Just my opinion, but we shot too high with OGD1... scale it back until you have something simple and cheap.
On Fri, Dec 7, 2012 at 8:15 PM, Dieter BSD <[email protected]> wrote: > Here is a summary of what I think we want in an open framebuffer board: > > PCIe Number of lanes to be determined > One > At least 2 heads How much would extra heads increase price? > I think 3 heads should cover most users' needs. > Many people won't use the extra heads. If a one-head board sells, make a two-head board. Two heads didn't help sell OGD1. > At least 2560x1600 Is 4096*2160 unreasonable? > That would require dual-link DVI and really really good DACs. Too expensive. Stick to the max for single-link DVI. This gets 90% of the users, even now. > At least 24 bit color > Yes. > Support for VGA, component, s-video, composite (others?) Support for DVI, HDMI, Displayport (others?) > I can be convinced about DVI, but perhaps we should consider VGA only. > Support for use as console > We have all the parts and firmware for this already. > > Support for adding an optional OGP-GPU (Open Graphics Project GPU): > If extra space on a PCB is cheap, maybe. My philosophy is to either go fully general (which we did with OGD1, and it failed), or optimize cost by making highly specific solutions. Consider doing the latter. > > Option 1) Socket for FPGA / DSP / other? > use the same pinout for OGP-GPU chip > It can be hard to find a really small and a really large FPGA that share the same pinout. I don't know of any socketed FPGAs. > > Option 2) a mini-PCIe slot or 2 > put OGP-GPU on mini-PCIe card > (allows adding an optional Broadcom Crystal HD video decoder, > or any other Mini-PCIe card) > > Option 2 is probably less expensive. Would it create a bottleneck > or other problem? > > Bracket connectors: > Option 1) DVI, s-video, HDMI, and (mini?) Displayport (if they all fit) > Option 2) LFH connector(s) [1] plus breakout cables > > Option 2 might allow having any combination of ports the user needs, > without adapters to convert one type of port to another. It isn't > always possible to convert a port to the type needed. This would require > the output drivers to have sufficient versatility. Is this doable, > oh analog gurus? > > There are existing video cards that use LFH connectors, so it might > be possible to use the existing off-the-shelf breakout cables. > > [1] http://en.wikipedia.org/wiki/Low-force_helix > > People that don't need a gpu can immediately use the board as is. > Once the OGP-GPU is available, users that want/need a gpu can just > plug it in. Depending on what socket(s)/slot(s) we include, users > can add FPGA, DSP, video decoder, etc. Some of these parts may not > be as open as we would like, but they are optional. It should be > possible for the board itself to be completely documented. > "Just plug it in"? No. There's a slight chance that we might be able to fit a different FPGA to the same board, but there's no "just plugging in" with a ball grid array. We could make a daughter board, but the signal integrity would be destroyed, and it would bloat the expense considerably. > > The idea is that we only need to design and build one board that > is versatile enough to serve a wide range of uses, and to remain > useful for a long time. (That's why I'm hoping we can support 4K > displays. 4K displays are rare and expensive today, but are likely > to become less expensive, as consumer electronics usually does.) > And it should be possible to keep the card relatively inexpensive. > I'm sure that price was a major reason that there wasn't a lot > of demand for the OGD1. The power consumption should be quite low. > No extra power cables, no "jet engine" fans, no heatsinks. > Too many conflicting constraints. > > So, who can fill in some of the details that I'm not sure about? > Have I left out anything? > This is a PCB project more than anything else. We already have a BoM for OGD1 that we can crib from. Whoever takes charge of this needs to be experienced with PCB design (certainly not me!) and parts selection. > _______________________________________________ > Open-graphics mailing list > [email protected] > http://lists.duskglow.com/mailman/listinfo/open-graphics > List service provided by Duskglow Consulting, LLC (www.duskglow.com) -- 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)
