On 15 Feb 01 at 10:07, Trent Piepho wrote: > On Thu, 15 Feb 2001, Alan Cox wrote: > > > would take a seperate sequencer instruction to the bt848 DMA controller each > > > time you want to reset the address back to the start. It's not going to work > > > if you want to do this for every pixel, but you had a window 16 or 32 pixels > > > wide in the FPGA's PCI address space, you could probably do it. > > > > Since the 848 will burst in order the obvious approach to me would be to map > > the FPGA 32bit wide FIFO register across an entire 2Mb of PCI space. Just > > forget to decode a few bits 8) > > Doesn't nvidia have a patent on this? I thought one of things they sued 3dfx > over was mapping a register over an address range, so burst transfers that are > designed to write blocks of memory can write to a single register efficiently. As matrox does it since Millennium I (at least, I never looked at their ISA cards) with their pseudoDMA window, I do not think that nvidia can ever win such lawsuit. Best regards, Petr Vandrovec vandrove@xxxxxxxxxx