tattar8

Probably not since the LCD is physically connected to the faulty internal video card.  You might be able to get video on another monitor connected to the eGPU.

In my experience, having the 750m enabled always gave me problems.  There's a hack you can perform to enable the integrated Intel graphics on Windows; you can then switch to it and disable the 750m.

In the Windows world, the NVIDIA drivers include Optimus, which allows an NVIDIA GPU to render images displayed on a monitor connected to an integrated GPU; this works on a Boot Camp installation on a MBP as well; you have to use a utility to force EFI to enable the integrated GPU for use by Windows to use Optimus, otherwise it will always default to the discrete GPU.  As far as I know, there's no Optimus for OSX, so you're out of luck there.  The graphics switching on discrete GPU-equipped MBPs seems to actually switch the card the monitor is connected to.

I'm trying to set up a GTX 770 I have laying around using an Akitio Thunder2 enclosure on my 2014 rMBP15, which has a discrete GT 750.  I've performed the barrel plug ATX mod (so I don't need the riser) with a 400W ATX PSU.  The setup works fine with a POS budget Radeon card (an R5 220), but not with the 770.  In OSX, after running the automate egpu script, the card is recognized as "NVIDIA Card" with way too little VRAM.  In Windows, the card is recognized fine as a GTX 770, but it throws an error 43 that I can't get rid of.  Is there some kind of conflict between the discrete 750M and the 770 (maybe because they're both Kepler based cards?)?

Is there some sort of conflict between a GTX 770 and a discrete GT 750M?  I'm on a MBP with the discrete card and I can't get the GTX 770 to work in either Mac or Windows.  Mac just reports the card as "NVIDIA Card" with way too little VRAM, and Windows recognizes the card but throws error 43.