Originally Posted by maarten12100
I re-discussed with Moddermike we still do not know whether or not the panel would accept the 120Hz.
The way I understood it the panel can take 120Hz and can display all those pixels but there might be an internal motion interpolation circuit so it can't actually go that high at native resolution due to the inputnot being accepted.
A second thing is that the panel has 4 channels and the dual link DVI receiver and LVDS repacker have a limit of 165M per channel. (only testing could show whether once each works at only 2 channels the rate per channel can be increased)
For reference 505M is needed for 3840x2160@60Hz and of course about an equal number is needed for half the screen @120Hz (slightly differs as blanking has more impact if the resolution is actually lower)
That makes 510M a safe number for each side of the panel.
The DVI dual link receiver and repacker I have in mind can do 450M in specs but the overclockable 1440P screens show that this is actually a fair bit higher.
Also the fact that the controller is still only 8 bit instead of 10(4 pairs instead of 5).
I will send Ineltek an Email they are located in Germany which is close to where I reside (on the map that is, still 500km apart) see if they can give me some sample board for a couple of hundred dollar.
Proof of concept is needed before proceeding.
Luckily for us 10 bit isn't creating an overhead since the LVDS channels were already meant to be used it will however put some extra strains on the chip so earlier "overclocks" might be lower.
Hm sounds similar to that ipad retina screens project I'm hoping to start poking in about a week when I get a little more time. By the spec sheet these also use internal interpolation (supported display modes are 120 and 240 Hz interpolated with 60 Hz input and "120 Hz with 120 Hz backlight") for these but the difference is that I'm going at them through displayport and poking at mine few weeks back did not reveal me a way to slap any higher frequency signal on mine than 60 Hz - I did not look very thoroughly as I could spare only about an hour so might be me just overlooking something or that the in-screen interpolation is too rigid. Granted, your situation is only vaguely similar with the screen and even links being all different (ipad one uses eDP) with just that being similar that your screen uses the interpolation circuitry as well.
As the DVI links carry also the reference clock it might be easier to play them along than fully digital links like DP in my impression. In my opinion the plan to use two dual link controllers is a lot better than to use four single link ones as DVI needs that reference clock and all current GFX cards can only provide two of these. So having to run the screen off two different GFX cards would be an huge increase in complexity in the project I would think. Granted, from top of my head I'm not aware if there is any graphics cards out there which provide two dual link DVI ports - I'm not aware of any from top of my head, but my awareness is mostly limited to AMD side as far as GFX cards go, I suspect nVidia should have some and perhaps even AMD has some and I just have not noticed as I have not looked for these. Ofc problem with nVidia is their spotty support (if you can call that even that) as far as 2 screen Surround goes.
Nevertheless, this project is something I'll be keeping an eye on - it sure will be awesome should it work.
As far as lightboost goes if you can get to the vsync signal and have access to backlight control then something similar to that could surely be possible to hack together for added awesomeness. Mmmm .. maybe. Dunno how hard it is to write custom firmware for that kind of stuff. One of the things I'm planning to look into in a week or two as part of my own smaller project.