2016 MBP 15" - GTX960 in Razer Core won't detect  

  RSS

demonguard
(@demonguard)
Active Member
Joined: 2 years  ago
Posts: 7
February 25, 2017 12:54 am  

Hi there! I have a 2016 15" MacBook Pro and a razer core. When I first started to set up the core, I successfully made it through the process of seeing it as an unsupported device, running tb3-enabler, and proceeding onward to automate-egpu. However, at that point I realized that I couldn't install the Nvidia web driver while on a beta build of Mac OS. I used the App Store to update to stable 10.12.3. Following this, I proceeded through the rest of the setup but never got the EGPU to actually detect in macOS. At some point in this ordeal, I ran tb3-enabler again and I believe that it used a backup of my beta kext or something because TB3 stopped working entirely. Even after a full format/reinstall from recovery the thunderbolt section of System Information will not show the core - even as "Unsupported". I even tried NVRAM/SMC reset and still have nothing. The core will charge the device, but no more.

 

How should I proceed from here? Is its possible that the kexts on the recovery partition are mismatched as well now?

 

Running TB3-enabler again, on a totally fresh install, yields:

tb3-enabler austin$ sudo python tb3-enabler.py apply

Password:

backing up... done

patching... done

clearing kext cache...kext-dev-mode allowing invalid signature -67061 0xFFFFFFFFFFFEFA0B for kext IOThunderboltFamily.kext

symlink("/System/Library/PrelinkedKernels/prelinkedkernel", "/System/Library/Caches/com.apple.kext.caches/Startup/kernelcache") failed 17 (File exists) <createPrelinkedKernel 2795>

 done


ReplyQuote
demonguard
(@demonguard)
Active Member
Joined: 2 years  ago
Posts: 7
February 25, 2017 1:34 am  

UPDATE: I'm an idiot and the core wasn't fully closed apparently. Anyways.

I've done tb3-enabler and automate-egpu now. On boot the GPU fans whoosh, but I get no output and no detection in macOS. I can see "NVIDIA Chip Model" in system info but it doesn't seem to do anything. 

Despite tons of restarting/unplugging/etc/etc I can't seem to get any further than NVIDIA Chip Model. Manual and automatic modes on the script seem to yield the same result.

 

Edit: For anyone that finds this, I ended up restarting like 50 more times and it started working around the time that I installed the CUDA driver for no apparent reason. I believe I'm using manual mode on the automate script. 


ReplyQuote