Mac Nvidia CUDA Updates (Top of Page are Mac Nvidia Graphics Driver Updates - later first.) CUDA 418.163 Update for macOS High Sierra (May 10, 2019) Download of Nvidia CUDA 418.163 for macOS High Sierra. Nvidia GPUs receive driver updates soon after each version update of OS X. Only one driver is released by Nvidia and it includes support for. There is still no fix for the CUDA problem with Mac OS Mojave. I will update the this section once one is available. I have switched to the Windows Notebook. I just updated to the latest macOS Sierra 10.12.2. The NVIDIA CUDA driver in the System Preferences pane shows the following (as you can see the driver shows 'No newer CUDA Driver available') but it shows in red as Update Required. Every time my Mac restarts, this CUDA System preference pane shows up. Any help would be appreciated.

Hi guys,

The macOS High Sierra 10.13 upgraded make me sick because of its incompatible with the GPU CUDA driver. I came here after weeks searching for a solution that fixes the annoying “Update Required” message showing everytime my Mac startups, also making all of my CUDA-required software I using like Adobe Premiere/After Effect runs and renders very slow. Finally, I must come here and register an account for only this, and glad I found it, the solution that saves my life XD

Ok so I won’t take your time anymore on how happy I did, let’s go straight to the point how I really made it:
The only information you need to know is: Right now the macOS’ native graphics driver that help the system communicate with the nVIDIA GPU, is still not updated to the really compatible version. So what we have to do here is to install the “web version” of the driver, which is an OFFICIAL version from nVIDIA, it’s just not a “native version” from Apple.

In quick view, what I have done to solve the problem:
1. Uninstall the incompatible “native version” CUDA from my Mac.
2. Download and install the latest “web version” of NVIDIA DRIVER and CUDA from nVIDIA’s website.

Let’s go in a more detailed how to fix this tutorial:
STEP 1. UNINSTALL THE MACOS CUDA DRIVER
Navigate and delete all CUDA existing native drivers and related components installed on your Mac (by right-clicking at Finder on the dock and selecting “Go to folder…”:

/System/Library/Extensions/CUDA.kext
/Library/Frameworks/CUDA.framework
/Library/LaunchAgents/com.nvidia.CUDASoftwareUpdate.plist
/Library/PreferencePanes/CUDA/Preferences.prefPane
/System/Library/StartupItems/CUDA/
/Developer/NVIDIA/CUDA-5.0
/usr/local/cuda
(Some of those will be empty and it’s normal)

OR you can try the method that use Terminal to do that from here: http://docs.nvidia.com/cuda/cuda-installation-guide-mac-os-x/index.html#uninstall

After step 1, make sure to do a restart to really get your Mac to a fresh state.

STEP 2. DOWNLOAD AND INSTALL THE NVIDIA DRIVER

First, download and install nVIDIA driver that supports up to macOS 10.13.3 here: http://www.nvidia.com/download/driverResults.aspx/130460/en-us
Select the option NVIDIA WEB DRIVER if you been asked. Do a restart after installation, then go to System Preferences > NVIDIA Driver Manager > Updates > Check Now - to check and update if the are newer version.
Second, download and install the lastest nVIDIA CUDA support for MacOS here: http://www.nvidia.com/object/macosx-cuda-387.128-driver.html (Updated Jan 2018. For the lastest version please check at http://www.nvidia.com/object/mac-driver-archive.html)

These installations may require you to restart your computer several times, just go ahead. And do it one more time when everything are all installed, too!

…aaaaand boooom, it’s done!
I think you may not see the annoying “Update Required” message anymore, also there’s now an nVIDIA icon on your menu bar - which showing the new nVIDIA web version driver is managing the GPU right now (Thank God!). You can also hide it in System Preferences > NVIDIA Driver Manager and uncheck the “Show NVIDIA Driver Manager in menu bar” checkbox. Try checking your CUDA-required software to make sure you have solved the problem, too.

If this could help you, please leave a reply so other people can see!
That’s all. I hope it helps! Sorry for my bad English :P

Thanks, and have a good day from now!!
Hoa Nguyen.

The best laptop ever produced was the 2012-2014 Macbook Pro Retina with 15 inch display. It has a CUDA-capable GPU, the NVIDIA GeForce GT 650M. This GPU has 384 cores and 1 GB of VRAM, and is CUDA capability 3. Although puny by modern standards, it provides about a 4X speedup over the cpu for Pytorch, and is fine for learning Pytorch and prototyping. If you have a newer MacBook Pro you are out of luck, because it either has a Radeon GPU or none at all.

The standard Mac distribution of Pytorch does not support CUDA, but it is supported if you compile Pytorch from source. There are numerous preliminary steps and 'gotchas'. Here is what you need to do. Thanks to Jack Dyson for this write up based on an earlier version that I published earlier. These instructions have been tested for:

OS : MacOS High Sierra 10.13.6 (17G14042)
GPU Driver: NVIDIA Web Driver 387.10.10.10.4.140
GPU CUDA Driver Version: 418.163
Xcode Version: 10.1 (10B61)

Downgrade to High Sierra

Check that you are running Mac OS X High Sierra (10.13.6). If you have an older version, upgrade. If you have a newer version you will need to downgrade; Apple banished CUDA with Mojave and later versions of the OS. Downgrading OS X requires creating a bootable USB memory stick installer and erasing your laptop's hard disk.

Install Xcode

Check that you have installed Xcode version 10.1. If you have a newer version or none at all, download it from the Apple Developer site. Rename any other version of Xcode you have installed, and then copy it to /Applications. Open Xcode, and under preferences, select the 10.1 command line tools. Close Xcode and open a terminal. Run

xcode-select --install

to reinstall the command line tools, because sometimes the Xcode application fails to install certain header files. Cuda for mac sierra free

Install NVIDIA Drivers

Install the NVIDIA Quadro and Geforce OS X Driver 387.10.10.10.40.140.

Add to your .profile and reboot:

export PATH=/Developer/NVIDIA/CUDA-10.0/bin${PATH:+:${PATH}}
export DYLD_LIBRARY_PATH=/usr/local/cuda/lib:$DYLD_LIBRARY_PATH

Install NVIDIA cuDNN 7.6.5.

Install Conda

Install Anaconda. Create an environment named ptc that includes pip, activate it, and install libraries:

conda create --name ptc python=3.7
conda activate ptc
conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing_extensions future six requests dataclasses
conda install pkg-config libuv

Build Pytorch

Now you are ready to build Pytorch with Cuda!

conda activate ptc
git clone --recursive https://github.com/pytorch/pytorch

Finally you build:

cd pytorch
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-'$(dirname $(which conda))/../'}
MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py install

Post Installation

conda activate ptc
pip install torchvision

Reboot and test that pytorch with CUDA is working:

conda activate ptc
python
import torch
torch.cuda.is_available()

If python does not print 'true', something has gone wrong. Mac

Install ptc as a kernel for jupyter notebooks.

conda deactivate
conda install ipykernel
python -m ipykernel install --user --name ptc --display-name 'Python 3.7 (ptc)'

When a program first invokes CUDA, a warning message will be printing stating that the GPU is too old. The message can be ignored - CUDA will indeed work! In order to eliminate the message, edit the file

Cuda Driver For Mac Sierra

~/anaconda3/envs/ptc/lib/python3.6/site-packages/torch/cuda/__init__.py

Cuda For Mac Sierra