Originally Posted by moddermike
Where is your Vsync signal coming from? The backlight diodes will take whatever you throw at them, but you'd need to have an extracted Vsync signal to push to the backlight driver to do so. This isn't exposed on eDP without packet decoding, and isn't exposed via the other interfaces (e.g. USB HID Monitor) either (at least I don't think so).
Short of another option you'd have to pray that there's an exposed Vsync testpoint on the panel and solder in a flyout back to the controller board. The Macbook panel has one, but I haven't yet bothered to probe it and see what's there.
Well - I have no idea from where the vsync would be coming from. I just pointed out the possibility in the things that backlight controller should be capable of doing by its spec sheet (although I think it was more intended to be an energy saving measure than visual enhancing measure in there). It raised my interest as on the ipad 4 LCD specs at http://www.panelook.com/LTL097QL01-W01_SAMSUNG_9.7_LCM_parameter_17823.html
it list supported display modes as 60 Hz input with interpolation at 120 and 240 Hz and 120 Hz + 120 Hz backlight. Now as far as I understand the backlight is driven by just DC usually so I have no idea what it means to have 120 Hz backlight. Closest thing I can think of is nVidia lightboost like tech with blinking backlight and as such that option caught my eye when skimming over the spec sheet. Ofc the screen will be also perfectly usable at 60 Hz but if it can be driven at native 120 Hz input it sure would be nice bonus.
Although with the eDP signal bypassing the controller stuff straight into panel I would not expect it to be exposed in there indeed. Closest thing without whipping out the soldering iron, perhaps, if possible would be faking the vsync somehow over the USB (don't think decoding AUX channel would be any easier than decoding the whole signal - which is not possible anyway on the board I currently have) but it would be probably exceptionally hard if not impossible to get it actually in sync with the main signal over USB and at best one could probably just make the backlight just blink at 120 Hz and hope its sufficient - should that to be needed to make the LCD to support native 120 Hz mode without interpolation. I'm currently optimistic and hoping that it would be sufficient to just slap the panel with 120 Hz input and that this would be sufficient to make it bypass interpolation. If not then probably making the backlight blink will not help a lot if it's not properly aligned with vsync.
Ofc as I have no prior experience whatsoever coding those chips then unless someone with prior experience ends up helping then this is all just purely theoretical. Unless coding these chips is exceptionally easy and one can get away by just sucking in existing firmware and changing few constants in there somehow. Current firmware is supposed to be rather basic and will not expose the controller board as USB display (which would be nice to have for software based backlight intensity control, for example).
Edit: after taking a good look at DisplayPort standard (http://www.vesa.org/wp-content/uploads/2011/01/ICCE-Presentation-on-VESA-DisplayPort.pdf
) I dont think DP1.1a has actually enough bandwidth for 2048x1536 @ 24 bit color and 120 Hz refresh rate - although it will be short by only a small amount. A 4 lane DP 1.1a bandwidth is 10.8 GBits/s with conding overhead, however, the actually usable bandwidth for the payload is only 8.64 Gbits/s. Problem is - 2044x1536 @ 120 Hz is 9.06 Gbit/s so approximately 5% short bandwidth wise. Well - 5% is not a lot, but I'm not aware of any possibility to actually overclock the DP links as that kind of access is not exposed anywhere as far as I'm aware. Perhaps if you go and start messing with custom BIOS for gfx card ...
Existing bandwidth would be sufficient for 114.4 Hz refresh rate.Edited by Carniflex - 7/23/13 at 6:38am