On 11/24/06, luc <[EMAIL PROTECTED]> wrote:
So, at least three levels of openess: - documented - core - firmware What is the most important of them ? It seems to be the core level, no ? What I mean by 'important' is the level which otherwise would have limited the community.
There are two different kinds of firmware: (a) Software like BIOS and Sun Forth code that are loaded from the device and executed on the host CPU. (b) Loaded either from the device or from the host, software that is executed on the device. For instance, code that runs in a DMA control engine, shader programs, etc. Having made the distinction, I'm going to anticlimactically say that I think they're both software, no different from driver software, and how they are licensed is a separate issue. I would have no qualms about reselling TRV10 chips to some company that wanted to bundle it with proprietary BIOS (I can't think of a good reason) and a proprietary driver (possibly necessary for Windows). Important question to consider: What does it mean to "execute" something on the device? Where is the line between a relatively dumb state machine that uses memory as part of the definition of the FSM and a general-purpose CPU? Is there one? Saying it's in "memory" doesn't help, because everything's in RAM in an FPGA, including everything that defines the function of the hardware. But again, with open hardware, the line is immaterial. If you are prevented from being able to replace a piece of the design (hardware or software), then it's not open hardware. So if I have a binary blob that gets loaded into a block ram in the FPGA to control a state machine, it doesn't matter if it's hardware or software, as long as I can alter or replace it. _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
