On Mon, Dec 05, 2005 at 04:41:54PM +0100, Attila Kinali wrote: > On Mon, 5 Dec 2005 15:16:27 +0100 > Luc Verhaegen <[EMAIL PROTECTED]> wrote: > > > Why not go for a pair of DVO like interfaces next to a single SIL178 and > > a single ramdac. Add a VIP like connection next to one of the DVOs and > > you can have video in. > > What exactly is DVO and how does it differ from DVI? > A quick google didn't reveal anything. > > BTW: we need at least a dual channel DVI port to support high > resolution displays (ie most TFTs these days). And we want > to have two DVI outputs, thus we need 2*2=4 SIL178 > > > You can find this sort of setup on many GeForce2 boards: 1 dvo like > > interface with a VIP bit next to it. Elsa Gladiacs were able to use a > > BT869 or an SAA711x on seperate daughterboards. Geforce2s and the early > > MXen are quite interesting pieces of hardware output/input wise. (Too > > bad that the X driver is a mess modesetting wise). > > This is for TV-out ? There we need only one of course, but > it should be somewhat seperated from the two screens as > people want to be able to assign the first or the second > screen to tv-out. > > Attila Kinali > This was said later on in this same mail:
> DVO is digital video output or something, and is used on most intel > i8xxen. The i9xxen, with pci-express, have moved to SDVO. The ADD cards > intel made for the i845 and i865 were possible because of 2 DVO busses > being multiplexed with AGP. You can find more information about this in > the intel northbridge datasheets, which are freely downloadable (in > contrast with actual intel IGP programming information). DVO is > basically a 12bit data bus, with H/VSync, enable and i2c. It allows connection of most TMDS, LVDS and TV encoders, if chosen right, and if the fpga (or implementation) is able to handle, it should even allow for direct input. It's a semi-standard, and if the open-graphics hardware uses a common header for all these modules (they probably should include power and ground), mixing and matching will become almost trivial. For instance: standard og fpga device comes with 1 ramdac/CRT controller, 2 TMDS encoders and one TV encoder (which usually can also be used as a CRT controller). This should allow for 2 single link DVI-Is, of which one can be used as a composite/s-video/RGB or SCART (composite sync)/YUV TV connection. You can see this trick being applied in the matrox Parhelias, where 2 Chrontel shared TMDS/TV encoders are being used per DVI connector. Since cvt reduced blanking (which will be fully supported in Xorg 7.1) allows for a decent resolution already, the dual link DVI is not an immediate requirement for everyone. If you do want dual link DVI, then there are 2 further DVO like connections, allowing for SIL178 daughterboards which connect to a header behind the DVI connectors for the secondary DVI links. The rest is handled in the FPGA and in the driver. Similarly, another TV encoder can be stuck on the same DVO like headers, although that will require a seperate cable. Same thing for LVDS encoders, same thing for framegrabbers, be it digital or be it from TV or RGB. Why i dragged in VIP there i don't know, the VIP/capture (in) ports on the unichromes are usually used for TV or TMDS (out) too. Luc Verhaegen. _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
