This is a most excellent place for technology news and articles.
I knew “scary fast” had to mean some sort of processor bump for hardware, but I was secretly hoping they’d kill off the remaining Lightning ports on their keyboards, trackpads, and mice.
And I was hoping they’d finally redesign that god awful mouse. I don’t know how people live with that thing.
The memory maximums are a tad silly. I’d expect …
The ray tracing is awesome, but minus that I am not eager to move up from my M1 Max.
The memory maximums are going to be more and more important when it comes to local AI applications.
Take language models for an example
To run a 30b model, you need 24gb of video ram to do it fully on the video card. That's a nvidia 3090 or 4090 today. But in the grand scheme of things, 30b is small. They are going to get much bigger, especially when you want larger contexts which allow the AI to remember more about its interactions with you.
Apples memory is unified, so it can be system ram, or video ram. You'll be able to easily load a 70b model into a MacBook with 64gb of ram for example, where you'd need 2 3090s or 4090s and a hefty PSU on a current Gen non Mac PC (if you even can with just that)
For the moment, things are better optimized for windows and nvidia hardware, but Apple is encroaching on this space, and their huge amounts of video memory will begin to unlock using and training larger and larger models with each hardware generation.
Expect to see nvidia starting to offer higher video ram cards as well for this exact reason. Maybe even cards tailored to that instead of gaming with really high amounts of ram.
I can't see local models or hardware needing to scale much past the sizes we already have. Recent models like mistral have shown that we are still far from saturation at current model sizes.
And we only ever needed 64kb of ram.
Even if we have a lot of room to optimize and grow within what we have, we still have so much more to do.
Fully coherent audio and video synthesis for a scene for example.
And these models are being trained on server farms, but thats just because video memory is so expensive to come by.
We're just starting to crawl, we haven't even started walking yet on where this is going.
I was mainly referring to language models which have somewhat predictable scaling laws. It doesn't make sense to continue scaling the parameters when you can scale the data instead.
Diffusion models are a completely different domain which is less established. Most advancements made in that space are related to the architecture and training methodology. In terms of scale they haven't changed much.
Large models will always be trained in datacenters because the compute will always be exponentially greater and cheaper than what you could get as an individual. Local finetuning already happens but it's expensive and limited.
Doesn't the m2 max allow 196gb of ram? Seems like an odd downgrade. The value in these for me is the unified memory for large ai models, but most consumers may not notice that. Who knows.
That's the m2 ultra which is only the desktop version right now.
What is "dynamic caching"? The article mentions that iris new, but nothing more.
Based on my 60 seconds reading on it, onboard GPUs typically share the systems RAM. It is usually a fixed amount from my understanding. Dynamic caching seems to allow the GPU to only consume what it needs. Without knowing more, I'm guessing this means it frees up more RAM for the system instead of holding a fixed chunk in reserve for the GPU, or, on the other side, allows the GPU to use more RAM than some predetermined fixed amount.
According to Apple's press release, the GPUs in the new Macs are already faster and more efficient than those that came before them. But they go further thanks to their support for Dynamic Caching, a feature that "unlike traditional GPUs, allocates the use of local memory in hardware in real time."
What does that mean? Apple says that "with Dynamic Caching, only the exact amount of memory needed is used for each task. This is an industry first, transparent to developers, and the cornerstone of the new GPU architecture."
Do we have a m3 air now? I can't find it. I was planning to buy m2 air but now we have m3 so probably best to wait ?
Not yet, it’s just the MacBook Pro and iMac for now
I was kinda bored for this announcement. I have a M2 Pro MBP for work and I really have no desire to get anything faster.
I was hoping for a new iPad Mini announcement
I’m hoping for a new iPhone mini announcement…
I’m hoping for an Apple Watch mini announcement*
*-not really, I just wanted to feel like I was a part of something for a moment there…
I'm am too wanting to be a part of this... So... Mini mac studio would be nice
So… a Mac mini…? 😀
pity qualcomm has just wiped the floor with them. even with the new M3 it's not even close I believe. please correct me if I'm wrong (who am I kidding you're gunna correct me)
