Machine Learning Framework PyTorch Enabling GPU-Accelerated Training on Apple Silicon Macs

In collaboration with the Metal engineering team at Apple, PyTorch today announced that its open source machine learning framework will soon support GPU-accelerated model training on Apple silicon Macs powered by M1, M1 Pro, M1 Max, or M1 Ultra chips.

Until now, PyTorch training on the Mac only leveraged the CPU, but an upcoming version will allow developers and researchers to take advantage of the integrated GPU in Apple silicon chips for "significantly faster" model training.

A preview build of PyTorch version 1.12 with GPU-accelerated training is available for Apple silicon Macs running macOS 12.3 or later with a native version of Python. Performance comparisons and additional details are available on PyTorch's website.

Tag: Metal

Popular Stories

iOS 18 CarPlay Feature

iOS 18 Adds These 5 New Features to CarPlay

Thursday June 13, 2024 7:44 am PDT by
Apple did not mention CarPlay during its WWDC keynote this week, but iOS 18 includes a handful of new features for the in-car software. Overall, there is not a whole lot new for CarPlay on iOS 18, with changes seemingly limited to the Messages and Settings apps so far. Below, we recap everything new for CarPlay on iOS 18. New for CarPlay on iOS 18 1. Contact Photos in Messages App...
ios 18 button bulge

iOS 18 Adds Pop-Out Bezel Animation When Pressing iPhone Buttons

Tuesday June 11, 2024 10:40 am PDT by
iOS 18 includes a small but interesting change for the buttons on the iPhone, adding more of a visual element when changing volume, activating the Action button, or locking the screen. When you press an iPhone button in iOS 18, the display bezel bulges outward slightly. This feature is available for the volume buttons, Action button and the power button, and it will also likely be used for...

First Look at Messages via Satellite in iOS 18

Thursday June 13, 2024 11:29 am PDT by
Apple has been gradually expanding its suite of satellite connectivity features for iPhone, and iOS 18 brings a significant new one in the form of Messages via satellite. The feature allows users to send and receive iMessages and SMS texts, including emoji and Tapbacks, while out of range of cellular and Wi-Fi networks. CNET met up with Apple's senior director of platform product marketing,...
iOS 18 Wallet Feature

Here's What's New in Apple Wallet on iOS 18 for Event Tickets and More

Friday June 14, 2024 7:32 am PDT by
iOS 18 includes a handful of enhancements to the Wallet app on the iPhone, with new features for Apple Pay, Apple Cash, event tickets, and more. Below, we outline everything new for the Wallet app on iOS 18, based on information from Apple's press release and a WWDC 2024 coding session. Redesigned Event Tickets Event tickets have an all-new design in the Wallet app on iOS 18, complete...

Top Rated Comments

innominato5090 Avatar
27 months ago
So unbelievably excited about this. Being able to test GPU-accelerated code locally (and even train some smaller networks) rather than having to rely on a unix-server will speed up ML development significantly.

Can't wait for it to become mature enough to be merged in a stable branch.
Score: 13 Votes (Like | Disagree)
mukiex Avatar
27 months ago
Personally I'm hoping to see support for the neural engine in the future. When they added NE support to Topaz, my M1 Macbook Air suddenly started performing on par with my desktop GTX 1080.
Score: 5 Votes (Like | Disagree)
27 months ago
I hope to be blown away with what wwdc has to announce
Score: 4 Votes (Like | Disagree)
adamw Avatar
27 months ago
Always good to see move native support for Metal in MacOS on M1 family of chips...
Score: 3 Votes (Like | Disagree)
seek3r Avatar
27 months ago
Nice step in the direction of finding out how capable these GPUs are for this kind of workload, may also give an idea if they work well enough for GPGPU and ML work to infer if Apple is likely to develop their own GPU for the eventual AS Mac Pro
Score: 3 Votes (Like | Disagree)
Luca1995it Avatar
27 months ago
I started collecting benchmarks of M1 Max on PyTorch here:
Score: 2 Votes (Like | Disagree)