If you've been around Macs long enough this is very much in Apple's wheelhouse. They already do something similar with Final Cut Pro rendering, Logic Pro, Grand Central Dispatch, and XCode distributed CI/CD. Apple HQ is probably sitting on 1000s of nodes, high powered MBPs, iMacs, that could do lots of inference work overnight.
Minions is an internal enterprise system already in use at Apple. It pools idle Mac GPUs into a shared compute cluster for ML inference.
There is no distributed execution of a single LLM across multiple GPUs using tensor parallelism here. Instead, it relies on task-level parallelism-breaking down a large workload into independent subtasks that run across separate machines. For example, generating embeddings for a million documents.
This pattern has been implemented across hundreds if not thousands of different organizations (and between organizations). I have a patent on data distribution strategies from my time at Google running a large (>1M cores 24/7 for 3 years) idle cycle harvesting project.
it's a good idea but you often end up with huge tail latencies, slight incompatibilities due to fleet heterogeneity, and a database/scheduler management problem.
As with all similar SETI@home approaches, this paper does not address the energy cost.
Yes, we can distribute tasks to idle machines in homes and on office desktops, but it is often much more expensive than sending it to a data center where power is cheaper and the hardware is more efficient.
I guess for companies this can be nice? Run some inference in the off-hours on the machines that you already have. Could be useful for analytical tasks that don't require to be instant. On the technical side this doesn't seem that groundbreaking for me though.
My warm take is that Apple is leaving a lot of money on the table by not investing in creating an independent supply chain for their chips and building their own cloud. They arguably had the first mover advantage when it came to general purpose arm compute; all they had to do was flood the market and make it easy for anyone to rent it. Not only would they corner a part of the market but perhaps ease demand pressure on other chip manufacturers too.
Apart from straight up buying TSMC's or Samsung's fabs I don't see a way of building a vertically integrated high-end chip supply chain for Apple. At least not on timeframes measured below a decade. And on those time frames the current supply issues will probably be resolved by companies who put all their effort into it instead of just tweaking their bottom line.
My guess is that they have challenges scaling the production anyway (6-16 weeks wait time for Studios), and also they would leave money in the table if they just released chips. I would assume that once they scale the production they will get back to the server game - MacOS server racks, or something more inventive like MacOS supercomputers running on a ton of prev gen chips ;)
They don't get enterprise, A/UX, mkLinux, OS X Server, Mac Pro tower, all gone now.
Swift's main reason to go open source and have good Linux support was naturally for app developers to be able to use it on server side, which naturally wasn't going to be macOS.
I macOS wasn't required for developing iDevices apps, they would surely sell even less desktops.
No, they really can't. There is no market for Apple Silicon GPU compute, largely because the fastest Mac GPUs you can buy are arguably a waste of TSMC's silicon.
Look at it like this: a maxed-out M5 Ultra Mac Studio starts at $6,799, and the iGPU benchmarks similar to an $800 RTX 4070 Ti SUPER. That's a 3nm SOC getting outperformed by a 5nm dGPU; Apple Silicon is years behind Nvidia's consumer offerings, and that's before we even compare with datacenter-grade DGX/Instinct GPUs.
It would be so incredibly cool if we finally had an accessible consumer-accessible way to turn a network of computers on a network into a supercomputer. A household has a child who gets into 3D artwork and wants to render something? People can opt their tablets, phones, laptops, whatever to crunch out some frames overnight while not being used.
There is a startup working on this called Distributive I used to work for - their compute nodes are based on js/wasm runtimes so you just need a web browser to turn a device into a compute node...
until you get a teen obsessed with 3d rendering then they're up all night finding a way to convert manga into 3d scenes and your old laptop is running 100% cpu trying to keep up. xDD
The idealized version of this in my mind has a consent-first interface which makes it easy to set boundaries and limits and wouldn't burn out the machine as a default
Yes! I remember it being a bit difficult and confusing to set up when I was 10 or so. And of course, you can't use it on e.g. your gaming consoles or phones, and AFAIK you can't set up BOINC for personal use (e.g. rendering your own blender projects on all your machines). (That said, Blender itself does provide good tooling here for Blender specifically).
I don't get it. In the paper it sounds like it's "just" a CRUD DB, then you have some fleet of Macs, and they store inputs and outputs in S3. What's the innovation here? Not trying to poo on this, but it looked like the most basic and obvious approach.
That seems to be the point: you leverage existing workstations on your corporate network and run compute batches at night while they're idle. You don't have to buy any extra servers for your low-priority AI workloads.
We were doing this decades ago with engineer's HP C3600s under their desks. It wasn't AI, just low-priority HPC / fluid stuff batched out to hundreds of workstations.
That was the original pitch for Ian Murdock's Progeny startup in the late 90s: a Linux-based distributed filesystem for corporate desktops to boot into overnight for distributed computing.
Future work will extend Minions to support distributed model training in addition to inference
That could allow enthusiasts to pool compute resources on devices like Mac Studios, Mac Minis, etc to perform large scale federated training. While it'll be massively inefficient compared to running in a data center, it could democratize open weights models even more.
It would be especially useful for RL post training of existing models, which is where much of the agentic abilities arise from.
There are a lot of parallels between the DGX (and clones) series devices and these Apple devices, maybe it would be possible to rope in the DGX boxes as well. There's lots of them out there and I'm pretty sure not all of them have high utilization.
The DGX are certainly making a case for a new workstation platform to compete with Apple. The user experience is on par, and the local model capabilities are very appealing in many use cases. I've been happy to have had a pair to play with anyway .. I sure hope the price comes down and Nvidia expand the concept ..
For now the prices seem to be only going up. It is complete madness. At the same time: I'm kind of happy about that. It means that lots of people and small to mid sized organization have a very nice incentive to make low end hardware perform better. If you look at how much effort is being poured into model compression and clever VRAM management schemes on the low end it is quite impressive. And plenty of those are compounding.
I wished they had a bit more memory bandwidth though, this is quite annoying.
Would love to build something like this but for classification models that run on CPU. Like a “Classify @ Home” p2p compute network, to which people could contribute their machines and see what work is being done
Reading the “paper”, this is just a job runner built on Ollama.
Exo, and the teams at Apple that built the foundations Exo uses, actually use modern distributed compute.
Apple has released some small models previously that seemed genuinely great, along with some quality research/efforts. I’m just not seeing the same level of depth here, am I missing something?
It's not a novel idea, but its provable that someone can take this same concept and do something truly groundbreaking. I wonder if there's value in the idea that you can join a supercomputer network of MAC/Nvidia GPUs as a way of getting off cloud and into "our cloud". As long as you offer compute, you have credit into the system to be able to support your own workloads that can't be done on just your hardware. Need to address concerns about privacy, abuse, etc, but Apple's idea here has some merit if applied right.
Imagine if Apple made a GENERIC distributed compute framework to leverage idle computing power from Macs for distributed Rendering/Compiling/etc use cases. I'm sure Apple could convince Adobe and others to support this framework for their compute-intensive products and create a really interesting ecosystem for it.
Then universities and business could reduce their need for cloud and have a reason to put that spending towards higher end Macs...
[0] https://en.wikipedia.org/wiki/Grand_Central_Dispatch
[1] https://developer.apple.com/documentation/dispatch
Basically SETI for AI workloads? At least, this is the impression the landingpage left me with.
I am a bit lost, tbh.
There is no distributed execution of a single LLM across multiple GPUs using tensor parallelism here. Instead, it relies on task-level parallelism-breaking down a large workload into independent subtasks that run across separate machines. For example, generating embeddings for a million documents.
it's a good idea but you often end up with huge tail latencies, slight incompatibilities due to fleet heterogeneity, and a database/scheduler management problem.
https://developer.nvidia.com/gpudirect
And this is because I can't be bothered to dive into older papers, so recent NVIDIA stuff will do.
Yes, we can distribute tasks to idle machines in homes and on office desktops, but it is often much more expensive than sending it to a data center where power is cheaper and the hardware is more efficient.
Swift's main reason to go open source and have good Linux support was naturally for app developers to be able to use it on server side, which naturally wasn't going to be macOS.
I macOS wasn't required for developing iDevices apps, they would surely sell even less desktops.
Maybe their chip team is already on the work, or already made, a server chip for AI purposes used internally, and doesn't have to talk about it yet.
Look at it like this: a maxed-out M5 Ultra Mac Studio starts at $6,799, and the iGPU benchmarks similar to an $800 RTX 4070 Ti SUPER. That's a 3nm SOC getting outperformed by a 5nm dGPU; Apple Silicon is years behind Nvidia's consumer offerings, and that's before we even compare with datacenter-grade DGX/Instinct GPUs.
I would definitely opt out.
Is Apple going to compensate the owners?
It would be especially useful for RL post training of existing models, which is where much of the agentic abilities arise from.
I wished they had a bit more memory bandwidth though, this is quite annoying.
https://arxiv.org/pdf/2610.12970
Shortest promoted paper I've seen from a FAANG.
Building the classification part now :) https://jeffyclassify.com/
Apple has released some small models previously that seemed genuinely great, along with some quality research/efforts. I’m just not seeing the same level of depth here, am I missing something?
Then universities and business could reduce their need for cloud and have a reason to put that spending towards higher end Macs...
https://developer.apple.com/documentation/dispatch
I’m very curious if that project evolved to this……..going to message a few friends now heh
What is it actually like?
nothing new here.