else already did it. I've looked at them. Saying nothing about the talents of their designers, I don't like them because they aren't structured to fit with the design I'm trying to make. In particular, I hate wishbone, because it isn't designed for efficient synchronous streaming of data -- too much combinatorial logic is required, or you can only clock data every other cycle; either way, there's too much of an impact on performance. _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
