How to get full H2D TB3 bandwidth in macOS?
Hi @fr34k. As far as you all have been testing, is there any H2D bandwith improvement with your new script? Just curious.
As it has been already said, thank you very very much for your dedication and hours and hours put into this. +1 to a way of donating! Can't wait to try it! Keep it up!
Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts
@aweenz, we’ve seen:
The problem appears to be the Mac firmware reserves 10/32 of the 22Gbps link, seemingly for 10Gbps USB 3.1 Gen2 if it detects a eGPU and boot into macOS. So try the strategies in the following thread to bypass Apple firmware seeing a eGPU:
Easiest method would be to *carefully* hotplug the video card into the powered eGPU enclosure while in macOS. If that works, then add a PERST# switch, as described in the above thread, to make this a more workable solution.
@goalque, could you create and test a EFI file to be run by rEFInd/USB stick that sets the OS to be Windows instead of macOS (ie: a apple_set_os_windows.efi instead of apple_set_os.efi)? Perhaps that may provide full H2D TB3 bandwidth in macOS?
@fr34k or anyone in the know, is there a way to get a full speed H2D connection on TB3 with this script? I read somewhere here a while back that it was a no with Nvidia cards. Cuda-z screenshot linked below with my current speeds displayed on a UHD LG 27" eM fed by eGPU GTX1080. Thank you!
One question: I installed my KFA2 eGPU in bootcamp on my MacMini 2018 and the H2D speed is the full (or almost) 2232 MiB/s.
Is the bug half H2D only applicable to MacOS then? I have not flashed this unit but a previous one that I returned due to a faulty power button.
Mac Mini 2012 and 2018 i7, Mac book Pro 2016 15" with TB, eGPU Galax/KFA2