On 9/8/07, Alex Beregszaszi <[EMAIL PROTECTED]> wrote: > Hi, > > On Fri, 2007-09-07 at 22:32 -0400, Timothy Normand Miller wrote: > > [...] > > > In practice, the VGA BIOS would be the first thing to program the VGA > > microcode into the nanocontroller. The Linux driver for our card > > would know about where to find that code in the PROM so that it can be > > reloaded when necessary. Our X11 driver is probably what would load > > the DMA microcode (and reload the VGA microcode when safely switching > > consoles). On exit from X11, we'd probably arrange it so that the > > kernel driver restored the VGA program, so that it gets restored > > properly even in the event of a crash. > > > > If there are memory blocks free in the XP10 when we've gotten far > > enough along, we might increase the size of the program file. This > > way, we can have both programs loaded at boot time and not require any > > reprogramming (except perhaps for downloadable updates installed when > > the OS comes up). Then switching is a matter of setting a mode flag; > > the currently running program would detect this in its main loop and > > just jump into the other program. > > Oh one thing regarding the bios. Please add some public interface to > program the bios flash chip, thus updating the bios would be possible > even if the chip is soldered. (Most of the mainboard chipsets and some > of the NICs have such support)
All of the interfaces will be public. They're also simple. > > And major updates could be programmed into the card. > > In the finished card will be the Xilinx and Lattice chips merged into > one ASIC, or two, or you'll leave the Lattice there? One chip. > Does the > development board currently has a flash chip for bios (SPI rom numbered > H?)? As I see Lattice has some space to program itself (I have to admin > I am not hardware guy). There are three flashes. One is in the XP10 so it can program itself. Another is for the firmware. The third contains the programming for the Spartan. We'll make sure all three are clearly documented. -- Timothy Normand Miller http://www.cse.ohio-state.edu/~millerti Open Graphics Project _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
