Contribute
Register

NVIDIA Releases Alternate Graphics Drivers for macOS High Sierra 10.13.3 (387.10.10.10.25)

Status
Not open for further replies.
Apply Clover kext patch from here and the drivers don't need to be patched. Download the patcher from post #11 and you can install any driver from the Nvidia Drivers page.

Nice! I actually really like the idea of approaching this issue from Clover. However, the solution may not be for everyone as you have to know what you're doing with Clover. Well, and use clover (this driver issue affects real Macs too).

Problem with latest driver? Reinstall previous working one. No bogus claims or black list required.

Another issue with a pure Clover approach is that what you've suggested simply won't work. You still can't install old packages on your system unless you patch the installer manually, or install with nvidia-update or web-driver.sh.

No bogus claims or black list required.

Calling the black list bogus is a bit odd though. Why do you believe it to be bogus?

It's a usability thing, trying to make this as simple as possible for your average user. People need to get information about which drivers are going to work for their system from somewhere. Sure, they can read a bunch of threads, or it can be automated for them with minimal fuss.
 
You still can't install old packages on your system unless you patch the installer manually, or install with nvidia-update or web-driver.sh.
You can drop the packages onto the applet in post #11 of that thread to remove the installation checks, and the 'liberated' packages can be saved for later use, offline use etc. The driver package and preference pane components are untouched.
 
You can drop the packages onto the applet in post #11 of that thread to remove the installation checks, and the 'liberated' packages can be saved for later use, offline use etc. The driver package and preference pane components are untouched.

Ah, I certainly do like the sound of that a lot more than some of the alternative approaches. Nonetheless, it still sounds like a multi-step procedure and requires a fair bit of knowledge about what you're doing (which drivers will work etc.)

However, I can't find the applet in post #11 you mentioned, if you can post a direct link I'm sure that'd be helpful.

I've been considering a few different alternatives myself, either a one-stop macOS Cocoa app updater/installer or using a signed codeless kext. However, my hope is that nVidia will simply stop messing stuff up and none of us will have to worry about this anymore ;)
 
Ah, I certainly do like the sound of that a lot more than some of the alternative approaches. Nonetheless, it still sounds like a multi-step procedure and requires a fair bit of knowledge about what you're doing (which drivers will work etc.)

However, I can't find the applet in post #11 you mentioned, if you can post a direct link I'm sure that'd be helpful.

I've been considering a few different alternatives myself, either one-stop macOS Cocoa app updated/installer or using a signed codeless kext. However, my hope is that nVidia will simply stop messing stuff up and none of us will have to worry about this anymore ;)
Here's the download link for the repackager. Credits: @bogdanw @Pavo

The problem with the codeless kext as you say - it needs to be signed to have any benefit. There's an xcode project here, NVDARequiredOS gets set at build time.
 
Last edited:
Here's the download link for the repackager. The problem with the codeless kext as you say - it needs to be signed to have any benefit. There's an xcode project here, NVDARequiredOS gets set at build time.

Excellent, I've linked to your codeless kext here, so if I do end up trying that approach we can avoid duplication of work. The company I run (Glass Echidna Pty Ltd) has code signing permissions.

I may be prepared to sign the kext. However, I can only do so if it's handled in a way that won't upset Apple. In the past they actually explicitly whitelisted one of our kexts (Heimdall - Samsung phone flashing) during the unsigned to signed kext transition period, so they know we sign kexts (that are somewhat unusual in nature).

However, in order to stay in Apple's good graces I think my company would need to release a public "Kext Signing Policy" etc. and we will unfortunately never be able to sign kexts that are specifically for Hackintosh.

However, seems as the kext in question is perfectly useful on real Macs (see egpu.io forums) then I think code signing could be arranged.
 
Last edited:
Excellent, I've linked to your codeless kext here, so if I do end up trying that approach we can avoid duplication of work. The company I run (Glass Echidna Pty Ltd) has code signing permissions.
Ok. The differences from stock NVDAStartup are only a bumped probe score and the 'os build variable' in NVDARequiredOS.
However, seems as the kext in question is perfectly useful on real Macs (see egpu.io forums) then I think code signing could be arranged.
So it is up to Apple...
 
So it is up to Apple...

Not quite. We have already been provided with valid code signing certificates with kext signing permissions and can sign whatever we want.

However
, we're bound by Apple's developer agreement etc. so don't want to accidentally do anything that may get us in trouble. So as I've said, the kext would need to be designed and distributed with real Macs in mind. However, I'm presently running old nVidia drivers (installed with nvidia-update) on my Macbook Pro (because both Apple's latest and nVidia's latest drivers lag). So there is a real-world benefit for such a kext.
 
Both of you don't seem to be using the hardware that causes the lag, namely Skylake/Coffeelake AND Pascal cards running on 10.13.3
Exactly. I have a Kaby Lake i7 CPU with a GTX 1080 and for me, the Nvidia web driver .158 is laggy. What is the method recommended to update/roll back to a driver with no lag?
 
Status
Not open for further replies.
Back
Top