How about this: Small Actel flash-based FPGA (the flash is in the CLBs, so the programming is instant). This one FPGA contains PCIe or an interface to an external PCIe, memory controller (got one), and video controller (got one). We could even support VGA text console (got that too).
Steps: 1. Figure out our PCIe solution 2. Cobble together OGD1 logic components into simplified solution 3. Synthesize to see how much logic are we need 4. Choose FPGA 5. Design board 6. ??? 7. Profit What's on the board? - Small FPGA - Maybe external PCIe chip - Video encoder (DAC, DVI) - Memory (if we can get cheap DDR1, we already have a controller) We need someone to take charge of this, and I can advise. From the perspective of the university and funding agencies, I'm chasing energy efficiency in CPUs, GPUs, and memory systems, and I'm asserting that the lack of research (simulation, etc.) infrastructure is an impediment, so they should fund that too. I'm trying to solve multiple problems at once, but that does constrain what I personally can spend time on. On Mon, Dec 10, 2012 at 12:35 AM, Troy Benjegerdes <[email protected]> wrote: > > Mind you, someone probably thought the whole Steam idea was a > distraction, > > while I think it's a good idea, so I'm not saying to stop brain-storming. > > I'm just saying that we should divert more of our energy towards > > developing an energy-efficient GPU. The more we get distracted by > > pie-in-the-sky ambitions, the longer it'll take us to make something even > > mediocre. First, we do the things we can do now, with the resources we > > have, and then we use those to achieve something else. > > Brainstorming is good. As for mediocre, what's the fastest path to getting > some sort of very basic framebuffer and a dumb resistor VGA output DAC, > and a simple single shader in VHDL that I can put on an XESS Xula2, or > Actel FPGA, or DE0-Nano board? > > Or maybe drive a 32x16 LED matrix? There's lots of interesting applications > with high potential margins that don't look like a PCI-E video card. > -- Timothy Normand Miller, PhD Assistant Professor of Computer Science, Binghamton University http://www.cs.binghamton.edu/~millerti/<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)
