Uuugz

Thanks, and as for your setup, the X360 power supply kinda worries me. It will probably work, but since it was developed specifically to work with the X360, it may not be able to properly power a graphics card that was designed to work with ATX power supplies. Not saying that I'll fail, but I'm not aware of it's specs, and I've never seen anyone use an X360 power supply for it. The rest seems fine, and personally I got a PSU that has a separate 6pin power cable for the GPU, however using a 6pin out on the dock works just as well. Hope everything works for you in the end!

Alright, I turned the most demanding settings down a notch (shadows, lighting, reflections) and had no crashes since then. My setup seems to be done at this point, and unless something noteworthy happens, I don't think I'm going to post much. Thanks everyone, and may your stuff work as intended!

Update: Installed 365.19 drivers (Doom-optimized) without using DDU and without modifying them, and interestingly, everything works now. Doom is running around 35-50 fps on Medium using the internal display, and around 50 quite stably using an external monitor. It crashed my system to a black screen once, which was unexpected. I'll try it some more and post if any crashes happen. Wish me luck ...

I'll look into it. In my case, the CPU doesn't seem to throttle, as it's sitting at 30%-ish usage and about 10°C under it's regular "fully utilised" temperature, so it's just not being used for some reason. Could be a driver issue, since 350.12 drivers are about a year older than Doom itself. And yes, I do get 60fps in less demanding games, I even see stretches of 50+ fps in Dying Light, which isn't particularly easy to run. Also, the 4GB version of GTX960 is quite a good choice, it's a safer bet than the 1060 in terms of getting it to work. And for now, avoid the 1050's and 1050Ti's, they do not work yet, I learned that the hard way. Traded it for my 960 in the end.

Update: Tested Doom (2016) today, and it's really weird. CPU usage howers around 30-40%, and framerate is identical 20-35 fps on low, medium and high presets. Performance monitoring shows that it's held back by the CPU, which takes up to 40ms to process each individual frame, while it only takes the GPU about 10-25ms. Tested on OpenGL, as Vulcan seems to prevent Doom from launching. Will try to update drivers while using an external monitor tomorrow. Any suggestions are appreciated, as always.

Update: The framerate drops & stuttering seem to be unrelated to VRAM usage, just tested Borderlands 2 and VRAM never exceeded 400MB, GPU usage is around 60-80%, CPU usage is around 80% at all times. RAM usage is just <4GB. During most of the heavy stutters CPU usage jumps to 100%, so the problem is CPU-related, however when CPu is around 70% the issue is still present. I'm going to test some more. Edit: Dying Light was indeed having VRAM issues, dropped the texture quality and stuttering is gone.

Hello, and first of all, you're welcome. As for tricking the laptop into thinking that eGPU is a dGPU, I believe that's exactly what happened. The Nvidia control panel only gives me an array of settings present for mobile GPUs, not the full set, which means the OS sees my eGPU as a dGPU. I have read the owen-lu eGPU guide, and considering my setup only worked when I installed 350.12 drivers, it was helpful, but if you open up some new drivers you'll see quite a lot changed in their .inf files, meaning you'll have to do some guesswork, as the guide is relevant for 353.xx drivers. As for the stuttering, I'll test it more extensively today, including VRAM monitoring, as it seems to be somewhat game-specific (Dying Light being the worst case, and Fallout 4 being somewhat tolerable). Will post once I do.

GPU-Z shows that it's using PCI-Express x1 v2.0, so everything's fine there. I'm running FurMark in the background by the way. Stuttering may be a bandwidth issue, since PCI-E is limited to x1.

and how do i do that?.. sorry if it's something really obvious, but i have no idea.

Also, do you buy chance also use the EXP GDC adapter? If so, is the "ATX PW" switch in the left or right position? Edit: moved the switch into the left position, things seem to be fine. Now the eGPU and the PSU turn on and off when I turn the laptop on and off. Previous freeze seems to be game-specific. Aside from some stuttering, games are stable to run. Will try connecting to a TV via HDMI to see if stuttering/framerate improves.

Alright, modded the 350.12 drivers with the correct lines for my GTX960, as well as downgraded the drivers for my HD4000 to 10.18.10.3621. The internal display is now properly recognised, however the game I was testing froze a few minutes after launch. Working on solving this at the moment.

Just tried the 8.1 drivers, and they refuse to install. However they had a folder with installation files for WIN10 alongside the folder for WIN8.1, which installed just fine but resulted in eGPU only working with an external monitor again. I'll try installing the old 350.xx drivers, apparently after them Optimus got broken on WIN10.

yeah, the problem for now is that the nvidia control panel is impossible to open. i'll try the WIN8.1 drivers tomorrow and post the results when i do. Thanks for the advice, since my laptop initially came with 8.1 on board it is worth a try.

will do and post as soon as I tried it

It does work with the external display, I tested. It is the laptop display issues i'm trying to sort out. So yeah,probably optimus' fault in some way. maybe i should try using laptop drivers? Edit: I'll try asking in the eGPU troubleshooting topic tomorrow, hopefully something will come of it.

seems that way. as for the display settings, there really is nothing in there in WIN10 going into display settings - advanced - adapter properties

Here's the Device manager, showing the active display adapters and on the second screenshot, the pci-e bus part that I disabled

No, unfortunately I am not familiar with it, however I have a general idea of how it's done. May try it if everything else fails. Also, did you by any chance mean screenshots? because if you actually mean snapshots, i would appreciate it if you told me how to make them edit: googling it at the moment

Tried the latest drivers just now, and nvidia control panel gives the "no monitors connected to an nvidia gpu" message. Will try the older version now.

Currently running Windows 10 64-bit.

You mean the latest one? Later than 372.70? Some people posted that anything later than it doesn't work, but I'll try.

Just tried disabling the part of PCI-E bus that the 740m is connected to, as you suggested, and wiped all drivers, rebooted and installed the modded eGPU ones, however the issue still persists, meaning the 960 doesn't work with the internal display... The 740m is nowhere to be found in the device manager. I will try a few other things, though. Will post as soon as i do. Any suggestions are still appreciated, maybe my driver modding work isn't done properly?

Actually, if the PCI-E lanes used for dGPU and eGPU are different and do not overlap, disabling the portion connected to the 740m might just work (fingers crossed). That's worth a try. Thank you for your time, and again, I'll post once I try doing this.

Alright, just DDU'd my drivers in safe mode, disabled the driver signature enforcement, disabled the "other device" in device manager (740m judging by ID) and installed the 372.54 drivers modded with my lines above. After the restart, both the 740m and the 960 had their drivers installed, even though I disabled the prior. 960 works as it used to, and Nvidia control panel gives the same message, so no luck this time... Maybe I should try removing the 740m from nvami.inf so that drivers for it don't get installed? It's not possible to disable it in BIOS, btw, option is not there. Edit: Maybe using the "uninstall" option in the device manager is what I should try?